fix(scheduler): prevent duplicate slot assignment from concurrent pipeline runs

Two bugs caused multiple articles to land on the same publish slot:

1. main.py: asyncio.create_task() returned immediately, allowing a second
   pipeline trigger (N8N + Telegram /run or two N8N calls) to start a
   second concurrent run. Added asyncio.Lock (_pipeline_lock) so any
   second trigger while the pipeline is running is rejected immediately.

2. scheduler.py: reserve_publish_slot() read the list of occupied slots
   and wrote the new slot in two separate DB connections. Concurrent threads
   could both see the same "free" slot before either committed its write.
   Fixed by wrapping the entire read-find-write cycle in a threading.Lock
   (_slot_lock) and a single DB connection, so the slot check and the
   slot assignment are atomic.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.dedupe/report.csv

@@ -0,0 +1 @@
+group_id,canonical_path,dup_path,dup_size_bytes

.env.example

@@ -0,0 +1,16 @@
+# Copy to .env and fill in values
+
+# WordPress base URL (required)
+WP_BASE_URL=https://your-site.tld
+
+# Authentication: prefer WP_AUTH_BASE64 OR use USERNAME+PASSWORD (Application Password)
+# Example to generate: base64(username:application_password)
+WP_AUTH_BASE64=
+
+# Alternatively provide username and application password
+WP_USERNAME=
+WP_PASSWORD=
+
+# OpenAI API key (optional, enables rewrite)
+OPENAI_API_KEY=
+

.github/workflows/deploy.yml

@@ -19,9 +19,16 @@ jobs:
           username: oliver
           key: ${{ secrets.HETZNER_SSH_KEY }}
           port: 22
+          envs: APP_ADMIN_USERNAME,APP_ADMIN_PASSWORD
           script: |
-            cd rss-news
+            cd /opt/rss-news
             git pull origin main
             source .venv/bin/activate
             pip install -r requirements.txt
-            sudo systemctl restart rss-app
+            pip install -r backend/requirements.txt || true
+            sudo systemctl restart rss-news-api
+            sleep 3
+            BASE_URL="https://news.vanityontour.de" APP_ADMIN_USERNAME="${APP_ADMIN_USERNAME}" APP_ADMIN_PASSWORD="${APP_ADMIN_PASSWORD}" bash scripts/smoke_backend.sh
+        env:
+          APP_ADMIN_USERNAME: ${{ secrets.NEWS_APP_ADMIN_USERNAME }}
+          APP_ADMIN_PASSWORD: ${{ secrets.NEWS_APP_ADMIN_PASSWORD }}

.github/workflows/test.yml

@@ -0,0 +1,39 @@
+name: Backend Tests
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  backend-tests:
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -r backend/requirements.txt
+          pip install -r backend/requirements-test.txt
+
+      - name: Run tests with coverage
+        env:
+          APP_DB_PATH: /tmp/rss_news_test.db
+        run: |
+          pytest backend/tests --cov=backend/app --cov-report=term-missing --cov-report=xml
+
+      - name: Upload coverage artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: coverage-xml
+          path: coverage.xml

.gitignore

@@ -35,3 +35,5 @@ internal/start.sh
 internal/copy_files.sh
 internal/_line.txt
 internal/push_commit.txt
+internal/git.sh
+CLAUDE.md

AGENTS.md

@@ -0,0 +1,40 @@
+# Repository Guidelines
+
+## Project Structure & Module Organization
+- `app.py`: Streamlit UI (entry point for the app).
+- `main.py`: RSS fetching, rewrite, and WordPress upload logic.
+- `utils/`: Helpers (image/article extraction, WP uploader, UI helpers).
+- `pages/`: Streamlit pages (e.g., `01_feed_manager.py`, `log_viewer.py`).
+- `data/`: JSON state (`articles.json`, `feeds.json`).
+- `logs/`: Runtime logs (`rss_tool.log`).
+- `docs/`: Project notes (e.g., roadmap).
+- `__version__.py`: Version string written by `versioning.py`.
+
+## Build, Test, and Development Commands
+- Create env: `python -m venv .venv && source .venv/bin/activate`
+- Install deps: `pip install -r requirements.txt`
+- Run app: `streamlit run app.py`
+- Version bump: `python versioning.py --level patch --push` (updates `__version__.py`, prepares `CHANGELOG.md`, creates tag; see `--help`).
+
+## Coding Style & Naming Conventions
+- Python 3.10+, PEP 8, 4-space indentation, type hints where practical.
+- Modules and functions: `snake_case`; classes: `PascalCase`.
+- Streamlit pages: numeric prefix for order, e.g., `pages/01_feature.py`.
+- Keep functions small and pure in `utils/`; isolate I/O in app layers.
+- Suggested tools (optional): Black (`black .`) and Ruff (`ruff check .`).
+
+## Testing Guidelines
+- Framework: pytest (recommended). Place tests under `tests/` with `test_*.py`.
+- Unit tests for `utils/*`; light integration checks for `main.py` with temporary files.
+- Run: `pytest -q`. Add coverage if needed (e.g., `pytest --cov=utils`).
+- Test data: avoid mutating files in `data/`; use temp dirs or fixtures.
+
+## Commit & Pull Request Guidelines
+- Commits: imperative mood, concise; examples: `Add feed dedupe`, `Fix WP upload retry`, `Bump version to v1.7.0`.
+- PRs: clear description, linked issue, screenshots/GIFs for UI changes, note env variables touched.
+- Update `CHANGELOG.md` and bump version via `versioning.py` before release PRs.
+
+## Security & Configuration Tips
+- Required env: `OPENAI_API_KEY`, `WP_BASE_URL`, `WP_USERNAME`, `WP_PASSWORD` or `WP_AUTH_BASE64` (see `.env`).
+- Never commit secrets; `.env` is git-ignored. Avoid hardcoded credentials; prefer `os.getenv`.
+- Logs and data may contain content; do not commit `logs/` or large `data/` snapshots.

CHANGELOG.md

@@ -1,3 +1,52 @@
+## [1.7.1] - 2025-08-24
+
+###  ✨ Security angepasst
+  - alle Credentials in die .env Datei verschoben
+  - beim Start der App werden die Credentials geprüft und beim fehlen entsprechende Meldungen ausgegeben
+
+---
+
+## [1.7.0] - 2025-08-24
+
+###  Multi-Select & Massenoperationen:
+  - ✅ Checkboxes für Artikel-Auswahl im "Artikel verwalten" Bereich
+  - ✅ "Alle auswählen" / "Auswahl aufheben" Buttons
+  - ✅ Massenoperationen für ausgewählte Artikel:
+    - Bulk Status-Änderung für mehrere Artikel gleichzeitig
+    - Bulk Artikel-Umschreibung mit automatischer Status-Verwaltung
+    - Bulk WordPress-Upload nur für "Process"-Artikel
+    - Bulk Papierkorb-Funktion
+
+###  Schnellaktionen Integration:
+  - ✅ Feed-Aktualisierung direkt im Artikel-Tab verfügbar
+  - ✅ Alle Dashboard-Schnellaktionen in Artikel-Verwaltung integriert
+  - ✅ Intelligente Anzeige nur relevanter Operationen (z.B. WordPress-Upload nur bei Process-Artikeln)
+
+###  🔧 Verbesserungen
+
+  - UI/UX: Verbesserte Artikel-Card-Layouts mit Checkbox-Integration
+  - Workflow: Streamlined Artikel-Management ohne Tab-Wechsel nötig
+  - Feedback: Detaillierte Statusmeldungen bei Massenoperationen
+  - Performance: Optimierte Session-State-Verwaltung für Artikel-Auswahl
+
+###  🏗️ Technische Änderungen
+
+  - Session State Erweiterung um selected_articles Set
+  - Neue Bulk-Operation-Funktionen in app.py:326-467
+  - Überarbeitetes Artikel-Card-Layout mit 3-Spalten-Design
+  - Integration bestehender WordPress-Upload und Rewrite-Funktionen
+
+---
+
+## [1.6.3] - 2025-08-18
+
+### 🔧 Verbesserungen
+- **SyleSheet erneut hinzugefügt
+  - Style wurde bei einem Release leider vergessen
+  - Style auf DarkMode angepasst
+
+---
+
 ## [v1.6.2] - 2025-08-16
 
 ### 🐛 Kritische Fehlerbehebung

README.md

@@ -1,76 +1,63 @@
-# 📰 RSS News Bot
+# rss-news (Rebuild)
 
-Ein intelligentes Tool zum Einlesen, Umschreiben und Veröffentlichen von Artikeln aus RSS-Feeds – mit automatischer Tag-Erkennung, KI-unterstütztem Rewrite via GPT-4, Bildextraktion aus Originalartikeln und optionaler DALL·E-Bildgenerierung.
+`rss-news` wird als bestehendes Repository weitergefuehrt und schrittweise zu einer robusten, rechtssicheren News-Pipeline neu aufgebaut.
 
-![Version](https://img.shields.io/badge/version-1.5.2-blue)
-![License](https://img.shields.io/badge/license-MIT-green)
-![Python](https://img.shields.io/badge/python-3.10+-yellow)
-![Streamlit](https://img.shields.io/badge/built%20with-Streamlit-ff4b4b)
+Aktueller Stand:
+- Alte Streamlit-App wird nicht produktiv genutzt.
+- `news.vanityontour.de` wird bis zum Go-Live der neuen App auf `https://vanityontour.de` umgeleitet.
+- Planung, Doku und Wiki werden als Grundlage fuer den Neuaufbau gepflegt.
 
----
+## Ziele
+- RSS-gestuetzte Artikelverarbeitung mit klaren Quellregeln
+- Rechtssichere Nutzung (Quellen, Attribution, Lizenzinformationen)
+- Zuverlaessige Automatisierung auf Hetzner
+- Publikation nach WordPress (IONOS aktuell, spaeter offen)
+- Zugriff nur nach Login (zunaechst User/Password)
 
-## 🚀 Features
+## Architektur-Richtung (MVP)
+- Backend: `Python + FastAPI`
+- Jobs: Queue-Worker (z. B. Redis + RQ/Celery)
+- Daten: SQLite fuer MVP, spaeter optional PostgreSQL
+- Auth: Session-Login mit einem Admin-User
+- Publishing: WordPress REST API (Status zunaechst `pending`)
 
-- 📡 **RSS-Feeds verwalten** (hinzufügen, aktualisieren)
-- ✍️ **Artikel automatisch umschreiben** mit GPT-4
-- 🏷️ **Tags automatisch generieren**
-- 🖼️ **Bilder aus Originalartikeln extrahieren**
-- 🪄 **Optionales DALL·E-Bild generieren**
-- 🔧 **Bearbeiten von Bildmetadaten**
-- 🗂️ **Statusverwaltung der Artikel (New, Rewrite, Process, etc.)**
-- 📜 **Log-Viewer-Seite integriert**
-- 📥 **Export zur Veröffentlichung auf WordPress vorbereitet**
-- 📋 Artikeltabelle mit Status-Filter
-- 🔍 Artikel-Expander mit Rewrite, Tags & Bildern
-- 🪄 Button für KI-Bildgenerierung
+Details: `docs/PROJECT_PLAN.md`
 
+## Projektsteuerung
+- GitHub Project: `https://github.com/users/OliverGiertz/projects/3/views/1`
+- Dieses Board ist die zentrale Steuerung fuer ToDos, Bugs, Verbesserungen.
+- Wiki-Struktur liegt unter `docs/wiki/`.
 
----
+## Dokumentation
+- Projektplan: `docs/PROJECT_PLAN.md`
+- ToDo-Liste: `docs/TODO.md`
+- Quell- und Lizenzpolicy: `docs/SOURCE_POLICY.md`
+- Wiki Home: `docs/wiki/Home.md`
 
-## 🧱 Projektstruktur
-
-ss-news/
-├── app.py # Haupt-UI mit Streamlit
-├── main.py # Logik für Feed-Import und Verarbeitung
-├── utils/
-│ └── image_extractor.py # Bilder aus Originalartikeln extrahieren
-│ └── dalle_generator.py # DALL·E-Integration (KI-Bild)
-├── pages/
-│ └── log_viewer.py # UI zur Anzeige der Logs
-├── data/
-│ └── articles.json # Gespeicherte Artikel
-│ └── feeds.json # Gespeicherte Feed-URLs
-├── logs/
-│ └── rss_tool.log # Logging der Verarbeitung
-├── versioning.py # CLI-Tool zur Versionierung & Release
-├── TEST-CHECKLIST.md # Manuelle Prüfliste für Releases
-├── version.py # Aktuelle Version
-└── CHANGELOG.md # Änderungsprotokoll
-
-
----
-
-## ⚙️ Installation
+## Lokale Entwicklung (Legacy-Code)
+Der vorhandene Legacy-Stand kann weiterhin lokal gestartet werden:
 
 ```bash
-git clone https://github.com/OliverGiertz/rss-news.git
-cd rss-news
 python -m venv .venv
 source .venv/bin/activate
 pip install -r requirements.txt
-```
-
----
-
-## Update
-Ein Update Script findest du hier: https://gist.github.com/OliverGiertz/ad33ae3de9aa1c1163dad5fe8affb6ca
-
-```bash
-bash update.sh
-```
-
-
-## ▶️ Starten der App
-
 streamlit run app.py
+```
+
+Hinweis: Diese App ist funktional historisch und wird durch die neue Architektur ersetzt.
+
+## Deployment-Zielbild
+- Betrieb auf Hetzner
+- Reverse Proxy via CloudPanel/Nginx
+- Produktive Domain: `news.vanityontour.de`
+- Bis zur Fertigstellung: Redirect auf `https://vanityontour.de`
+
+## Sicherheit
+- Keine Secrets im Repository
+- `.env` lokal/auf Server, nie committen
+- Auth-Pflicht fuer die neue WebApp
+- spaeter optional: Passkeys/WebAuthn
+
+## Rechtlicher Hinweis
+Dieses Projekt verarbeitet nur Quellen mit dokumentierter Nutzungsgrundlage. Vor produktiver Nutzung ist eine finale rechtliche Pruefung der ausgewaehlten Feeds notwendig.
 

__version__.py

@@ -1 +1 @@
-VERSION = "1.6.2"
+VERSION = "1.7.1"

app.py

@@ -13,6 +13,8 @@ from main import (
 )
 from utils.dalle_generator import generate_dalle_image
 from utils.wordpress_uploader import WordPressUploader
+from utils.css_loader import load_css, apply_dark_theme
+from utils.config import validate_env
 import os
 from collections import Counter
 import time
@@ -24,103 +26,22 @@ st.set_page_config(
     initial_sidebar_state="collapsed"
 )
 
-# === Custom CSS für modernes Design ===
-st.markdown("""
-<style>
-    /* Hauptcontainer */
-    .main-header {
-        background: linear-gradient(90deg, #667eea 0%, #764ba2 100%);
-        padding: 2rem;
-        border-radius: 10px;
-        margin-bottom: 2rem;
-        color: white;
-        text-align: center;
-    }
-    
-    /* Artikel Cards */
-    .article-card {
-        background: white;
-        border-radius: 12px;
-        padding: 1.5rem;
-        margin-bottom: 1rem;
-        box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
-        border-left: 4px solid #667eea;
-        transition: transform 0.2s;
-    }
-    
-    .article-card:hover {
-        transform: translateY(-2px);
-        box-shadow: 0 8px 15px rgba(0, 0, 0, 0.15);
-    }
-    
-    /* Status Badges */
-    .status-badge {
-        padding: 0.3rem 0.8rem;
-        border-radius: 20px;
-        font-size: 0.8rem;
-        font-weight: bold;
-        margin-right: 0.5rem;
-    }
-    
-    .status-new { background-color: #e3f2fd; color: #1976d2; }
-    .status-rewrite { background-color: #fff3e0; color: #f57c00; }
-    .status-process { background-color: #f3e5f5; color: #7b1fa2; }
-    .status-online { background-color: #e8f5e8; color: #388e3c; }
-    .status-hold { background-color: #fce4ec; color: #c2185b; }
-    .status-trash { background-color: #ffebee; color: #d32f2f; }
-    .status-wp-pending { background-color: #e1f5fe; color: #0277bd; }
-    
-    /* Filter Section */
-    .filter-section {
-        background: #f8f9fa;
-        padding: 1.5rem;
-        border-radius: 10px;
-        margin-bottom: 2rem;
-    }
-    
-    /* Stats Cards */
-    .stats-card {
-        background: white;
-        padding: 1.5rem;
-        border-radius: 10px;
-        text-align: center;
-        box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
-    }
-    
-    .stats-number {
-        font-size: 2rem;
-        font-weight: bold;
-        color: #667eea;
-    }
-    
-    /* Action Buttons */
-    .action-button {
-        margin: 0.25rem;
-    }
-    
-    /* Image Gallery */
-    .image-gallery {
-        display: flex;
-        gap: 1rem;
-        overflow-x: auto;
-        padding: 1rem 0;
-    }
-    
-    .image-item {
-        min-width: 200px;
-        text-align: center;
-    }
-    
-    /* WordPress Upload Status */
-    .wp-status {
-        background: #e3f2fd;
-        padding: 1rem;
-        border-radius: 8px;
-        margin: 1rem 0;
-        border-left: 4px solid #2196f3;
-    }
-</style>
-""", unsafe_allow_html=True)
+# === CSS & Theme laden ===
+load_css()
+apply_dark_theme()
+
+# === Environment-Validierung (.env) ===
+env_check = validate_env()
+if not env_check.get("ok"):
+    st.error("🔒 Sicherheits-/Konfigurationshinweis: Bitte .env korrekt konfigurieren.")
+    for msg in env_check.get("errors", []):
+        st.markdown(f"- ❌ {msg}")
+    for msg in env_check.get("warnings", []):
+        st.markdown(f"- ⚠️ {msg}")
+elif env_check.get("warnings"):
+    st.info("ℹ️ Hinweise zur Konfiguration:")
+    for msg in env_check.get("warnings", []):
+        st.markdown(f"- ⚠️ {msg}")
 
 # === Initialize Session State ===
 if 'selected_articles' not in st.session_state:
@@ -340,9 +261,8 @@ with tab1:
         <div class="article-card">
             <div style="display: flex; justify-content: space-between; align-items: center;">
                 <div>
-                    <strong>{article.get('title', 'Kein Titel')}</strong>
-                    <br>
-                    <small>{format_date(article.get('date', ''))}</small>
+                    <h3 class="article-title">{article.get('title', 'Kein Titel')}</h3>
+                    <div class="article-meta">{format_date(article.get('date', ''))}</div>
                 </div>
                 <div>
                     {get_status_badge(article.get('status', 'New'))}
@@ -417,8 +337,148 @@ with tab2:
             or any(query in tag.lower() for tag in a.get("tags", []))
         ]
     
-    # Ergebnisse anzeigen
-    st.write(f"**{len(filtered_articles)} Artikel gefunden**")
+    # Ergebnisse und Massenoperationen
+    col1, col2 = st.columns([2, 1])
+    
+    with col1:
+        st.write(f"**{len(filtered_articles)} Artikel gefunden**")
+    
+    with col2:
+        # Select All / None Buttons
+        if filtered_articles:
+            col_select_1, col_select_2 = st.columns(2)
+            with col_select_1:
+                if st.button("✓ Alle auswählen", key="select_all"):
+                    for article in filtered_articles:
+                        st.session_state.selected_articles.add(article['id'])
+                    st.rerun()
+            
+            with col_select_2:
+                if st.button("✗ Auswahl aufheben", key="select_none"):
+                    st.session_state.selected_articles.clear()
+                    st.rerun()
+    
+    # Bulk Operations Section
+    selected_count = len(st.session_state.selected_articles)
+    if selected_count > 0:
+        st.markdown(f"""
+        <div class="filter-section" style="margin-top: 10px;">
+            <h4>⚡ Massenoperationen ({selected_count} Artikel ausgewählt)</h4>
+        </div>
+        """, unsafe_allow_html=True)
+        
+        # Quick Actions für ausgewählte Artikel
+        col1, col2, col3, col4, col5 = st.columns(5)
+        
+        with col1:
+            if st.button("🔄 Feeds aktualisieren", use_container_width=True, key="bulk_update_feeds"):
+                with st.spinner("Feeds werden aktualisiert..."):
+                    existing_ids = [a["id"] for a in all_articles]
+                    process_articles(existing_ids)
+                    show_notification("Feeds erfolgreich aktualisiert!")
+                    time.sleep(1)
+                    st.rerun()
+        
+        with col2:
+            # Bulk Status Change
+            bulk_status = st.selectbox(
+                "Status ändern", 
+                ["--Auswählen--"] + ["New", "Rewrite", "Process", "Online", "On Hold", "Trash", "WordPress Pending"],
+                key="bulk_status"
+            )
+            
+            if bulk_status != "--Auswählen--" and st.button("Status anwenden", key="apply_bulk_status"):
+                changed_count = 0
+                for article in all_articles:
+                    if article["id"] in st.session_state.selected_articles:
+                        article["status"] = bulk_status
+                        changed_count += 1
+                
+                if changed_count > 0:
+                    save_articles(all_articles)
+                    show_notification(f"{changed_count} Artikel auf '{bulk_status}' gesetzt!")
+                    st.session_state.selected_articles.clear()
+                    st.rerun()
+        
+        with col3:
+            # Bulk Rewrite
+            rewrite_selected_count = len([a for a in all_articles if a["id"] in st.session_state.selected_articles and a.get("status") != "Rewrite"])
+            if st.button(f"✍️ Artikel umschreiben ({rewrite_selected_count})", use_container_width=True, key="bulk_rewrite"):
+                # Ausgewählte Artikel auf "Rewrite" setzen
+                for article in all_articles:
+                    if article["id"] in st.session_state.selected_articles:
+                        article["status"] = "Rewrite"
+                
+                save_articles(all_articles)
+                
+                # Umschreiben starten
+                with st.spinner(f"{rewrite_selected_count} Artikel werden umgeschrieben..."):
+                    rewrite_articles()
+                    show_notification(f"{rewrite_selected_count} Artikel erfolgreich umgeschrieben!")
+                    st.session_state.selected_articles.clear()
+                    time.sleep(1)
+                    st.rerun()
+        
+        with col4:
+            # Bulk WordPress Upload
+            wp_ready_selected = len([a for a in all_articles if a["id"] in st.session_state.selected_articles and a.get("status") == "Process"])
+            if wp_ready_selected > 0:
+                if st.button(f"📤 WordPress Upload ({wp_ready_selected})", use_container_width=True, key="bulk_wp_upload"):
+                    with st.spinner(f"{wp_ready_selected} Artikel werden zu WordPress hochgeladen..."):
+                        # Nur die ausgewählten "Process" Artikel hochladen
+                        selected_process_articles = [a for a in all_articles if a["id"] in st.session_state.selected_articles and a.get("status") == "Process"]
+                        
+                        if selected_process_articles:
+                            from utils.wordpress_uploader import upload_articles_to_wordpress
+                            upload_results = upload_articles_to_wordpress(selected_process_articles)
+                            
+                            if upload_results.get('error'):
+                                show_notification(f"Fehler beim WordPress-Upload: {upload_results['error']}", "error")
+                            else:
+                                successful = upload_results.get('successful', 0)
+                                failed = upload_results.get('failed', 0)
+                                duplicates = upload_results.get('duplicates', 0)
+                                
+                                # Status der erfolgreich hochgeladenen Artikel ändern
+                                if successful > 0:
+                                    for detail in upload_results.get('details', []):
+                                        if detail.get('success'):
+                                            article_id = detail.get('article_id')
+                                            for article in all_articles:
+                                                if article.get('id') == article_id:
+                                                    article['status'] = "WordPress Pending"
+                                                    article['wp_upload_date'] = datetime.now().isoformat()
+                                                    article['wp_post_id'] = detail.get('wp_post_id')
+                                                    break
+                                    save_articles(all_articles)
+                                
+                                if successful > 0:
+                                    show_notification(f"✅ {successful} Artikel erfolgreich zu WordPress hochgeladen!")
+                                if failed > 0:
+                                    show_notification(f"⚠️ {failed} Artikel konnten nicht hochgeladen werden.", "warning")
+                                if duplicates > 0:
+                                    show_notification(f"ℹ️ {duplicates} Duplikate übersprungen.", "info")
+                        
+                        st.session_state.selected_articles.clear()
+                        time.sleep(2)
+                        st.rerun()
+            else:
+                st.markdown("*Keine Process-Artikel ausgewählt*")
+        
+        with col5:
+            # Bulk Delete/Trash
+            if st.button("🗑️ In Papierkorb", use_container_width=True, key="bulk_trash"):
+                trash_count = 0
+                for article in all_articles:
+                    if article["id"] in st.session_state.selected_articles:
+                        article["status"] = "Trash"
+                        trash_count += 1
+                
+                if trash_count > 0:
+                    save_articles(all_articles)
+                    show_notification(f"{trash_count} Artikel in Papierkorb verschoben!")
+                    st.session_state.selected_articles.clear()
+                    st.rerun()
     
     # Artikel Cards
     for article in filtered_articles:
@@ -430,40 +490,48 @@ with tab2:
         # Article Card
         st.markdown('<div class="article-card">', unsafe_allow_html=True)
         
-        # Header
-        col1, col2 = st.columns([3, 1])
+        # Header with Checkbox
+        col_check, col_content, col_status = st.columns([0.3, 2.7, 1])
         
-        with col1:
+        with col_check:
+            # Checkbox für Artikel-Auswahl
+            is_selected = article["id"] in st.session_state.selected_articles
+            if st.checkbox("", value=is_selected, key=f"check_{article['id']}"):
+                st.session_state.selected_articles.add(article['id'])
+            else:
+                st.session_state.selected_articles.discard(article['id'])
+        
+        with col_content:
             title = article.get("title", "Kein Titel")
             if has_incomplete_images:
                 title += " ⚠️"
-            st.markdown(f"**{title}**")
-            st.markdown(f"📅 {format_date(article.get('date', ''))}")
+            st.markdown(f'<h3 class="article-title">{title}</h3>', unsafe_allow_html=True)
+            st.markdown(f'<div class="article-meta">📅 {format_date(article.get("date", ""))}</div>', unsafe_allow_html=True)
             
             # WordPress-Info anzeigen falls vorhanden
             if article.get("wp_post_id"):
-                st.markdown(f"🔗 WordPress ID: {article.get('wp_post_id')} | Upload: {format_date(article.get('wp_upload_date', ''))}")
+                st.markdown(f'<div class="article-meta">🔗 WordPress ID: {article.get("wp_post_id")} | Upload: {format_date(article.get("wp_upload_date", ""))}</div>', unsafe_allow_html=True)
             
-        with col2:
+        with col_status:
             st.markdown(get_status_badge(article.get("status", "New")), unsafe_allow_html=True)
         
         # Content Preview
         summary = article.get("summary", "")[:200]
         if len(summary) == 200:
             summary += "..."
-        st.markdown(summary)
+        st.markdown(f'<div class="article-summary">{summary}</div>', unsafe_allow_html=True)
         
         # Meta Info
         col1, col2, col3 = st.columns(3)
         with col1:
-            st.markdown(f"📝 **{get_word_count(article.get('text', ''))} Wörter**")
+            st.markdown(f'<div class="article-footer">📝 **{get_word_count(article.get("text", ""))} Wörter**</div>', unsafe_allow_html=True)
         with col2:
             tags = article.get("tags", [])
             if tags:
-                st.markdown(f"🏷️ {', '.join(tags[:3])}{'...' if len(tags) > 3 else ''}")
+                st.markdown(f'<div class="article-footer">🏷️ {", ".join(tags[:3])}{"..." if len(tags) > 3 else ""}</div>', unsafe_allow_html=True)
         with col3:
             source_name = source_to_name.get(article.get("source", ""), "Unbekannt")
-            st.markdown(f"📡 {source_name}")
+            st.markdown(f'<div class="article-footer">📡 {source_name}</div>', unsafe_allow_html=True)
         
         # Actions
         col1, col2, col3, col4, col5 = st.columns(5)
@@ -642,11 +710,9 @@ with tab3:
         <div class="article-card">
             <div style="display: flex; justify-content: space-between; align-items: center;">
                 <div>
-                    <strong>{feed_name}</strong>
-                    <br>
-                    <small>{feed_url}</small>
-                    <br>
-                    <span style="color: #667eea;">📰 {article_count} Artikel</span>
+                    <h3 class="article-title">{feed_name}</h3>
+                    <div class="article-meta">{feed_url}</div>
+                    <div class="article-footer">📰 {article_count} Artikel</div>
                 </div>
                 <div>
                     <span class="status-badge status-online">{article_count} Artikel</span>
@@ -712,13 +778,15 @@ with tab4:
         cols = st.columns(3)
         for idx, img in enumerate(all_images):
             with cols[idx % 3]:
+                st.markdown('<div class="image-item">', unsafe_allow_html=True)
                 st.image(img["url"], use_column_width=True)
-                st.markdown(f"**{img.get('caption', 'Kein Titel')}**")
-                st.markdown(f"📰 {img['article_title']}")
-                st.markdown(f"©️ {img.get('copyright', 'Unbekannt')}")
+                st.markdown(f'<strong class="text-primary">{img.get("caption", "Kein Titel")}</strong>', unsafe_allow_html=True)
+                st.markdown(f'<div class="text-secondary">📰 {img["article_title"]}</div>', unsafe_allow_html=True)
+                st.markdown(f'<div class="text-muted">©️ {img.get("copyright", "Unbekannt")}</div>', unsafe_allow_html=True)
                 
                 if img.get("copyright_url") and img["copyright_url"] != "#":
-                    st.markdown(f"[🔗 Quelle]({img['copyright_url']})")
+                    st.markdown(f'<a href="{img["copyright_url"]}" target="_blank" class="text-accent">🔗 Quelle</a>', unsafe_allow_html=True)
+                st.markdown('</div>', unsafe_allow_html=True)
     else:
         st.info("Keine Bilder gefunden.")
 
@@ -735,13 +803,13 @@ with tab5:
         st.subheader("📈 Status Verteilung")
         for status, count in status_counts.items():
             percentage = (count / len(all_articles) * 100) if all_articles else 0
-            st.markdown(f"{get_status_badge(status)} {count} ({percentage:.1f}%)", unsafe_allow_html=True)
+            st.markdown(f"{get_status_badge(status)} <span class='text-primary'>{count} ({percentage:.1f}%)</span>", unsafe_allow_html=True)
     
     with col2:
         st.subheader("📡 Artikel pro Feed")
         feed_counts = Counter([source_to_name.get(a.get("source", ""), "Unbekannt") for a in all_articles])
         for feed_name, count in feed_counts.most_common():
-            st.markdown(f"**{feed_name}:** {count} Artikel")
+            st.markdown(f'<div class="text-primary"><strong>{feed_name}:</strong> {count} Artikel</div>', unsafe_allow_html=True)
     
     # WordPress-Statistiken
     st.subheader("🔗 WordPress-Statistiken")
@@ -751,15 +819,30 @@ with tab5:
         col1, col2, col3 = st.columns(3)
         
         with col1:
-            st.metric("WordPress Artikel", len(wp_articles))
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>WordPress Artikel</div>
+            </div>
+            """.format(len(wp_articles)), unsafe_allow_html=True)
         
         with col2:
             pending_count = len([a for a in wp_articles if a.get("status") == "WordPress Pending"])
-            st.metric("Ausstehend", pending_count)
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>Ausstehend</div>
+            </div>
+            """.format(pending_count), unsafe_allow_html=True)
         
         with col3:
             online_wp_count = len([a for a in wp_articles if a.get("status") == "Online"])
-            st.metric("Online", online_wp_count)
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>Online</div>
+            </div>
+            """.format(online_wp_count), unsafe_allow_html=True)
         
         # Neueste WordPress-Uploads
         recent_wp = sorted([a for a in wp_articles if a.get("wp_upload_date")], 
@@ -769,9 +852,11 @@ with tab5:
             st.subheader("🕒 Neueste WordPress-Uploads")
             for article in recent_wp:
                 st.markdown(f"""
-                **{article.get('title', 'Kein Titel')}** {get_status_badge(article.get('status', 'Unknown'))}
-                
-                WP ID: {article.get('wp_post_id')} | Upload: {format_date(article.get('wp_upload_date', ''))}
+                <div class="article-card">
+                    <h3 class="article-title">{article.get('title', 'Kein Titel')}</h3>
+                    {get_status_badge(article.get('status', 'Unknown'))}
+                    <div class="article-meta">WP ID: {article.get('wp_post_id')} | Upload: {format_date(article.get('wp_upload_date', ''))}</div>
+                </div>
                 """, unsafe_allow_html=True)
     else:
         st.info("Noch keine Artikel zu WordPress hochgeladen.")
@@ -784,13 +869,28 @@ with tab5:
         col1, col2, col3 = st.columns(3)
         
         with col1:
-            st.metric("Durchschnittliche Wortanzahl", f"{sum(word_counts) // len(word_counts)}")
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>Durchschnittliche Wortanzahl</div>
+            </div>
+            """.format(sum(word_counts) // len(word_counts)), unsafe_allow_html=True)
         
         with col2:
-            st.metric("Längster Artikel", f"{max(word_counts)} Wörter")
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>Längster Artikel (Wörter)</div>
+            </div>
+            """.format(max(word_counts)), unsafe_allow_html=True)
         
         with col3:
-            st.metric("Kürzester Artikel", f"{min(word_counts)} Wörter")
+            st.markdown("""
+            <div class="stats-card">
+                <div class="stats-number">{}</div>
+                <div>Kürzester Artikel (Wörter)</div>
+            </div>
+            """.format(min(word_counts)), unsafe_allow_html=True)
     
     # Tag Cloud Simulation
     st.subheader("🏷️ Häufigste Tags")
@@ -801,7 +901,7 @@ with tab5:
     if all_tags:
         tag_counts = Counter(all_tags)
         for tag, count in tag_counts.most_common(10):
-            st.markdown(f"**{tag}:** {count}x verwendet")
+            st.markdown(f'<div class="text-primary"><strong>{tag}:</strong> {count}x verwendet</div>', unsafe_allow_html=True)
     else:
         st.info("Keine Tags gefunden.")
 
@@ -830,28 +930,19 @@ with tab6:
         wp_user = os.getenv("WP_USERNAME", "Nicht konfiguriert")
         wp_base64 = os.getenv("WP_AUTH_BASE64", "")
         
-        st.info(f"""
-        **WordPress-Konfiguration:**
-        - URL: {wp_url}
-        - Benutzer: {wp_user}
-        - Passwort: {'✅ Konfiguriert' if os.getenv("WP_PASSWORD") else '❌ Nicht konfiguriert'}
-        - Base64 Auth: {'✅ Konfiguriert' if wp_base64 else '❌ Nicht konfiguriert'}
-        """)
+        st.markdown(f"""
+        <div class="wp-status">
+            <strong>WordPress-Konfiguration:</strong><br>
+            <div class="text-secondary">
+                URL: {wp_url}<br>
+                Benutzer: {wp_user}<br>
+                Passwort: {'✅ Konfiguriert' if os.getenv("WP_PASSWORD") else '❌ Nicht konfiguriert'}<br>
+                Base64 Auth: {'✅ Konfiguriert' if wp_base64 else '❌ Nicht konfiguriert'}
+            </div>
+        </div>
+        """, unsafe_allow_html=True)
     
-    # WordPress Auth Debug (nur für Entwicklung)
-    if st.checkbox("🔧 Debug-Modus (Auth-Details anzeigen)", value=False):
-        st.warning("⚠️ Nur für Entwicklung - zeigt Auth-Details!")
-        
-        wp_base64 = os.getenv("WP_AUTH_BASE64", "")
-        if wp_base64:
-            try:
-                import base64
-                decoded = base64.b64decode(wp_base64).decode('utf-8')
-                st.code(f"Base64: {wp_base64}\nDecoded: {decoded}")
-            except Exception as e:
-                st.error(f"Fehler beim Dekodieren: {e}")
-        else:
-            st.info("Kein Base64-String konfiguriert")
+    # Sicherheit: Kein Anzeigen sensibler Auth-Details mehr
     
     # Bulk Upload
     st.subheader("📦 Massenupload")
@@ -863,10 +954,10 @@ with tab6:
         
         # Artikel-Vorschau
         for article in process_articles_list[:5]:  # Nur die ersten 5 anzeigen
-            st.markdown(f"• **{article.get('title', 'Kein Titel')}** ({get_word_count(article.get('text', ''))} Wörter)")
+            st.markdown(f'<div class="text-primary">• <strong>{article.get("title", "Kein Titel")}</strong> ({get_word_count(article.get("text", ""))} Wörter)</div>', unsafe_allow_html=True)
         
         if len(process_articles_list) > 5:
-            st.markdown(f"... und {len(process_articles_list) - 5} weitere")
+            st.markdown(f'<div class="text-muted">... und {len(process_articles_list) - 5} weitere</div>', unsafe_allow_html=True)
         
         col1, col2 = st.columns(2)
         
@@ -884,24 +975,46 @@ with tab6:
                         col1, col2, col3 = st.columns(3)
                         
                         with col1:
-                            st.metric("Erfolgreich", upload_results.get('successful', 0))
+                            st.markdown("""
+                            <div class="stats-card">
+                                <div class="stats-number">{}</div>
+                                <div>Erfolgreich</div>
+                            </div>
+                            """.format(upload_results.get('successful', 0)), unsafe_allow_html=True)
                         with col2:
-                            st.metric("Fehlgeschlagen", upload_results.get('failed', 0))
+                            st.markdown("""
+                            <div class="stats-card">
+                                <div class="stats-number">{}</div>
+                                <div>Fehlgeschlagen</div>
+                            </div>
+                            """.format(upload_results.get('failed', 0)), unsafe_allow_html=True)
                         with col3:
-                            st.metric("Duplikate", upload_results.get('duplicates', 0))
+                            st.markdown("""
+                            <div class="stats-card">
+                                <div class="stats-number">{}</div>
+                                <div>Duplikate</div>
+                            </div>
+                            """.format(upload_results.get('duplicates', 0)), unsafe_allow_html=True)
                         
                         # Details anzeigen
                         if upload_results.get('details'):
                             st.subheader("📋 Upload-Details")
                             for detail in upload_results['details']:
                                 status_icon = "✅" if detail['success'] else "❌"
-                                st.markdown(f"{status_icon} **{detail['title']}**: {detail['message']}")
+                                st.markdown(f'<div class="text-primary">{status_icon} <strong>{detail["title"]}:</strong> {detail["message"]}</div>', unsafe_allow_html=True)
                     
                     time.sleep(2)
                     st.rerun()
         
         with col2:
-            st.info("💡 Artikel erhalten den Status 'WordPress Pending' nach erfolgreichem Upload.")
+            st.markdown("""
+            <div class="wp-status">
+                <strong>💡 Info:</strong><br>
+                <div class="text-secondary">
+                    Artikel erhalten den Status 'WordPress Pending' nach erfolgreichem Upload.
+                </div>
+            </div>
+            """, unsafe_allow_html=True)
     
     else:
         st.info("Keine Artikel mit Status 'Process' gefunden. Artikel müssen zuerst umgeschrieben werden.")
@@ -931,8 +1044,7 @@ with tab6:
             <div class="wp-status">
                 <div style="display: flex; justify-content: space-between; align-items: center;">
                     <div>
-                        <strong>{article.get('title', 'Kein Titel')}</strong>
-                        <br>
+                        <strong>{article.get('title', 'Kein Titel')}</strong><br>
                         <small>WP ID: {article.get('wp_post_id')} | Upload: {format_date(article.get('wp_upload_date', ''))}</small>
                     </div>
                     <div>
@@ -951,53 +1063,47 @@ with tab6:
     with st.expander("📋 .env-Datei Vorlage", expanded=False):
         st.code("""
 # WordPress-Konfiguration
-WP_BASE_URL=https://vanityontour.de
-WP_USERNAME=ogiertz
-WP_PASSWORD=whNEx9aZCIUXViV89Z3e7Z03
+WP_BASE_URL=https://your-site.tld
 
-# WordPress Base64-Authentifizierung (bevorzugte Methode)
-WP_AUTH_BASE64=b2dpZXJ0ejp3aE5FeDlhWkNJVVhWaVY4OVozZTdaMDM=
+# Entweder Base64 (empfohlen) ODER Benutzername/Passwort (Application Password)
+WP_AUTH_BASE64=
+# Oder alternativ:
+WP_USERNAME=
+WP_PASSWORD=
 
-# OpenAI-Konfiguration (für Artikel-Umschreibung)
-OPENAI_API_KEY=sk-...
+# OpenAI-Konfiguration (optional für Umschreibung)
+OPENAI_API_KEY=
         """, language="bash")
     
     with st.expander("🔑 Base64-Authentifizierung verstehen", expanded=False):
         st.markdown("""
-        **WordPress REST API Authentifizierung:**
-        
-        Die WordPress REST API erfordert eine Base64-kodierte Authentifizierung im Format:
-        ```
-        Authorization: Basic <base64_encoded_credentials>
-        ```
-        
-        **Ihr bereitgestellter Base64-String:**
-        - `b2dpZXJ0ejp3aE5FeDlhWkNJVVhWaVY4OVozZTdaMDM=`
-        - Dekodiert: `ogiertz:whNEx9aZCIUXViV89Z3e7Z03`
-        
-        **So funktioniert es:**
-        1. Benutzername und Anwendungspasswort werden kombiniert: `username:password`
-        2. Dieser String wird Base64-kodiert
-        3. Im Authorization-Header verwendet: `Basic <base64_string>`
-        
-        **Fallback-Verhalten:**
-        - Wenn `WP_AUTH_BASE64` gesetzt ist → Direkter Base64-String verwendet
-        - Wenn nicht gesetzt → Base64 wird aus `WP_USERNAME:WP_PASSWORD` generiert
-        """)
+        <div class="article-card">
+            <h3 class="article-title">WordPress REST API Authentifizierung:</h3>
+            <div class="article-summary">
+                Die WordPress REST API nutzt <code>Basic</code>-Auth mit Base64-kodierten Zugangsdaten:<br>
+                <code>Authorization: Basic &lt;base64(username:password)&gt;</code><br><br>
+                Empfehlung: In der .env <code>WP_AUTH_BASE64</code> setzen (aus <code>username:application_password</code> erzeugt).<br>
+                Alternativ können <code>WP_USERNAME</code> und <code>WP_PASSWORD</code> gesetzt werden; dann wird Base64 zur Laufzeit generiert.
+            </div>
+        </div>
+        """, unsafe_allow_html=True)
     
     with st.expander("📖 WordPress-API Berechtigungen", expanded=False):
         st.markdown("""
-        **Erforderliche Berechtigungen für den WordPress-Benutzer:**
-        
-        - `edit_posts` - Beiträge erstellen und bearbeiten
-        - `publish_posts` - Beiträge veröffentlichen (für Status-Änderungen)
-        - `upload_files` - Dateien hochladen (für spätere Bild-Uploads)
-        - `edit_categories` - Kategorien verwalten
-        - `edit_tags` - Tags verwalten
-        
-        **Anwendungspasswort erstellen:**
-        1. WordPress Admin → Benutzer → Profil
-        2. Unter "Anwendungspasswörter" neues Passwort erstellen
-        3. Name: "RSS Feed Manager"
-        4. Generiertes Passwort in .env-Datei eintragen
-        """)
+        <div class="article-card">
+            <h3 class="article-title">Erforderliche Berechtigungen für den WordPress-Benutzer:</h3>
+            <div class="article-summary">
+                • <code>edit_posts</code> - Beiträge erstellen und bearbeiten<br>
+                • <code>publish_posts</code> - Beiträge veröffentlichen (für Status-Änderungen)<br>
+                • <code>upload_files</code> - Dateien hochladen (für spätere Bild-Uploads)<br>
+                • <code>edit_categories</code> - Kategorien verwalten<br>
+                • <code>edit_tags</code> - Tags verwalten
+                <br><br>
+                <strong>Anwendungspasswort erstellen:</strong><br>
+                1. WordPress Admin → Benutzer → Profil<br>
+                2. Unter "Anwendungspasswörter" neues Passwort erstellen<br>
+                3. Name: "RSS Feed Manager"<br>
+                4. Generiertes Passwort in .env-Datei eintragen
+            </div>
+        </div>
+        """, unsafe_allow_html=True)

backend/.env.example

@@ -0,0 +1,45 @@
+# ─── App ────────────────────────────────────────────────────────────────────
+APP_ENV=development
+APP_NAME=rss-news-backend
+APP_SECRET_KEY=replace-with-a-long-random-secret
+APP_DB_PATH=backend/data/rss_news.db
+
+APP_ADMIN_USERNAME=admin
+APP_ADMIN_PASSWORD=change-me
+
+SESSION_COOKIE_NAME=rss_news_session
+SESSION_MAX_AGE_SECONDS=28800
+
+# ─── WordPress ──────────────────────────────────────────────────────────────
+WP_BASE_URL=https://your-site.tld
+WP_USERNAME=your-wp-username
+WP_PASSWORD=your-wp-app-password
+# Status für neue Beiträge: draft | future | publish
+WORDPRESS_DEFAULT_STATUS=draft
+
+# ─── OpenAI ─────────────────────────────────────────────────────────────────
+OPENAI_API_KEY=sk-...
+# gpt-4o-mini empfohlen (Kosten/Qualität)
+OPENAI_MODEL=gpt-4o-mini
+
+# ─── Telegram Bot ────────────────────────────────────────────────────────────
+# Bot-Token von @BotFather
+TELEGRAM_BOT_TOKEN=123456789:ABC...
+# Chat-ID deines persönlichen Chats oder einer Gruppe
+TELEGRAM_CHAT_ID=123456789
+# Zufälliger Secret-Token zur Webhook-Absicherung (mindestens 20 Zeichen)
+TELEGRAM_WEBHOOK_SECRET=replace-with-random-secret-min-20-chars
+
+# ─── N8N API-Key ─────────────────────────────────────────────────────────────
+# Wird von N8N im Header X-API-Key mitgeschickt
+N8N_API_KEY=replace-with-strong-random-key
+
+# ─── Pipeline-Einstellungen ──────────────────────────────────────────────────
+# Relevanz-Score >= dieser Wert: automatisch verarbeiten (0-100)
+PIPELINE_RELEVANCE_AUTO=80
+# Relevanz-Score >= dieser Wert, aber < AUTO: Telegram-Warnung senden
+PIPELINE_RELEVANCE_WARN=60
+# Maximale Drafts/Veröffentlichungen pro Tag
+PIPELINE_MAX_DRAFTS_PER_DAY=2
+# Bevorzugte Veröffentlichungszeiten (Stunden, kommagetrennt, CET)
+PIPELINE_PUBLISH_HOURS=9,14

backend/README.md

@@ -0,0 +1,82 @@
+# Backend Skeleton (FastAPI)
+
+Dieses Verzeichnis enthaelt das technische Grundgeruest fuer den Rebuild von `rss-news`.
+
+## Start (lokal)
+
+```bash
+python -m venv .venv
+source .venv/bin/activate
+pip install -r backend/requirements.txt
+uvicorn backend.app.main:app --reload --port 8501
+```
+
+## Admin UI
+- Login: `http://127.0.0.1:8501/admin/login`
+- Dashboard: `http://127.0.0.1:8501/admin/dashboard`
+
+## Environment
+- Datei: `backend/.env`
+- Vorlage: `backend/.env.example`
+
+## Endpoints
+- `GET /health` - Healthcheck
+- `POST /auth/login` - Login mit Admin-User
+- `POST /auth/logout` - Logout
+- `GET /auth/me` - Aktiver User
+- `GET /api/protected` - Geschuetzter Test-Endpoint
+- `GET /api/pipeline/status` - Basisstatus inkl. Datensatzzaehler
+- `GET /api/sources` - Quellenliste
+- `POST /api/sources` - Quelle anlegen
+- `GET /api/sources/{source_id}/policy-check` - Policy-Pruefung fuer Quelle
+- `GET /api/feeds` - Feedliste
+- `POST /api/feeds` - Feed anlegen
+- `GET /api/feeds/{feed_id}/policy-check` - Policy-Pruefung fuer Feed
+- `GET /api/runs` - Import-/Job-Runs anzeigen
+- `GET /api/runs/{run_id}` - Detailansicht eines Runs
+- `POST /api/runs` - Run starten
+- `POST /api/runs/{run_id}/finish` - Run abschliessen
+- `GET /api/articles` - Artikel anzeigen
+- `GET /api/articles/{article_id}` - Artikeldetail
+- `POST /api/articles/upsert` - Artikel idempotent anlegen/aktualisieren
+- `POST /api/articles/{article_id}/transition` - Statuswechsel nach Workflow-Regeln
+- `POST /api/articles/{article_id}/review` - Review-Entscheidung (approve/reject)
+- `POST /api/ingestion/run` - Feed-Ingestion starten (optional pro Feed)
+
+## Datenbank
+- SQLite-Datei unter `backend/data/rss_news.db`
+- Tabellen werden beim App-Start initialisiert.
+- Tabellen: `sources`, `feeds`, `runs`, `articles`
+- Dedupe-Strategie Artikel: `source_url` -> `(feed_id, source_article_id)` -> `source_hash`
+
+## Policy-Enforcement
+- Ingestion blockiert Feeds automatisch, wenn die zugeordnete Quelle nicht policy-konform ist.
+- Mindestanforderungen: `risk_level=green`, `terms_url`, `license_name`, `last_reviewed_at`, `is_enabled=1`.
+- Pro importiertem Artikel wird ein `attribution`-Block in `meta_json` gespeichert.
+
+## Review-Workflow
+- Statuskette: `new -> review -> approved -> published`
+- Ablehnung im Review setzt auf `rewrite`
+- Ungueltige Statuswechsel werden per API blockiert
+
+## Verifikation
+```bash
+python -m unittest backend.tests.test_db_repositories
+python -m unittest backend.tests.test_ingestion
+python -m unittest backend.tests.test_api_auth
+```
+
+## CI / Online-Auswertung
+- GitHub Actions Workflow: `.github/workflows/test.yml`
+- Fuehrt Tests inkl. Coverage auf Push/PR gegen `main` aus.
+
+## Hetzner Smoketest
+```bash
+BASE_URL="https://news.vanityontour.de" \
+APP_ADMIN_USERNAME="admin" \
+APP_ADMIN_PASSWORD="..." \
+bash scripts/smoke_backend.sh
+```
+
+## Hinweis
+Passwort-Hashing und CSRF/Rate-Limit sind als naechste Ausbaustufe vorgesehen.

backend/__init__.py

@@ -0,0 +1 @@
+"""Backend package for rss-news rebuild."""

backend/app/__init__.py

@@ -0,0 +1 @@
+"""Application package."""

backend/app/auth.py

@@ -0,0 +1,31 @@
+import hmac
+from typing import Optional
+
+from itsdangerous import URLSafeTimedSerializer, BadSignature, SignatureExpired
+
+from .config import get_settings
+
+
+def _serializer() -> URLSafeTimedSerializer:
+    settings = get_settings()
+    return URLSafeTimedSerializer(settings.app_secret_key, salt="rss-news-session")
+
+
+def verify_credentials(username: str, password: str) -> bool:
+    settings = get_settings()
+    user_ok = hmac.compare_digest(username, settings.app_admin_username)
+    pw_ok = hmac.compare_digest(password, settings.app_admin_password)
+    return user_ok and pw_ok
+
+
+def create_session_token(username: str) -> str:
+    return _serializer().dumps({"username": username})
+
+
+def verify_session_token(token: str) -> Optional[str]:
+    settings = get_settings()
+    try:
+        payload = _serializer().loads(token, max_age=settings.session_max_age_seconds)
+    except (BadSignature, SignatureExpired):
+        return None
+    return payload.get("username")

backend/app/config.py

@@ -0,0 +1,65 @@
+from functools import lru_cache
+from pathlib import Path
+
+from dotenv import load_dotenv
+from pydantic import AliasChoices, Field
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    # Prefer backend-specific env file to avoid collisions with legacy root .env
+    model_config = SettingsConfigDict(
+        env_file=("backend/.env", ".env"),
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    app_env: str = "development"
+    app_name: str = "rss-news-backend"
+    app_secret_key: str = "replace-with-a-long-random-secret"
+
+    app_admin_username: str = "admin"
+    app_admin_password: str = "change-me"
+
+    session_cookie_name: str = "rss_news_session"
+    session_max_age_seconds: int = 28800
+
+    app_db_path: str = "backend/data/rss_news.db"
+
+    wordpress_base_url: str | None = Field(default=None, validation_alias=AliasChoices("WORDPRESS_BASE_URL", "WP_BASE_URL"))
+    wordpress_username: str | None = Field(default=None, validation_alias=AliasChoices("WORDPRESS_USERNAME", "WP_USERNAME"))
+    wordpress_app_password: str | None = Field(default=None, validation_alias=AliasChoices("WORDPRESS_APP_PASSWORD", "WP_PASSWORD"))
+    wordpress_default_status: str = "draft"
+    openai_api_key: str | None = Field(default=None, validation_alias=AliasChoices("OPENAI_API_KEY"))
+    openai_model: str = "gpt-4o-mini"
+
+    # Telegram Bot
+    telegram_bot_token: str | None = Field(default=None, validation_alias=AliasChoices("TELEGRAM_BOT_TOKEN"))
+    telegram_chat_id: str | None = Field(default=None, validation_alias=AliasChoices("TELEGRAM_CHAT_ID"))
+    telegram_webhook_secret: str | None = Field(default=None, validation_alias=AliasChoices("TELEGRAM_WEBHOOK_SECRET"))
+
+    # N8N API authentication
+    n8n_api_key: str | None = Field(default=None, validation_alias=AliasChoices("N8N_API_KEY"))
+
+    # Pipeline behaviour
+    pipeline_relevance_auto: int = 80    # >= this: auto-process
+    pipeline_relevance_warn: int = 60    # >= this: Telegram warning, else reject
+    pipeline_max_drafts_per_day: int = 2
+    pipeline_publish_hours: str = "9,14"  # comma-separated preferred publish hours (CET)
+    pipeline_min_words_raw: int = 120    # minimum words in raw content before rewrite (else reject)
+    pipeline_min_words_rewritten: int = 150  # minimum words in rewritten content (else reject)
+    pipeline_max_article_age_days: int = 7   # skip articles older than N days during ingestion (0 = no limit)
+
+
+@lru_cache(maxsize=1)
+def get_settings() -> Settings:
+    # Prefer shared legacy env from the original rss-news workspace if present.
+    env_candidates = (
+        Path("/Users/oliver/Documents/rss-news/.env"),
+        Path("backend/.env"),
+        Path(".env"),
+    )
+    for env_path in env_candidates:
+        if env_path.exists():
+            load_dotenv(env_path, override=False)
+    return Settings()

backend/app/db.py

@@ -0,0 +1,293 @@
+import sqlite3
+from contextlib import contextmanager
+from pathlib import Path
+from typing import Any, Iterator
+
+from .config import get_settings
+
+
+def _db_path() -> Path:
+    settings = get_settings()
+    path = Path(settings.app_db_path)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    return path
+
+
+@contextmanager
+def get_conn() -> Iterator[sqlite3.Connection]:
+    conn = sqlite3.connect(_db_path())
+    conn.row_factory = sqlite3.Row
+    conn.execute("PRAGMA foreign_keys=ON;")
+    try:
+        yield conn
+        conn.commit()
+    finally:
+        conn.close()
+
+
+def init_db() -> None:
+    with get_conn() as conn:
+        conn.executescript(
+            """
+            PRAGMA journal_mode=WAL;
+
+            CREATE TABLE IF NOT EXISTS sources (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                name TEXT NOT NULL,
+                base_url TEXT,
+                terms_url TEXT,
+                license_name TEXT,
+                risk_level TEXT NOT NULL DEFAULT 'yellow' CHECK (risk_level IN ('green', 'yellow', 'red')),
+                is_enabled INTEGER NOT NULL DEFAULT 0,
+                notes TEXT,
+                last_reviewed_at TEXT,
+                created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                updated_at TEXT NOT NULL DEFAULT (datetime('now'))
+            );
+
+            CREATE TABLE IF NOT EXISTS feeds (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                source_id INTEGER,
+                name TEXT NOT NULL,
+                url TEXT NOT NULL UNIQUE,
+                is_enabled INTEGER NOT NULL DEFAULT 1,
+                etag TEXT,
+                last_modified TEXT,
+                last_checked_at TEXT,
+                created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+                FOREIGN KEY(source_id) REFERENCES sources(id) ON DELETE SET NULL
+            );
+
+            CREATE TABLE IF NOT EXISTS runs (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                run_type TEXT NOT NULL,
+                status TEXT NOT NULL CHECK (status IN ('queued', 'running', 'success', 'failed')),
+                started_at TEXT NOT NULL DEFAULT (datetime('now')),
+                finished_at TEXT,
+                details TEXT
+            );
+
+            CREATE TABLE IF NOT EXISTS publish_jobs (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                article_id INTEGER NOT NULL,
+                status TEXT NOT NULL CHECK (status IN ('queued', 'running', 'success', 'failed')),
+                attempts INTEGER NOT NULL DEFAULT 0,
+                max_attempts INTEGER NOT NULL DEFAULT 3,
+                error_message TEXT,
+                wp_post_id INTEGER,
+                wp_post_url TEXT,
+                created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                started_at TEXT,
+                finished_at TEXT,
+                FOREIGN KEY(article_id) REFERENCES articles(id) ON DELETE CASCADE
+            );
+
+            CREATE TABLE IF NOT EXISTS articles (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                feed_id INTEGER,
+                source_article_id TEXT,
+                source_hash TEXT,
+                title TEXT NOT NULL,
+                source_url TEXT NOT NULL,
+                canonical_url TEXT,
+                published_at TEXT,
+                author TEXT,
+                summary TEXT,
+                content_raw TEXT,
+                content_rewritten TEXT,
+                image_urls_json TEXT,
+                press_contact TEXT,
+                source_name_snapshot TEXT,
+                source_terms_url_snapshot TEXT,
+                source_license_name_snapshot TEXT,
+                legal_checked INTEGER NOT NULL DEFAULT 0,
+                legal_checked_at TEXT,
+                legal_note TEXT,
+                wp_post_id INTEGER,
+                wp_post_url TEXT,
+                publish_attempts INTEGER NOT NULL DEFAULT 0,
+                publish_last_error TEXT,
+                published_to_wp_at TEXT,
+                word_count INTEGER DEFAULT 0,
+                status TEXT NOT NULL DEFAULT 'new' CHECK (status IN ('new', 'rewrite', 'review', 'approved', 'published', 'error', 'no_image')),
+                meta_json TEXT,
+                created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+                FOREIGN KEY(feed_id) REFERENCES feeds(id) ON DELETE SET NULL,
+                UNIQUE(source_url)
+            );
+
+            CREATE INDEX IF NOT EXISTS idx_articles_source_article_id ON articles(source_article_id);
+            CREATE INDEX IF NOT EXISTS idx_articles_source_hash ON articles(source_hash);
+            CREATE UNIQUE INDEX IF NOT EXISTS uq_articles_feed_source_article_id
+              ON articles(feed_id, source_article_id)
+              WHERE source_article_id IS NOT NULL;
+            CREATE UNIQUE INDEX IF NOT EXISTS uq_articles_source_hash
+              ON articles(source_hash)
+              WHERE source_hash IS NOT NULL;
+            CREATE INDEX IF NOT EXISTS idx_articles_status ON articles(status);
+            CREATE INDEX IF NOT EXISTS idx_feeds_source_id ON feeds(source_id);
+            CREATE INDEX IF NOT EXISTS idx_runs_started_at ON runs(started_at);
+            CREATE INDEX IF NOT EXISTS idx_articles_published_at ON articles(published_at);
+            CREATE INDEX IF NOT EXISTS idx_publish_jobs_status_created_at ON publish_jobs(status, created_at);
+
+            CREATE TRIGGER IF NOT EXISTS trg_sources_updated_at
+            AFTER UPDATE ON sources
+            FOR EACH ROW
+            BEGIN
+                UPDATE sources SET updated_at = datetime('now') WHERE id = OLD.id;
+            END;
+
+            CREATE TRIGGER IF NOT EXISTS trg_feeds_updated_at
+            AFTER UPDATE ON feeds
+            FOR EACH ROW
+            BEGIN
+                UPDATE feeds SET updated_at = datetime('now') WHERE id = OLD.id;
+            END;
+
+            CREATE TRIGGER IF NOT EXISTS trg_articles_updated_at
+            AFTER UPDATE ON articles
+            FOR EACH ROW
+            BEGIN
+                UPDATE articles SET updated_at = datetime('now') WHERE id = OLD.id;
+            END;
+            """
+        )
+
+        # Lightweight migration for existing DBs created before source_hash was introduced.
+        existing_columns = {
+            row["name"] for row in conn.execute("PRAGMA table_info(articles)").fetchall()
+        }
+        migration_columns = {
+            "relevance_score": "ALTER TABLE articles ADD COLUMN relevance_score INTEGER",
+            "scheduled_publish_at": "ALTER TABLE articles ADD COLUMN scheduled_publish_at TEXT",
+            "source_hash": "ALTER TABLE articles ADD COLUMN source_hash TEXT",
+            "image_urls_json": "ALTER TABLE articles ADD COLUMN image_urls_json TEXT",
+            "press_contact": "ALTER TABLE articles ADD COLUMN press_contact TEXT",
+            "source_name_snapshot": "ALTER TABLE articles ADD COLUMN source_name_snapshot TEXT",
+            "source_terms_url_snapshot": "ALTER TABLE articles ADD COLUMN source_terms_url_snapshot TEXT",
+            "source_license_name_snapshot": "ALTER TABLE articles ADD COLUMN source_license_name_snapshot TEXT",
+            "legal_checked": "ALTER TABLE articles ADD COLUMN legal_checked INTEGER NOT NULL DEFAULT 0",
+            "legal_checked_at": "ALTER TABLE articles ADD COLUMN legal_checked_at TEXT",
+            "legal_note": "ALTER TABLE articles ADD COLUMN legal_note TEXT",
+            "wp_post_id": "ALTER TABLE articles ADD COLUMN wp_post_id INTEGER",
+            "wp_post_url": "ALTER TABLE articles ADD COLUMN wp_post_url TEXT",
+            "publish_attempts": "ALTER TABLE articles ADD COLUMN publish_attempts INTEGER NOT NULL DEFAULT 0",
+            "publish_last_error": "ALTER TABLE articles ADD COLUMN publish_last_error TEXT",
+            "published_to_wp_at": "ALTER TABLE articles ADD COLUMN published_to_wp_at TEXT",
+        }
+        for column, ddl in migration_columns.items():
+            if column not in existing_columns:
+                conn.execute(ddl)
+
+        # Migration: add 'no_image' to the status CHECK constraint if not present.
+        # SQLite cannot modify CHECK constraints in-place, so we recreate the table.
+        table_sql_row = conn.execute(
+            "SELECT sql FROM sqlite_master WHERE type='table' AND name='articles'"
+        ).fetchone()
+        if table_sql_row and "'no_image'" not in (table_sql_row["sql"] or ""):
+            conn.executescript(
+                """
+                PRAGMA foreign_keys=OFF;
+
+                CREATE TABLE articles_v2 (
+                    id INTEGER PRIMARY KEY AUTOINCREMENT,
+                    feed_id INTEGER,
+                    source_article_id TEXT,
+                    source_hash TEXT,
+                    title TEXT NOT NULL,
+                    source_url TEXT NOT NULL,
+                    canonical_url TEXT,
+                    published_at TEXT,
+                    author TEXT,
+                    summary TEXT,
+                    content_raw TEXT,
+                    content_rewritten TEXT,
+                    image_urls_json TEXT,
+                    press_contact TEXT,
+                    source_name_snapshot TEXT,
+                    source_terms_url_snapshot TEXT,
+                    source_license_name_snapshot TEXT,
+                    legal_checked INTEGER NOT NULL DEFAULT 0,
+                    legal_checked_at TEXT,
+                    legal_note TEXT,
+                    wp_post_id INTEGER,
+                    wp_post_url TEXT,
+                    publish_attempts INTEGER NOT NULL DEFAULT 0,
+                    publish_last_error TEXT,
+                    published_to_wp_at TEXT,
+                    word_count INTEGER DEFAULT 0,
+                    status TEXT NOT NULL DEFAULT 'new' CHECK (status IN ('new', 'rewrite', 'review', 'approved', 'published', 'error', 'no_image')),
+                    meta_json TEXT,
+                    relevance_score INTEGER,
+                    scheduled_publish_at TEXT,
+                    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                    updated_at TEXT NOT NULL DEFAULT (datetime('now')),
+                    FOREIGN KEY(feed_id) REFERENCES feeds(id) ON DELETE SET NULL,
+                    UNIQUE(source_url)
+                );
+
+                INSERT INTO articles_v2 SELECT
+                    id, feed_id, source_article_id, source_hash, title, source_url,
+                    canonical_url, published_at, author, summary, content_raw,
+                    content_rewritten, image_urls_json, press_contact,
+                    source_name_snapshot, source_terms_url_snapshot, source_license_name_snapshot,
+                    legal_checked, legal_checked_at, legal_note,
+                    wp_post_id, wp_post_url, publish_attempts, publish_last_error,
+                    published_to_wp_at, word_count, status, meta_json,
+                    relevance_score, scheduled_publish_at, created_at, updated_at
+                FROM articles;
+
+                DROP TABLE articles;
+                ALTER TABLE articles_v2 RENAME TO articles;
+
+                CREATE INDEX IF NOT EXISTS idx_articles_source_article_id ON articles(source_article_id);
+                CREATE INDEX IF NOT EXISTS idx_articles_source_hash ON articles(source_hash);
+                CREATE UNIQUE INDEX IF NOT EXISTS uq_articles_feed_source_article_id
+                  ON articles(feed_id, source_article_id)
+                  WHERE source_article_id IS NOT NULL;
+                CREATE UNIQUE INDEX IF NOT EXISTS uq_articles_source_hash
+                  ON articles(source_hash)
+                  WHERE source_hash IS NOT NULL;
+                CREATE INDEX IF NOT EXISTS idx_articles_status ON articles(status);
+                CREATE INDEX IF NOT EXISTS idx_articles_published_at ON articles(published_at);
+
+                CREATE TRIGGER IF NOT EXISTS trg_articles_updated_at
+                AFTER UPDATE ON articles
+                FOR EACH ROW
+                BEGIN
+                    UPDATE articles SET updated_at = datetime('now') WHERE id = OLD.id;
+                END;
+
+                PRAGMA foreign_keys=ON;
+                """
+            )
+
+        table_rows = conn.execute(
+            "SELECT name FROM sqlite_master WHERE type = 'table' AND name = 'publish_jobs'"
+        ).fetchall()
+        if not table_rows:
+            conn.executescript(
+                """
+                CREATE TABLE IF NOT EXISTS publish_jobs (
+                    id INTEGER PRIMARY KEY AUTOINCREMENT,
+                    article_id INTEGER NOT NULL,
+                    status TEXT NOT NULL CHECK (status IN ('queued', 'running', 'success', 'failed')),
+                    attempts INTEGER NOT NULL DEFAULT 0,
+                    max_attempts INTEGER NOT NULL DEFAULT 3,
+                    error_message TEXT,
+                    wp_post_id INTEGER,
+                    wp_post_url TEXT,
+                    created_at TEXT NOT NULL DEFAULT (datetime('now')),
+                    started_at TEXT,
+                    finished_at TEXT,
+                    FOREIGN KEY(article_id) REFERENCES articles(id) ON DELETE CASCADE
+                );
+                CREATE INDEX IF NOT EXISTS idx_publish_jobs_status_created_at ON publish_jobs(status, created_at);
+                """
+            )
+
+
+def rows_to_dicts(rows: list[sqlite3.Row]) -> list[dict[str, Any]]:
+    return [dict(r) for r in rows]

backend/app/ingestion.py

@@ -0,0 +1,486 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import datetime, timedelta, timezone
+import hashlib
+import json
+import re
+import time
+from typing import Any
+from urllib.parse import unquote, urlencode, urlparse, parse_qs
+import urllib.error
+import urllib.request as _urllib_req
+
+import feedparser
+
+from .repositories import (
+    ArticleUpsert,
+    RunCreate,
+    create_run,
+    find_existing_article_for_upsert,
+    finish_run,
+    get_feed_by_id,
+    list_enabled_feeds,
+    update_feed_fetch_state,
+    upsert_article,
+)
+from .source_extraction import extract_article, extracted_article_to_meta
+
+
+@dataclass(frozen=True)
+class IngestionStats:
+    run_id: int
+    feeds_processed: int
+    entries_seen: int
+    articles_upserted: int
+    status: str
+    message: str
+
+
+MAX_FEED_FETCH_RETRIES = 3
+
+
+def _normalize_article_url(url: str) -> str:
+    """Strip AMP and tracking query parameters from article URLs.
+
+    Removes ?outputType=valid_amp and other AMP/tracking params so that
+    AMP and non-AMP versions of the same article are deduplicated.
+    """
+    _AMP_PARAMS = {"outputtype", "amp", "outputformat"}
+    try:
+        from urllib.parse import parse_qs, urlencode
+        parsed = urlparse(url)
+        if not parsed.query:
+            return url
+        params = parse_qs(parsed.query, keep_blank_values=True)
+        filtered = {k: v for k, v in params.items() if k.lower() not in _AMP_PARAMS}
+        new_query = urlencode(filtered, doseq=True)
+        return parsed._replace(query=new_query).geturl()
+    except Exception:
+        return url
+
+
+def _resolve_google_redirect(url: str) -> str:
+    """Extract the real article URL from Google redirect URLs.
+
+    Google Alerts feed entries use tracking links like:
+      https://www.google.com/url?rct=j&sa=t&url=<encoded_real_url>&ct=ga&...
+
+    This function returns the decoded real URL if detected, otherwise the
+    original URL unchanged.
+    """
+    try:
+        parsed = urlparse(url)
+        host = (parsed.hostname or "").lower()
+        if host not in ("www.google.com", "google.com"):
+            return url
+        if parsed.path not in ("/url", "/url/"):
+            return url
+        params = parse_qs(parsed.query, keep_blank_values=False)
+        real_urls = params.get("url")
+        if real_urls:
+            return unquote(real_urls[0])
+    except Exception:
+        pass
+    return url
+
+
+def _entry_published_iso(entry: dict) -> str | None:
+    published = entry.get("published_parsed") or entry.get("updated_parsed")
+    if not published:
+        return None
+    return datetime(*published[:6], tzinfo=timezone.utc).isoformat()
+
+
+def _entry_text(entry: dict) -> tuple[str, str]:
+    summary = entry.get("summary", "") or ""
+    content = ""
+    if entry.get("content") and isinstance(entry.get("content"), list):
+        first = entry["content"][0]
+        content = first.get("value", "") if isinstance(first, dict) else ""
+    if not content:
+        content = summary
+    return summary, content
+
+
+def _entry_hash(entry: dict, feed_id: int, link: str, title: str, summary: str) -> str:
+    source_id = entry.get("id") or entry.get("guid") or ""
+    published = _entry_published_iso(entry) or ""
+    fingerprint = f"{feed_id}|{source_id}|{link}|{title.strip()}|{summary.strip()}|{published}"
+    return hashlib.sha256(fingerprint.encode("utf-8")).hexdigest()
+
+
+def _parsed_get(parsed: object, key: str, default: object = None) -> object:
+    if isinstance(parsed, dict):
+        return parsed.get(key, default)
+    return getattr(parsed, key, default)
+
+
+def _normalize_tokens(text: str) -> set[str]:
+    normalized = re.sub(r"[^a-z0-9]+", " ", text.lower())
+    return {token for token in normalized.split() if len(token) >= 4}
+
+
+def _probe_image_url(url: str, timeout: int = 5) -> bool:
+    """Return True if URL responds without a 4xx/5xx error (HEAD request).
+
+    Returns True on network/connection errors so that a flaky server does not
+    cause a valid image to be silently dropped.
+    """
+    try:
+        req = _urllib_req.Request(
+            url,
+            method="HEAD",
+            headers={"User-Agent": "Mozilla/5.0 (compatible; rss-news/1.0)"},
+        )
+        with _urllib_req.urlopen(req, timeout=timeout) as resp:
+            return resp.status < 400
+    except urllib.error.HTTPError as exc:
+        return exc.code < 400  # 3xx redirects are OK; 4xx/5xx are not
+    except Exception:
+        return True  # network error → don't filter, let WP try later
+
+
+def _rank_image_candidates(source_url: str, title: str, images: list[str]) -> list[dict[str, Any]]:
+    source_host = (urlparse(source_url).hostname or "").lower()
+    is_presseportal = "presseportal.de" in source_host
+    title_tokens = _normalize_tokens(title)
+    blocked_patterns = ("logo", "badge", "app-store", "google-play", "na-logo", "sprite", "icon", "favicon", "tracking", "pixel", ".svg", ".ico", ".gif")
+    # Known placeholder/default images that should never be used as featured image
+    placeholder_patterns = ("some-default.jpg", "default-image", "placeholder", "no-image", "noimage")
+
+
+    ranked: list[dict[str, Any]] = []
+    for url in images:
+        # Skip inline data: URIs (e.g. base64-encoded SVG placeholders)
+        if url.startswith("data:"):
+            continue
+
+        parsed = urlparse(url)
+        path = unquote(parsed.path.lower())
+        full = f"{parsed.netloc.lower()}{path}"
+        score = 0
+        reasons: list[str] = []
+
+        if any(token in full for token in placeholder_patterns):
+            score -= 300
+            reasons.append("placeholder-image")
+
+        if any(token in full for token in blocked_patterns):
+            score -= 150
+            reasons.append("blocked-pattern")
+
+        if is_presseportal and "/thumbnail/story_big/" in path:
+            score += 120
+            reasons.append("presseportal-story-big")
+        elif is_presseportal and "/thumbnail/highlight/" in path:
+            score += 45
+            reasons.append("presseportal-highlight")
+        elif is_presseportal and "/thumbnail/liste/" in path:
+            score -= 40
+            reasons.append("presseportal-list")
+
+        if "crop=" in (parsed.query or "").lower():
+            score -= 10
+            reasons.append("cropped-preview")
+
+        path_tokens = _normalize_tokens(path.replace("-", " "))
+        overlap = len(title_tokens.intersection(path_tokens))
+        if overlap > 0:
+            score += min(30, overlap * 6)
+            reasons.append(f"title-match:{overlap}")
+
+        ranked.append({"url": url, "score": score, "reasons": reasons})
+
+    ranked.sort(key=lambda item: item["score"], reverse=True)
+    return ranked
+
+
+def _select_relevant_images(source_url: str, title: str, images: list[str], max_keep: int = 3) -> tuple[list[str], str | None, list[dict[str, Any]]]:
+    # dedupe incoming order first
+    deduped: list[str] = []
+    seen: set[str] = set()
+    for image in images:
+        if image and image not in seen:
+            seen.add(image)
+            deduped.append(image)
+
+    ranked = _rank_image_candidates(source_url, title, deduped)
+    candidates = [item["url"] for item in ranked if item["score"] > -100]
+
+    # Probe top candidates (max 4) to skip definitively broken URLs (HTTP 4xx).
+    # Network errors are treated as OK to avoid false negatives on flaky servers.
+    primary = None
+    kept: list[str] = []
+    for url in candidates[:4]:
+        if _probe_image_url(url):
+            if primary is None:
+                primary = url
+            kept.append(url)
+            if len(kept) >= max_keep:
+                break
+
+    # Fallback: if all probes failed with network errors, use best candidate anyway
+    if not kept and candidates:
+        primary = candidates[0]
+        kept = candidates[:max_keep]
+
+    return kept, primary, ranked
+
+
+def _merge_ingestion_meta(existing_meta_json: str | None, attribution: dict[str, Any], extraction_meta: dict[str, Any]) -> str:
+    meta: dict[str, Any] = {}
+    if existing_meta_json:
+        try:
+            parsed = json.loads(existing_meta_json)
+            if isinstance(parsed, dict):
+                meta = parsed
+        except Exception:
+            meta = {}
+    meta["attribution"] = attribution
+    meta["extraction"] = extraction_meta
+    return json.dumps(meta, ensure_ascii=False)
+
+
+def run_ingestion(feed_id: int | None = None) -> IngestionStats:
+    run_id = create_run(RunCreate(run_type="ingestion", status="running", details="started"))
+    feeds_processed = 0
+    entries_seen = 0
+    articles_upserted = 0
+    feed_results: list[dict[str, object]] = []
+
+    try:
+        if feed_id is not None:
+            feed = get_feed_by_id(feed_id)
+            feeds = [feed] if feed and int(feed.get("is_enabled", 0)) == 1 else []
+        else:
+            feeds = list_enabled_feeds()
+
+        for feed in feeds:
+            if not feed:
+                continue
+            feeds_processed += 1
+
+            parsed = None
+            feed_error = None
+            for attempt in range(1, MAX_FEED_FETCH_RETRIES + 1):
+                try:
+                    parsed = feedparser.parse(
+                        feed["url"],
+                        etag=feed.get("etag"),
+                        modified=feed.get("last_modified"),
+                    )
+                    break
+                except Exception as exc:
+                    feed_error = str(exc)
+                    if attempt < MAX_FEED_FETCH_RETRIES:
+                        time.sleep(0.5 * attempt)
+
+            if parsed is None:
+                feed_results.append(
+                    {
+                        "feed_id": int(feed["id"]),
+                        "feed_url": feed["url"],
+                        "status": "failed",
+                        "error": feed_error or "unknown",
+                        "entries_seen": 0,
+                        "upserts": 0,
+                    }
+                )
+                continue
+
+            # Persist ETag/Last-Modified for conditional requests.
+            parsed_etag = _parsed_get(parsed, "etag")
+            parsed_modified = _parsed_get(parsed, "modified")
+            if parsed_modified and not isinstance(parsed_modified, str):
+                parsed_modified = str(parsed_modified)
+            update_feed_fetch_state(
+                feed_id=int(feed["id"]),
+                etag=parsed_etag if isinstance(parsed_etag, str) else None,
+                last_modified=parsed_modified if isinstance(parsed_modified, str) else None,
+            )
+
+            feed_entries_seen = 0
+            feed_upserts = 0
+            from .config import get_settings as _get_settings
+            _max_age_days = _get_settings().pipeline_max_article_age_days
+            for entry in _parsed_get(parsed, "entries", []):
+                entries_seen += 1
+                feed_entries_seen += 1
+                link = entry.get("link")
+                if not link:
+                    continue
+
+                # Age filter: skip articles older than max_age_days (0 = no limit)
+                if _max_age_days > 0:
+                    published_iso = _entry_published_iso(entry)
+                    if published_iso:
+                        try:
+                            published_dt = datetime.fromisoformat(published_iso)
+                            age = datetime.now(timezone.utc) - published_dt
+                            if age > timedelta(days=_max_age_days):
+                                continue
+                        except Exception:
+                            pass  # can't parse date → allow through
+
+                # Resolve Google redirect URLs (google.com/url?...&url=<actual_url>&...)
+                link = _resolve_google_redirect(link)
+                # Normalize AMP/tracking params (e.g. ?outputType=valid_amp)
+                link = _normalize_article_url(link)
+
+                summary, content_raw = _entry_text(entry)
+                # Strip HTML tags from title (Google Alerts wraps matched keywords in <b>)
+                raw_title = entry.get("title") or "Ohne Titel"
+                title = re.sub(r"<[^>]+>", "", raw_title).strip() or "Ohne Titel"
+                extracted = extract_article(link)
+
+                final_title = extracted.title or title
+                final_author = extracted.author or entry.get("author")
+                final_summary = extracted.summary or (summary[:1000] if summary else None)
+                final_content_raw = extracted.content_text or content_raw
+                final_canonical = extracted.canonical_url or entry.get("link")
+                selected_images, primary_image, ranked_images = _select_relevant_images(
+                    link,
+                    final_title,
+                    extracted.images,
+                    max_keep=3,
+                )
+
+                source_hash = _entry_hash(
+                    entry,
+                    int(feed["id"]),
+                    link,
+                    final_title,
+                    final_summary or "",
+                )
+                attribution = {
+                    "source_name": feed.get("source_name"),
+                    "source_base_url": feed.get("source_base_url"),
+                    "source_terms_url": feed.get("source_terms_url"),
+                    "source_license_name": feed.get("source_license_name"),
+                    "source_risk_level": feed.get("source_risk_level"),
+                    "original_link": link,
+                    "feed_name": feed.get("name"),
+                    "feed_id": int(feed["id"]),
+                    "imported_at": datetime.now(timezone.utc).isoformat(),
+                }
+                extraction_meta: dict[str, Any] = extracted_article_to_meta(extracted)
+                extraction_meta["fetched_from"] = link
+                extraction_meta["image_selection"] = {
+                    "primary": primary_image,
+                    "selected_count": len(selected_images),
+                    "total_candidates": len(extracted.images),
+                    "ranked": ranked_images,
+                }
+                base_payload = ArticleUpsert(
+                    feed_id=int(feed["id"]),
+                    source_article_id=entry.get("id") or entry.get("guid"),
+                    source_hash=source_hash,
+                    title=final_title,
+                    source_url=link,
+                    canonical_url=final_canonical,
+                    published_at=_entry_published_iso(entry),
+                    author=final_author,
+                    summary=final_summary,
+                    content_raw=final_content_raw,
+                    content_rewritten=None,
+                    image_urls_json=json.dumps(selected_images, ensure_ascii=False) if selected_images else None,
+                    press_contact=extracted.press_contact,
+                    source_name_snapshot=feed.get("source_name"),
+                    source_terms_url_snapshot=feed.get("source_terms_url"),
+                    source_license_name_snapshot=feed.get("source_license_name"),
+                    legal_checked=False,
+                    legal_checked_at=None,
+                    legal_note=None,
+                    wp_post_id=None,
+                    wp_post_url=None,
+                    publish_attempts=0,
+                    publish_last_error=None,
+                    published_to_wp_at=None,
+                    word_count=len((final_content_raw or "").split()),
+                    status="new",
+                    meta_json=json.dumps({"attribution": attribution, "extraction": extraction_meta}, ensure_ascii=False),
+                )
+                existing = find_existing_article_for_upsert(base_payload)
+                if existing and existing.get("status") == "error":
+                    # Explicitly closed article: ignore on subsequent ingestion runs.
+                    continue
+
+                payload = base_payload
+                if existing:
+                    payload = ArticleUpsert(
+                        feed_id=base_payload.feed_id,
+                        source_article_id=base_payload.source_article_id,
+                        source_hash=base_payload.source_hash,
+                        title=base_payload.title,
+                        source_url=base_payload.source_url,
+                        canonical_url=base_payload.canonical_url,
+                        published_at=base_payload.published_at,
+                        author=base_payload.author,
+                        summary=base_payload.summary,
+                        content_raw=base_payload.content_raw,
+                        content_rewritten=existing.get("content_rewritten"),
+                        image_urls_json=base_payload.image_urls_json,
+                        press_contact=base_payload.press_contact or existing.get("press_contact"),
+                        source_name_snapshot=base_payload.source_name_snapshot,
+                        source_terms_url_snapshot=base_payload.source_terms_url_snapshot,
+                        source_license_name_snapshot=base_payload.source_license_name_snapshot,
+                        legal_checked=bool(int(existing.get("legal_checked", 0))),
+                        legal_checked_at=existing.get("legal_checked_at"),
+                        legal_note=existing.get("legal_note"),
+                        wp_post_id=existing.get("wp_post_id"),
+                        wp_post_url=existing.get("wp_post_url"),
+                        publish_attempts=int(existing.get("publish_attempts", 0)),
+                        publish_last_error=existing.get("publish_last_error"),
+                        published_to_wp_at=existing.get("published_to_wp_at"),
+                        word_count=base_payload.word_count,
+                        status=existing.get("status") or "new",
+                        meta_json=_merge_ingestion_meta(existing.get("meta_json"), attribution, extraction_meta),
+                    )
+
+                article_id = upsert_article(payload)
+                if article_id:
+                    articles_upserted += 1
+                    feed_upserts += 1
+
+            feed_results.append(
+                {
+                    "feed_id": int(feed["id"]),
+                    "feed_url": feed["url"],
+                    "status": "success",
+                    "entries_seen": feed_entries_seen,
+                    "upserts": feed_upserts,
+                }
+            )
+
+        finish_run(
+            run_id=run_id,
+            status="success",
+            details=json.dumps(
+                {
+                    "feeds_processed": feeds_processed,
+                    "entries_seen": entries_seen,
+                    "upserts": articles_upserted,
+                    "feeds": feed_results,
+                },
+                ensure_ascii=False,
+            ),
+        )
+        return IngestionStats(
+            run_id=run_id,
+            feeds_processed=feeds_processed,
+            entries_seen=entries_seen,
+            articles_upserted=articles_upserted,
+            status="success",
+            message="Ingestion abgeschlossen",
+        )
+    except Exception as exc:
+        finish_run(run_id=run_id, status="failed", details=str(exc))
+        return IngestionStats(
+            run_id=run_id,
+            feeds_processed=feeds_processed,
+            entries_seen=entries_seen,
+            articles_upserted=articles_upserted,
+            status="failed",
+            message=str(exc),
+        )

backend/app/main.py

@@ -0,0 +1,727 @@
+import asyncio
+from contextlib import asynccontextmanager
+import csv
+from datetime import datetime, timezone
+import io
+import json
+import logging
+from pathlib import Path
+
+from fastapi import Depends, FastAPI, HTTPException, Request, Response, status
+from fastapi.responses import JSONResponse
+from pydantic import BaseModel, Field
+from fastapi.staticfiles import StaticFiles
+
+from .admin_ui import router as admin_router
+from .auth import create_session_token, verify_credentials, verify_session_token
+from .config import get_settings
+from .db import init_db
+from .ingestion import run_ingestion
+from .pipeline import run_auto_pipeline
+from .policy import evaluate_source_policy, is_source_allowed
+from .publisher import enqueue_publish, run_publisher
+from .relevance import article_age_days, article_relevance
+from .rewrite import generate_article_tags, merge_generated_tags, rewrite_article_text
+from .telegram_bot import handle_update, setup_webhook
+from .repositories import (
+    ArticleUpsert,
+    FeedCreate,
+    RunCreate,
+    SourceCreate,
+    create_feed as repo_create_feed,
+    create_run,
+    create_source as repo_create_source,
+    finish_run,
+    get_article_by_id,
+    get_feed_by_id,
+    get_run_by_id,
+    get_source_by_id,
+    list_publish_jobs,
+    list_articles as repo_list_articles,
+    list_feeds as repo_list_feeds,
+    list_runs,
+    list_sources as repo_list_sources,
+    set_article_legal_review,
+    update_article_status,
+    upsert_article as repo_upsert_article,
+)
+from .workflow import ALLOWED_UI_TRANSITIONS, UI_STATUSES, internal_to_ui_status, ui_to_internal_status
+
+settings = get_settings()
+
+
+@asynccontextmanager
+async def app_lifespan(_: FastAPI):
+    init_db()
+    yield
+
+
+app = FastAPI(title=settings.app_name, lifespan=app_lifespan)
+app.include_router(admin_router)
+app.mount(
+    "/admin/static",
+    StaticFiles(directory=str(Path(__file__).resolve().parent.parent / "static")),
+    name="admin-static",
+)
+
+
+class LoginRequest(BaseModel):
+    username: str
+    password: str
+
+
+class SourceCreateRequest(BaseModel):
+    name: str = Field(min_length=1, max_length=200)
+    base_url: str | None = None
+    terms_url: str | None = None
+    license_name: str | None = None
+    risk_level: str = Field(default="yellow", pattern="^(green|yellow|red)$")
+    is_enabled: bool = False
+    notes: str | None = None
+    last_reviewed_at: str | None = None
+
+
+class FeedCreateRequest(BaseModel):
+    name: str = Field(min_length=1, max_length=200)
+    url: str = Field(min_length=5, max_length=1000)
+    source_id: int | None = None
+    is_enabled: bool = True
+
+
+class RunCreateRequest(BaseModel):
+    run_type: str = Field(min_length=2, max_length=100)
+    status: str = Field(default="queued", pattern="^(queued|running|success|failed)$")
+    details: str | None = None
+
+
+class RunFinishRequest(BaseModel):
+    status: str = Field(pattern="^(success|failed)$")
+    details: str | None = None
+
+
+class ArticleUpsertRequest(BaseModel):
+    feed_id: int | None = None
+    source_article_id: str | None = None
+    source_hash: str | None = None
+    title: str = Field(min_length=1, max_length=500)
+    source_url: str = Field(min_length=5, max_length=2000)
+    canonical_url: str | None = None
+    published_at: str | None = None
+    author: str | None = None
+    summary: str | None = None
+    content_raw: str | None = None
+    content_rewritten: str | None = None
+    image_urls_json: str | None = None
+    press_contact: str | None = None
+    source_name_snapshot: str | None = None
+    source_terms_url_snapshot: str | None = None
+    source_license_name_snapshot: str | None = None
+    legal_checked: bool = False
+    legal_checked_at: str | None = None
+    legal_note: str | None = None
+    wp_post_id: int | None = None
+    wp_post_url: str | None = None
+    publish_attempts: int = 0
+    publish_last_error: str | None = None
+    published_to_wp_at: str | None = None
+    word_count: int = 0
+    status: str = Field(default="new", pattern="^(new|rewrite|publish|published|close|review|approved|error|no_image)$")
+    meta_json: str | None = None
+
+
+class IngestionRunRequest(BaseModel):
+    feed_id: int | None = None
+
+
+class ArticleTransitionRequest(BaseModel):
+    target_status: str = Field(pattern="^(new|rewrite|publish|published|close|review|approved|error|no_image)$")
+    note: str | None = None
+
+
+class ArticleReviewRequest(BaseModel):
+    decision: str = Field(pattern="^(approve|reject)$")
+    note: str | None = None
+
+
+class ArticleLegalReviewRequest(BaseModel):
+    approved: bool
+    note: str | None = None
+
+
+class PublisherEnqueueRequest(BaseModel):
+    article_id: int
+    max_attempts: int = 3
+
+
+class PublisherRunRequest(BaseModel):
+    max_jobs: int = 10
+
+
+ALLOWED_ARTICLE_TRANSITIONS: dict[str, set[str]] = {
+    "new": {"rewrite", "error"},
+    "rewrite": {"approved", "error"},
+    "approved": {"published", "error"},
+    "published": {"error"},
+    "error": {"rewrite"},
+}
+
+
+def require_auth(request: Request) -> str:
+    token = request.cookies.get(settings.session_cookie_name)
+    if not token:
+        raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Nicht angemeldet")
+
+    username = verify_session_token(token)
+    if not username:
+        raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Session ungueltig oder abgelaufen")
+
+    return username
+
+
+@app.get("/health")
+def health() -> dict:
+    return {"status": "ok", "service": settings.app_name, "db_path": settings.app_db_path}
+
+
+@app.post("/auth/login")
+def login(payload: LoginRequest, response: Response) -> dict:
+    if not verify_credentials(payload.username, payload.password):
+        raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Ungueltige Zugangsdaten")
+
+    token = create_session_token(payload.username)
+    response.set_cookie(
+        key=settings.session_cookie_name,
+        value=token,
+        max_age=settings.session_max_age_seconds,
+        httponly=True,
+        secure=False,
+        samesite="lax",
+    )
+    return {"ok": True, "username": payload.username}
+
+
+@app.post("/auth/logout")
+def logout(response: Response) -> dict:
+    response.delete_cookie(settings.session_cookie_name)
+    return {"ok": True}
+
+
+@app.get("/auth/me")
+def me(username: str = Depends(require_auth)) -> dict:
+    return {"authenticated": True, "username": username}
+
+
+@app.get("/api/protected")
+def protected(username: str = Depends(require_auth)) -> dict:
+    return {"ok": True, "message": "Protected endpoint", "username": username}
+
+
+@app.get("/api/pipeline/status")
+def pipeline_status(username: str = Depends(require_auth)) -> dict:
+    feeds_total = len(repo_list_feeds())
+    sources_total = len(repo_list_sources())
+    articles_total = len(repo_list_articles(limit=500))
+    return {
+        "ok": True,
+        "stage": "skeleton+db",
+        "requested_by": username,
+        "counts": {
+            "sources": sources_total,
+            "feeds": feeds_total,
+            "articles": articles_total,
+        },
+    }
+
+
+@app.get("/api/sources")
+def list_sources(username: str = Depends(require_auth)) -> dict:
+    return {"ok": True, "items": repo_list_sources(), "requested_by": username}
+
+
+@app.get("/api/sources/{source_id}/policy-check")
+def source_policy_check(source_id: int, username: str = Depends(require_auth)) -> dict:
+    source = get_source_by_id(source_id)
+    if not source:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Quelle nicht gefunden")
+    issues = evaluate_source_policy(source)
+    return {
+        "ok": True,
+        "source_id": source_id,
+        "allowed": is_source_allowed(source),
+        "issues": issues,
+        "requested_by": username,
+    }
+
+
+@app.post("/api/sources")
+def create_source(payload: SourceCreateRequest, username: str = Depends(require_auth)) -> dict:
+    source_id = repo_create_source(
+        SourceCreate(
+            name=payload.name,
+            base_url=payload.base_url,
+            terms_url=payload.terms_url,
+            license_name=payload.license_name,
+            risk_level=payload.risk_level,
+            is_enabled=payload.is_enabled,
+            notes=payload.notes,
+            last_reviewed_at=payload.last_reviewed_at,
+        )
+    )
+    return {"ok": True, "id": source_id, "requested_by": username}
+
+
+@app.get("/api/feeds")
+def list_feeds(username: str = Depends(require_auth)) -> dict:
+    return {"ok": True, "items": repo_list_feeds(), "requested_by": username}
+
+
+@app.get("/api/feeds/{feed_id}/policy-check")
+def feed_policy_check(feed_id: int, username: str = Depends(require_auth)) -> dict:
+    feed = get_feed_by_id(feed_id)
+    if not feed:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Feed nicht gefunden")
+
+    source_snapshot = {
+        "id": feed.get("source_id"),
+        "name": feed.get("source_name"),
+        "base_url": feed.get("source_base_url"),
+        "terms_url": feed.get("source_terms_url"),
+        "license_name": feed.get("source_license_name"),
+        "risk_level": feed.get("source_risk_level"),
+        "last_reviewed_at": feed.get("source_last_reviewed_at"),
+        "is_enabled": feed.get("source_is_enabled"),
+    }
+    issues = evaluate_source_policy(source_snapshot)
+    return {
+        "ok": True,
+        "feed_id": feed_id,
+        "allowed": len(issues) == 0,
+        "issues": issues,
+        "requested_by": username,
+    }
+
+
+@app.post("/api/feeds")
+def create_feed(payload: FeedCreateRequest, username: str = Depends(require_auth)) -> dict:
+    try:
+        feed_id = repo_create_feed(
+            FeedCreate(
+                name=payload.name,
+                url=payload.url,
+                source_id=payload.source_id,
+                is_enabled=payload.is_enabled,
+            )
+        )
+    except Exception as exc:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=f"Feed konnte nicht angelegt werden: {exc}") from exc
+
+    return {"ok": True, "id": feed_id, "requested_by": username}
+
+
+@app.get("/api/runs")
+def api_list_runs(limit: int = 50, username: str = Depends(require_auth)) -> dict:
+    return {"ok": True, "items": list_runs(limit=limit), "requested_by": username}
+
+
+@app.get("/api/runs/{run_id}")
+def api_get_run(run_id: int, username: str = Depends(require_auth)) -> dict:
+    run = get_run_by_id(run_id)
+    if not run:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Run nicht gefunden")
+    return {"ok": True, "item": run, "requested_by": username}
+
+
+@app.post("/api/runs")
+def api_create_run(payload: RunCreateRequest, username: str = Depends(require_auth)) -> dict:
+    run_id = create_run(RunCreate(run_type=payload.run_type, status=payload.status, details=payload.details))
+    return {"ok": True, "id": run_id, "requested_by": username}
+
+
+@app.post("/api/runs/{run_id}/finish")
+def api_finish_run(run_id: int, payload: RunFinishRequest, username: str = Depends(require_auth)) -> dict:
+    finish_run(run_id=run_id, status=payload.status, details=payload.details)
+    return {"ok": True, "id": run_id, "requested_by": username}
+
+
+@app.get("/api/articles")
+def api_list_articles(limit: int = 100, status_filter: str | None = None, username: str = Depends(require_auth)) -> dict:
+    internal_filter = ui_to_internal_status(status_filter) if status_filter else None
+    items = repo_list_articles(limit=limit, status_filter=internal_filter)
+    for item in items:
+        item["status_ui"] = internal_to_ui_status(item.get("status"))
+    return {"ok": True, "items": items, "requested_by": username}
+
+
+@app.get("/api/articles/export")
+def api_export_articles(
+    format: str = "json",
+    status_filter: str | None = None,
+    username: str = Depends(require_auth),
+):
+    internal_filter = ui_to_internal_status(status_filter) if status_filter else None
+    articles = repo_list_articles(limit=500, status_filter=internal_filter)
+    rows = []
+    for article in articles:
+        meta: dict = {}
+        if article.get("meta_json"):
+            try:
+                parsed = json.loads(article["meta_json"])
+                if isinstance(parsed, dict):
+                    meta = parsed
+            except Exception:
+                meta = {}
+        image_review = meta.get("image_review") if isinstance(meta.get("image_review"), dict) else {}
+        selected_image_url = image_review.get("selected_url") if isinstance(image_review.get("selected_url"), str) else None
+
+        days_old = article_age_days(article.get("published_at"))
+        rows.append(
+            {
+                "id": article.get("id"),
+                "title": article.get("title"),
+                "status": article.get("status"),
+                "published_at": article.get("published_at"),
+                "days_old": days_old,
+                "relevance": article_relevance(article.get("published_at")),
+                "author": article.get("author"),
+                "source_url": article.get("source_url"),
+                "canonical_url": article.get("canonical_url"),
+                "source_name_snapshot": article.get("source_name_snapshot"),
+                "source_license_name_snapshot": article.get("source_license_name_snapshot"),
+                "source_terms_url_snapshot": article.get("source_terms_url_snapshot"),
+                "press_contact": article.get("press_contact"),
+                "image_urls_json": article.get("image_urls_json"),
+                "selected_image_url": selected_image_url,
+                "legal_checked": bool(int(article.get("legal_checked", 0))),
+                "legal_checked_at": article.get("legal_checked_at"),
+                "legal_note": article.get("legal_note"),
+            }
+        )
+
+    generated_at = datetime.now(timezone.utc).isoformat()
+    if format == "csv":
+        out = io.StringIO()
+        fieldnames = [
+            "id",
+            "title",
+            "status",
+            "published_at",
+            "days_old",
+            "relevance",
+            "author",
+            "source_url",
+            "canonical_url",
+            "source_name_snapshot",
+            "source_license_name_snapshot",
+            "source_terms_url_snapshot",
+            "press_contact",
+            "image_urls_json",
+            "selected_image_url",
+            "legal_checked",
+            "legal_checked_at",
+            "legal_note",
+        ]
+        writer = csv.DictWriter(out, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(rows)
+        return Response(
+            content=out.getvalue(),
+            media_type="text/csv; charset=utf-8",
+            headers={"Content-Disposition": 'attachment; filename="articles_export.csv"'},
+        )
+
+    return JSONResponse(
+        {
+            "ok": True,
+            "count": len(rows),
+            "generated_at": generated_at,
+            "status_filter": status_filter,
+            "items": rows,
+            "requested_by": username,
+        }
+    )
+
+
+@app.get("/api/articles/{article_id}")
+def api_get_article(article_id: int, username: str = Depends(require_auth)) -> dict:
+    article = get_article_by_id(article_id)
+    if not article:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+    article["status_ui"] = internal_to_ui_status(article.get("status"))
+    return {"ok": True, "item": article, "requested_by": username}
+
+
+@app.post("/api/articles/upsert")
+def api_upsert_article(payload: ArticleUpsertRequest, username: str = Depends(require_auth)) -> dict:
+    article_id = repo_upsert_article(
+        ArticleUpsert(
+            feed_id=payload.feed_id,
+            source_article_id=payload.source_article_id,
+            source_hash=payload.source_hash,
+            title=payload.title,
+            source_url=payload.source_url,
+            canonical_url=payload.canonical_url,
+            published_at=payload.published_at,
+            author=payload.author,
+            summary=payload.summary,
+            content_raw=payload.content_raw,
+            content_rewritten=payload.content_rewritten,
+            image_urls_json=payload.image_urls_json,
+            press_contact=payload.press_contact,
+            source_name_snapshot=payload.source_name_snapshot,
+            source_terms_url_snapshot=payload.source_terms_url_snapshot,
+            source_license_name_snapshot=payload.source_license_name_snapshot,
+            legal_checked=payload.legal_checked,
+            legal_checked_at=payload.legal_checked_at,
+            legal_note=payload.legal_note,
+            wp_post_id=payload.wp_post_id,
+            wp_post_url=payload.wp_post_url,
+            publish_attempts=payload.publish_attempts,
+            publish_last_error=payload.publish_last_error,
+            published_to_wp_at=payload.published_to_wp_at,
+            word_count=payload.word_count,
+            status=ui_to_internal_status(payload.status),
+            meta_json=payload.meta_json,
+        )
+    )
+    return {"ok": True, "id": article_id, "requested_by": username}
+
+
+@app.post("/api/articles/{article_id}/transition")
+def api_article_transition(article_id: int, payload: ArticleTransitionRequest, username: str = Depends(require_auth)) -> dict:
+    article = get_article_by_id(article_id)
+    if not article:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+
+    current_status = article.get("status")
+    current_ui = internal_to_ui_status(current_status)
+    target_internal = ui_to_internal_status(payload.target_status)
+    target_ui = internal_to_ui_status(target_internal)
+    allowed_targets = ALLOWED_UI_TRANSITIONS.get(current_ui, set())
+    if target_ui not in allowed_targets:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"Ungueltiger Statuswechsel: {current_ui} -> {target_ui}",
+        )
+
+    updated = update_article_status(article_id, target_internal, actor=username, note=payload.note)
+    if not updated:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+    return {"ok": True, "id": article_id, "from_status": current_ui, "to_status": target_ui}
+
+
+@app.post("/api/articles/{article_id}/rewrite-run")
+def api_article_rewrite_run(article_id: int, username: str = Depends(require_auth)) -> dict:
+    article = get_article_by_id(article_id)
+    if not article:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+    if internal_to_ui_status(article.get("status")) not in {"rewrite", "new"}:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="Rewrite nur aus Status 'new' oder 'rewrite'")
+
+    rewritten = rewrite_article_text(article)
+    tags: list[str] = []
+    try:
+        tags = generate_article_tags(article, rewritten_text=rewritten)
+    except Exception:
+        tags = []
+    merged_meta = merge_generated_tags(article.get("meta_json"), tags)
+    # upsert via status update + existing fields by lightweight path:
+    repo_upsert_article(
+        ArticleUpsert(
+            feed_id=article.get("feed_id"),
+            source_article_id=article.get("source_article_id"),
+            source_hash=article.get("source_hash"),
+            title=article.get("title"),
+            source_url=article.get("source_url"),
+            canonical_url=article.get("canonical_url"),
+            published_at=article.get("published_at"),
+            author=article.get("author"),
+            summary=article.get("summary"),
+            content_raw=article.get("content_raw"),
+            content_rewritten=rewritten,
+            image_urls_json=article.get("image_urls_json"),
+            press_contact=article.get("press_contact"),
+            source_name_snapshot=article.get("source_name_snapshot"),
+            source_terms_url_snapshot=article.get("source_terms_url_snapshot"),
+            source_license_name_snapshot=article.get("source_license_name_snapshot"),
+            legal_checked=bool(int(article.get("legal_checked", 0))),
+            legal_checked_at=article.get("legal_checked_at"),
+            legal_note=article.get("legal_note"),
+            wp_post_id=article.get("wp_post_id"),
+            wp_post_url=article.get("wp_post_url"),
+            publish_attempts=int(article.get("publish_attempts", 0)),
+            publish_last_error=article.get("publish_last_error"),
+            published_to_wp_at=article.get("published_to_wp_at"),
+            word_count=len(rewritten.split()),
+            status="approved",
+            meta_json=merged_meta,
+        )
+    )
+    return {"ok": True, "id": article_id, "status": "publish", "tags": tags}
+
+
+@app.post("/api/articles/{article_id}/legal-review")
+def api_article_legal_review(article_id: int, payload: ArticleLegalReviewRequest, username: str = Depends(require_auth)) -> dict:
+    article = get_article_by_id(article_id)
+    if not article:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+
+    updated = set_article_legal_review(article_id, approved=payload.approved, note=payload.note, actor=username)
+    if not updated:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+    return {
+        "ok": True,
+        "id": article_id,
+        "legal_checked": payload.approved,
+    }
+
+
+@app.get("/api/publisher/jobs")
+def api_publisher_jobs(limit: int = 100, username: str = Depends(require_auth)) -> dict:
+    return {"ok": True, "items": list_publish_jobs(limit=limit), "requested_by": username}
+
+
+@app.post("/api/publisher/enqueue")
+def api_publisher_enqueue(payload: PublisherEnqueueRequest, username: str = Depends(require_auth)) -> dict:
+    article = get_article_by_id(payload.article_id)
+    if not article:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Artikel nicht gefunden")
+    job_id = enqueue_publish(article_id=payload.article_id, max_attempts=payload.max_attempts)
+    return {"ok": True, "job_id": job_id, "article_id": payload.article_id, "requested_by": username}
+
+
+@app.post("/api/publisher/run")
+def api_publisher_run(payload: PublisherRunRequest, username: str = Depends(require_auth)) -> dict:
+    stats = run_publisher(max_jobs=payload.max_jobs)
+    return {
+        "ok": True,
+        "requested_by": username,
+        "stats": {
+            "processed": stats.processed,
+            "success": stats.success,
+            "failed": stats.failed,
+            "requeued": stats.requeued,
+        },
+    }
+
+
+@app.post("/api/articles/{article_id}/review")
+def api_article_review(article_id: int, payload: ArticleReviewRequest, username: str = Depends(require_auth)) -> dict:
+    raise HTTPException(status_code=status.HTTP_410_GONE, detail="Review-Endpoint ersetzt durch Rewrite-Workflow")
+
+
+@app.post("/api/ingestion/run")
+def api_run_ingestion(payload: IngestionRunRequest, username: str = Depends(require_auth)) -> dict:
+    stats = run_ingestion(feed_id=payload.feed_id)
+    return {
+        "ok": stats.status == "success",
+        "run_id": stats.run_id,
+        "status": stats.status,
+        "message": stats.message,
+        "stats": {
+            "feeds_processed": stats.feeds_processed,
+            "entries_seen": stats.entries_seen,
+            "articles_upserted": stats.articles_upserted,
+        },
+        "requested_by": username,
+    }
+
+
+# ---------------------------------------------------------------------------
+# N8N Automation endpoint (API-Key auth, no session cookie required)
+# ---------------------------------------------------------------------------
+
+def _require_api_key(request: Request) -> None:
+    api_key = request.headers.get("X-API-Key") or request.query_params.get("api_key")
+    expected = settings.n8n_api_key
+    if not expected:
+        raise HTTPException(status_code=status.HTTP_501_NOT_IMPLEMENTED, detail="N8N_API_KEY nicht konfiguriert")
+    if api_key != expected:
+        raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Ungültiger API-Key")
+
+
+_pipeline_lock = asyncio.Lock()
+
+
+@app.post("/api/n8n/pipeline")
+async def api_n8n_pipeline(request: Request) -> dict:
+    """Trigger the full auto pipeline in background. Returns immediately.
+    Called by N8N (2x/day or on demand). Results arrive via Telegram."""
+    _require_api_key(request)
+
+    if _pipeline_lock.locked():
+        logging.getLogger(__name__).warning("Pipeline bereits aktiv – Trigger ignoriert")
+        return {"ok": False, "message": "Pipeline läuft bereits – Trigger ignoriert"}
+
+    async def _run():
+        async with _pipeline_lock:
+            loop = asyncio.get_event_loop()
+            try:
+                await loop.run_in_executor(None, lambda: run_auto_pipeline(trigger="n8n"))
+            except Exception as exc:
+                logging.getLogger(__name__).error("Background pipeline error: %s", exc)
+
+    asyncio.create_task(_run())
+    return {"ok": True, "message": "Pipeline gestartet – Ergebnisse kommen per Telegram"}
+
+
+@app.post("/api/n8n/ingest")
+def api_n8n_ingest(request: Request) -> dict:
+    """Run only the ingestion step (no rewrite/publish). For N8N."""
+    _require_api_key(request)
+    stats = run_ingestion()
+    return {
+        "ok": stats.status == "success",
+        "stats": {
+            "feeds_processed": stats.feeds_processed,
+            "entries_seen": stats.entries_seen,
+            "articles_upserted": stats.articles_upserted,
+        },
+    }
+
+
+# ---------------------------------------------------------------------------
+# Telegram Webhook
+# ---------------------------------------------------------------------------
+
+@app.post("/telegram/webhook")
+async def telegram_webhook(request: Request) -> dict:
+    """Receive updates from Telegram Bot API.
+
+    Returns 200 immediately so Telegram never retries the same update.
+    Actual processing runs in a background task.
+    """
+    import asyncio
+    import logging
+
+    # Verify secret token
+    secret = settings.telegram_webhook_secret
+    if secret:
+        incoming = request.headers.get("X-Telegram-Bot-Api-Secret-Token", "")
+        if incoming != secret:
+            raise HTTPException(status_code=status.HTTP_403_FORBIDDEN, detail="Invalid secret")
+
+    body = await request.body()
+    try:
+        update = json.loads(body.decode("utf-8"))
+    except Exception:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="Invalid JSON")
+
+    async def _process():
+        loop = asyncio.get_event_loop()
+        try:
+            await loop.run_in_executor(None, lambda: handle_update(update))
+        except Exception as exc:
+            logging.getLogger(__name__).error("Telegram update handler error: %s", exc)
+
+    asyncio.create_task(_process())
+    return {"ok": True}
+
+
+@app.post("/api/telegram/setup-webhook")
+def api_setup_telegram_webhook(request: Request) -> dict:
+    """Register the Telegram webhook URL. Call once after deployment."""
+    username = require_auth(request)
+    base_url = str(request.base_url).rstrip("/")
+    webhook_url = f"{base_url}/telegram/webhook"
+    result = setup_webhook(webhook_url)
+    return {"ok": True, "webhook_url": webhook_url, "telegram_response": result, "requested_by": username}

backend/app/pipeline.py

@@ -0,0 +1,516 @@
+"""Autonomous RSS-News pipeline.
+
+Full automated flow:
+1. Run RSS ingestion
+2. For each new article:
+   - Auto-select primary image
+   - Score relevance via GPT
+   - < warn threshold: reject (error status) → Telegram rejected summary
+   - warn..auto threshold: Telegram warning with override button
+   - >= auto threshold: rewrite → create WP draft → Telegram notification
+3. Send pipeline summary to Telegram
+"""
+from __future__ import annotations
+
+import json
+import logging
+import time
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from typing import Any
+
+from .config import get_settings
+from .ingestion import run_ingestion
+from .publisher import enqueue_publish, run_publisher
+from .repositories import (
+    ArticleUpsert,
+    get_article_by_id,
+    list_articles,
+    set_article_image_decision,
+    update_article_status,
+    upsert_article as repo_upsert_article,
+)
+from .rewrite import generate_article_tags, merge_generated_tags, rewrite_article_text, score_article_relevance
+from .scheduler import reserve_publish_slot
+from .wordpress import publish_article_draft, selected_image_exists
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class PipelineStats:
+    ingested: int = 0
+    processed: int = 0
+    drafts_created: int = 0
+    rejected: int = 0
+    quality_gate_rejected: int = 0
+    warnings: int = 0
+    errors: int = 0
+    no_image: int = 0
+    rejected_articles: list[dict[str, Any]] = field(default_factory=list)
+
+
+# ---------------------------------------------------------------------------
+# Internal helpers
+# ---------------------------------------------------------------------------
+
+def _auto_select_image(article: dict[str, Any]) -> bool:
+    """Auto-select the primary image from ingestion metadata if not already selected."""
+    meta_json = article.get("meta_json") or "{}"
+    try:
+        meta = json.loads(meta_json)
+    except Exception:
+        return False
+
+    # Already selected?
+    image_review = meta.get("image_review") or {}
+    if isinstance(image_review, dict) and image_review.get("selected_url"):
+        return True
+
+    # Try to get primary from ingestion extraction
+    extraction = meta.get("extraction") or {}
+    image_selection = extraction.get("image_selection") or {}
+    primary = image_selection.get("primary")
+
+    if not primary:
+        # Fallback: use first URL from image_urls_json
+        image_urls_json = article.get("image_urls_json") or "[]"
+        try:
+            urls = json.loads(image_urls_json)
+            if urls:
+                primary = urls[0]
+        except Exception:
+            pass
+
+    if primary:
+        set_article_image_decision(int(article["id"]), primary, "select", actor="pipeline")
+        return True
+    return False
+
+
+def _store_relevance(article_id: int, relevance: dict[str, Any]) -> None:
+    """Persist relevance score and reason in article meta_json and relevance_score column."""
+    article = get_article_by_id(article_id)
+    if not article:
+        return
+    try:
+        meta = json.loads(article.get("meta_json") or "{}")
+    except Exception:
+        meta = {}
+    meta["relevance"] = relevance
+    new_meta = json.dumps(meta, ensure_ascii=False)
+    from .db import get_conn
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE articles SET meta_json = ?, relevance_score = ? WHERE id = ?",
+            (new_meta, relevance.get("score", 0), article_id),
+        )
+
+
+def _do_rewrite_and_draft(article: dict[str, Any]) -> tuple[int, str | None]:
+    """Rewrite article and create WP draft. Returns (wp_post_id, wp_post_url)."""
+    article_id = int(article["id"])
+    settings = get_settings()
+
+    # ── Quality gate 1: raw content length ──────────────────────────────────
+    import re as _re
+    raw_text = _re.sub(r"<[^>]+>", " ", article.get("content_raw") or "")
+    raw_words = len(raw_text.split())
+    if raw_words < settings.pipeline_min_words_raw:
+        note = (
+            f"Zu wenig Rohinhalt: {raw_words} Wörter "
+            f"(Minimum: {settings.pipeline_min_words_raw})"
+        )
+        logger.warning("_do_rewrite_and_draft #%d: %s — überspringe", article_id, note)
+        update_article_status(article_id, "error", actor="pipeline", note=note)
+        raise ValueError(note)
+
+    # Rewrite
+    logger.info("_do_rewrite_and_draft #%d: starte OpenAI-Rewrite (%d Roh-Wörter)", article_id, raw_words)
+    rewritten = rewrite_article_text(article)
+
+    # ── Quality gate 2: rewritten content length ─────────────────────────────
+    rewritten_words = len(rewritten.split())
+    if rewritten_words < settings.pipeline_min_words_rewritten:
+        note = (
+            f"Rewrite zu kurz: {rewritten_words} Wörter "
+            f"(Minimum: {settings.pipeline_min_words_rewritten})"
+        )
+        logger.warning("_do_rewrite_and_draft #%d: %s — überspringe", article_id, note)
+        update_article_status(article_id, "error", actor="pipeline", note=note)
+        raise ValueError(note)
+    logger.info("_do_rewrite_and_draft #%d: Rewrite fertig (%d Wörter), generiere Tags", article_id, len(rewritten.split()))
+    tags: list[str] = []
+    try:
+        tags = generate_article_tags(article, rewritten_text=rewritten)
+    except Exception:
+        pass
+    merged_meta = merge_generated_tags(article.get("meta_json"), tags)
+
+    # Save rewritten content + approved status
+    repo_upsert_article(
+        ArticleUpsert(
+            feed_id=article.get("feed_id"),
+            source_article_id=article.get("source_article_id"),
+            source_hash=article.get("source_hash"),
+            title=article.get("title", ""),
+            source_url=article.get("source_url", ""),
+            canonical_url=article.get("canonical_url"),
+            published_at=article.get("published_at"),
+            author=article.get("author"),
+            summary=article.get("summary"),
+            content_raw=article.get("content_raw"),
+            content_rewritten=rewritten,
+            image_urls_json=article.get("image_urls_json"),
+            press_contact=article.get("press_contact"),
+            source_name_snapshot=article.get("source_name_snapshot"),
+            source_terms_url_snapshot=article.get("source_terms_url_snapshot"),
+            source_license_name_snapshot=article.get("source_license_name_snapshot"),
+            legal_checked=bool(int(article.get("legal_checked", 0))),
+            legal_checked_at=article.get("legal_checked_at"),
+            legal_note=article.get("legal_note"),
+            wp_post_id=article.get("wp_post_id"),
+            wp_post_url=article.get("wp_post_url"),
+            publish_attempts=int(article.get("publish_attempts", 0)),
+            publish_last_error=article.get("publish_last_error"),
+            published_to_wp_at=article.get("published_to_wp_at"),
+            word_count=len(rewritten.split()),
+            status="approved",
+            meta_json=merged_meta,
+        )
+    )
+
+    # Reload after save to get updated meta_json
+    fresh = get_article_by_id(article_id)
+    if not fresh:
+        raise RuntimeError(f"Artikel #{article_id} nach Rewrite nicht gefunden")
+
+    # Ensure a publish slot is reserved — reserve one now if not yet set
+    if not fresh.get("scheduled_publish_at"):
+        from .scheduler import reserve_publish_slot
+        logger.info("_do_rewrite_and_draft #%d: kein Slot gesetzt, reserviere jetzt", article_id)
+        reserve_publish_slot(article_id)
+        fresh = get_article_by_id(article_id)
+        if not fresh:
+            raise RuntimeError(f"Artikel #{article_id} nach Slot-Reservierung nicht gefunden")
+
+    # Create WP draft
+    logger.info("_do_rewrite_and_draft #%d: erstelle/aktualisiere WP Draft (wp_post_id=%s, sched=%s)", article_id, fresh.get("wp_post_id"), fresh.get("scheduled_publish_at"))
+    wp_post_id, wp_post_url = publish_article_draft(fresh)
+    logger.info("_do_rewrite_and_draft #%d: WP Draft fertig (post_id=%s)", article_id, wp_post_id)
+
+    # Update WP info in DB
+    from .repositories import mark_article_publish_result
+    mark_article_publish_result(
+        article_id,
+        wp_post_id=wp_post_id,
+        wp_post_url=wp_post_url,
+        error=None,
+        increment_attempts=True,
+        set_published_status=False,
+    )
+
+    return wp_post_id, wp_post_url
+
+
+# ---------------------------------------------------------------------------
+# Public pipeline functions
+# ---------------------------------------------------------------------------
+
+def run_auto_pipeline(trigger: str = "auto") -> dict[str, Any]:
+    """Run the full automated pipeline and return stats dict."""
+    from . import telegram_bot as tg
+
+    settings = get_settings()
+    stats = PipelineStats()
+
+    tg.notify_pipeline_started(trigger)
+
+    # Step 1: Ingestion
+    try:
+        ingest_result = run_ingestion()
+        stats.ingested = ingest_result.articles_upserted
+    except Exception as exc:
+        tg.notify_error(f"Ingestion fehlgeschlagen: {exc}")
+        logger.error("Ingestion error: %s", exc)
+        stats.errors += 1
+
+    # Step 2: Process new articles
+    new_articles = list_articles(limit=100, status_filter="new")
+
+    for article in new_articles:
+        article_id = int(article["id"])
+        try:
+            _process_article(article, stats, settings)
+        except Exception as exc:
+            logger.error("Fehler bei Artikel #%d: %s", article_id, exc)
+            tg.notify_error(f"Fehler bei Artikel #{article_id} ({article.get('title','?')[:50]}): {exc}")
+            stats.errors += 1
+        # Rate limiting between OpenAI calls
+        time.sleep(1)
+
+    # Step 3: Send rejected summary if any
+    if stats.rejected_articles:
+        try:
+            tg.notify_rejected_summary(stats.rejected_articles)
+        except Exception as exc:
+            logger.warning("Telegram rejected summary fehlgeschlagen: %s", exc)
+
+    # Step 4: Summary
+    result = {
+        "ingested": stats.ingested,
+        "processed": stats.processed,
+        "drafts_created": stats.drafts_created,
+        "rejected": stats.rejected,
+        "quality_gate_rejected": stats.quality_gate_rejected,
+        "no_image": stats.no_image,
+        "warnings": stats.warnings,
+        "errors": stats.errors,
+    }
+    tg.notify_pipeline_done(result)
+    return result
+
+
+def _process_article(article: dict[str, Any], stats: PipelineStats, settings: Any) -> None:
+    """Process a single new article through the pipeline."""
+    from . import telegram_bot as tg
+
+    article_id = int(article["id"])
+
+    # Auto-select image
+    _auto_select_image(article)
+
+    # Reload to get updated image_review
+    article = get_article_by_id(article_id) or article
+
+    # Exclude articles without a usable image
+    try:
+        meta = json.loads(article.get("meta_json") or "{}")
+    except Exception:
+        meta = {}
+    has_image = bool((meta.get("image_review") or {}).get("selected_url"))
+    if not has_image:
+        update_article_status(
+            article_id,
+            "no_image",
+            actor="pipeline",
+            note="Kein Bild vorhanden – Artikel ausgeschlossen",
+        )
+        stats.no_image += 1
+        logger.info("Artikel #%d ausgeschlossen: kein Bild gefunden", article_id)
+        try:
+            tg.send_message(
+                f"🖼️ <b>Kein Bild</b> – Artikel #{article_id} ausgeschlossen\n"
+                f"📰 {(article.get('title') or '')[:80]}"
+            )
+        except Exception:
+            pass
+        return
+
+    # Score relevance
+    try:
+        relevance = score_article_relevance(article)
+    except Exception as exc:
+        logger.warning("Relevanz-Scoring für #%d fehlgeschlagen: %s", article_id, exc)
+        relevance = {"score": 0, "reason": f"Scoring-Fehler: {exc}", "topics": []}
+
+    score = relevance.get("score", 0)
+    reason = relevance.get("reason", "")
+    _store_relevance(article_id, relevance)
+
+    stats.processed += 1
+
+    if score < settings.pipeline_relevance_warn:
+        # Reject
+        update_article_status(
+            article_id,
+            "error",
+            actor="pipeline",
+            note=f"Abgelehnt: Score {score}/100 — {reason}",
+        )
+        stats.rejected += 1
+        # Reload for summary (now has relevance in meta)
+        updated = get_article_by_id(article_id)
+        if updated:
+            stats.rejected_articles.append(updated)
+
+    elif score < settings.pipeline_relevance_auto:
+        # Warning zone: set status to "review" so repeated /run calls don't re-warn
+        update_article_status(
+            article_id,
+            "review",
+            actor="pipeline",
+            note=f"Niedrige Relevanz: Score {score}/100 — {reason}",
+        )
+        stats.warnings += 1
+        try:
+            tg.notify_relevance_warning(article, score, reason)
+        except Exception as exc:
+            logger.warning("Telegram warning für #%d fehlgeschlagen: %s", article_id, exc)
+
+    else:
+        # Auto-process: rewrite + WP draft
+        try:
+            # Reserve publish slot FIRST so it's available when WP draft is created
+            slot = reserve_publish_slot(article_id)
+
+            # Reload article to get updated image_review + scheduled_publish_at
+            fresh = get_article_by_id(article_id)
+            if not fresh:
+                return
+            wp_post_id, wp_post_url = _do_rewrite_and_draft(fresh)
+            stats.drafts_created += 1
+
+            # Reload for notification
+            final = get_article_by_id(article_id)
+            if final:
+                try:
+                    tg.notify_new_draft(final, score=score, suggested_publish_at=slot)
+                except Exception as exc:
+                    logger.warning("Telegram draft-Benachrichtigung für #%d fehlgeschlagen: %s", article_id, exc)
+
+        except ValueError as exc:
+            # Quality gate rejection (too short etc.) — status already set in _do_rewrite_and_draft
+            # Release the reserved slot so it's available for the next article
+            from .scheduler import release_publish_slot
+            release_publish_slot(article_id)
+            # Clean up any stale WP draft from a previous pipeline run
+            stale = get_article_by_id(article_id)
+            if stale and stale.get("wp_post_id"):
+                try:
+                    from .wordpress import delete_wp_post
+                    delete_wp_post(int(stale["wp_post_id"]))
+                    logger.info("Artikel #%d: veralteten WP-Draft #%s gelöscht", article_id, stale["wp_post_id"])
+                except Exception as del_exc:
+                    logger.warning("Artikel #%d: WP-Draft konnte nicht gelöscht werden: %s", article_id, del_exc)
+            stats.quality_gate_rejected += 1
+            logger.info("Artikel #%d wegen Qualitätsprüfung abgelehnt: %s", article_id, exc)
+            # Individual Telegram notification for quality gate rejection
+            try:
+                title = (article.get("title") or "Ohne Titel")[:80]
+                tg.send_message(
+                    f"✂️ <b>Qualitätsprüfung nicht bestanden</b>\n"
+                    f"📰 {title}\n"
+                    f"💯 Score: {score}/100\n"
+                    f"⚠️ {exc}"
+                )
+            except Exception as tg_exc:
+                logger.warning("Telegram QG-Benachrichtigung für #%d fehlgeschlagen: %s", article_id, tg_exc)
+
+        except Exception as exc:
+            logger.error("Draft-Erstellung für #%d fehlgeschlagen: %s", article_id, exc)
+            update_article_status(article_id, "error", actor="pipeline", note=f"Draft-Fehler: {exc}")
+            # Release reserved slot so it's not permanently blocked by a failed article
+            from .scheduler import release_publish_slot
+            release_publish_slot(article_id)
+            raise
+
+
+# ---------------------------------------------------------------------------
+# Callback actions (called from telegram_bot._handle_callback)
+# ---------------------------------------------------------------------------
+
+def rewrite_and_update_draft(article_id: int) -> None:
+    """Rewrite article and update the existing WP draft."""
+    article = get_article_by_id(article_id)
+    if not article:
+        raise RuntimeError(f"Artikel #{article_id} nicht gefunden")
+    _auto_select_image(article)
+    fresh = get_article_by_id(article_id)
+    _do_rewrite_and_draft(fresh)
+
+
+def discard_article(article_id: int) -> None:
+    """Discard a draft: delete WP post if exists, set article to error."""
+    article = get_article_by_id(article_id)
+    if not article:
+        return
+
+    wp_post_id = article.get("wp_post_id")
+    if wp_post_id:
+        try:
+            from .wordpress import delete_wp_post
+            delete_wp_post(int(wp_post_id))
+        except Exception as exc:
+            logger.warning("WP Post #%d konnte nicht gelöscht werden: %s", wp_post_id, exc)
+
+    update_article_status(article_id, "error", actor="telegram", note="Via Telegram verworfen")
+
+
+def override_rejected_article(article_id: int) -> None:
+    """Force-process a previously rejected article."""
+    from . import telegram_bot as tg
+
+    article = get_article_by_id(article_id)
+    if not article:
+        raise RuntimeError(f"Artikel #{article_id} nicht gefunden")
+
+    # Reset to new so processing is allowed
+    update_article_status(article_id, "new", actor="telegram", note="Manuell übernommen via Telegram")
+
+    # Reload
+    fresh = get_article_by_id(article_id)
+    if not fresh:
+        return
+
+    _auto_select_image(fresh)
+    fresh = get_article_by_id(article_id)
+
+    # Get existing score or re-score
+    try:
+        meta = json.loads(fresh.get("meta_json") or "{}")
+        score = int((meta.get("relevance") or {}).get("score", 0))
+    except Exception:
+        score = 0
+
+    # Reserve publish slot FIRST so it's in the DB when WP draft is created
+    slot = reserve_publish_slot(article_id)
+    fresh = get_article_by_id(article_id)
+
+    wp_post_id, wp_post_url = _do_rewrite_and_draft(fresh)
+
+    final = get_article_by_id(article_id)
+    if final:
+        tg.notify_new_draft(final, score=score, suggested_publish_at=slot)
+
+
+# ---------------------------------------------------------------------------
+# Status helpers (used by /status command)
+# ---------------------------------------------------------------------------
+
+def get_recently_rejected(days: int = 3) -> list[dict[str, Any]]:
+    """Return articles rejected in the last N days."""
+    from .db import get_conn
+    from .db import rows_to_dicts
+    cutoff = datetime.now(timezone.utc).isoformat()[:10]
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT id, title, meta_json, source_url, created_at
+            FROM articles
+            WHERE status IN ('error', 'review')
+            AND json_extract(meta_json, '$.relevance.score') IS NOT NULL
+            AND date(updated_at) >= date('now', ?)
+            ORDER BY updated_at DESC
+            LIMIT 20
+            """,
+            (f"-{days} days",),
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def get_pipeline_status_text() -> str:
+    """Return a text summary of current pipeline state."""
+    from .repositories import list_articles as _list
+    new_count = len(_list(limit=500, status_filter="new"))
+    approved_count = len(_list(limit=500, status_filter="approved"))
+    published_count = len(_list(limit=500, status_filter="published"))
+    error_count = len(_list(limit=500, status_filter="error"))
+
+    return (
+        f"📊 <b>Pipeline-Status</b>\n"
+        f"🆕 Neu / wartend: {new_count}\n"
+        f"✅ Draft / freigegeben: {approved_count}\n"
+        f"📢 Veröffentlicht: {published_count}\n"
+        f"🚫 Fehler / abgelehnt: {error_count}"
+    )

backend/app/policy.py

@@ -0,0 +1,35 @@
+from __future__ import annotations
+
+from typing import Any
+
+
+def evaluate_source_policy(source: dict[str, Any] | None) -> list[str]:
+    issues: list[str] = []
+    if not source:
+        issues.append("Keine Quelle zugeordnet")
+        return issues
+
+    risk_level = (source.get("risk_level") or "").strip().lower()
+    if risk_level != "green":
+        issues.append(f"Quelle nicht freigegeben (risk_level={risk_level or 'unset'})")
+
+    terms_url = (source.get("terms_url") or "").strip()
+    if not terms_url:
+        issues.append("terms_url fehlt")
+
+    license_name = (source.get("license_name") or "").strip()
+    if not license_name:
+        issues.append("license_name fehlt")
+
+    last_reviewed_at = (source.get("last_reviewed_at") or "").strip()
+    if not last_reviewed_at:
+        issues.append("last_reviewed_at fehlt")
+
+    if int(source.get("is_enabled", 0) or 0) != 1:
+        issues.append("Quelle ist deaktiviert")
+
+    return issues
+
+
+def is_source_allowed(source: dict[str, Any] | None) -> bool:
+    return len(evaluate_source_policy(source)) == 0

backend/app/publisher.py

@@ -0,0 +1,101 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from .repositories import (
+    claim_next_publish_job,
+    complete_publish_job,
+    create_publish_job,
+    fail_publish_job,
+    get_article_by_id,
+    mark_article_publish_result,
+    PublishJobCreate,
+)
+from .wordpress import publish_article_draft, selected_image_exists
+
+
+@dataclass(frozen=True)
+class PublisherStats:
+    processed: int
+    success: int
+    failed: int
+    requeued: int
+
+
+def enqueue_publish(article_id: int, max_attempts: int = 3) -> int:
+    return create_publish_job(PublishJobCreate(article_id=article_id, max_attempts=max_attempts))
+
+
+def _can_publish(article: dict) -> tuple[bool, str | None]:
+    if article.get("status") not in {"approved", "published"}:
+        return False, "Artikelstatus muss 'publish' sein"
+    if not selected_image_exists(article):
+        return False, "Hauptbild nicht gesetzt"
+    return True, None
+
+
+def run_publisher(max_jobs: int = 10) -> PublisherStats:
+    processed = 0
+    success = 0
+    failed = 0
+    requeued = 0
+
+    for _ in range(max(1, max_jobs)):
+        job = claim_next_publish_job()
+        if not job:
+            break
+        processed += 1
+        job_id = int(job["id"])
+        article_id = int(job["article_id"])
+
+        article = get_article_by_id(article_id)
+        if not article:
+            fail_publish_job(job_id, "Artikel nicht gefunden", requeue=False)
+            failed += 1
+            continue
+
+        allowed, reason = _can_publish(article)
+        if not allowed:
+            fail_publish_job(job_id, reason or "Publish-Bedingungen nicht erfüllt", requeue=False)
+            mark_article_publish_result(
+                article_id,
+                wp_post_id=article.get("wp_post_id"),
+                wp_post_url=article.get("wp_post_url"),
+                error=reason or "blocked",
+                increment_attempts=True,
+                set_published_status=False,
+            )
+            failed += 1
+            continue
+
+        try:
+            wp_post_id, wp_post_url = publish_article_draft(article)
+            complete_publish_job(job_id, wp_post_id=wp_post_id, wp_post_url=wp_post_url)
+            mark_article_publish_result(
+                article_id,
+                wp_post_id=wp_post_id,
+                wp_post_url=wp_post_url,
+                error=None,
+                increment_attempts=True,
+                set_published_status=True,
+            )
+            success += 1
+        except Exception as exc:
+            attempts = int(job.get("attempts", 1))
+            max_attempts = int(job.get("max_attempts", 3))
+            should_requeue = attempts < max_attempts
+            fail_publish_job(job_id, str(exc), requeue=should_requeue)
+            mark_article_publish_result(
+                article_id,
+                wp_post_id=article.get("wp_post_id"),
+                wp_post_url=article.get("wp_post_url"),
+                error=str(exc),
+                increment_attempts=True,
+                set_published_status=False,
+            )
+            if should_requeue:
+                requeued += 1
+            else:
+                failed += 1
+
+    return PublisherStats(processed=processed, success=success, failed=failed, requeued=requeued)

backend/app/relevance.py

@@ -0,0 +1,44 @@
+from __future__ import annotations
+
+from datetime import datetime, timezone
+
+
+def _parse_iso_datetime(value: str | None) -> datetime | None:
+    if not value:
+        return None
+    raw = value.strip()
+    if not raw:
+        return None
+    if raw.endswith("Z"):
+        raw = raw[:-1] + "+00:00"
+    try:
+        parsed = datetime.fromisoformat(raw)
+    except ValueError:
+        return None
+    if parsed.tzinfo is None:
+        parsed = parsed.replace(tzinfo=timezone.utc)
+    return parsed
+
+
+def article_age_days(published_at: str | None, now: datetime | None = None) -> int | None:
+    published = _parse_iso_datetime(published_at)
+    if not published:
+        return None
+    ref = now or datetime.now(timezone.utc)
+    delta = ref - published
+    if delta.total_seconds() < 0:
+        return 0
+    return delta.days
+
+
+def article_relevance(published_at: str | None, now: datetime | None = None) -> str:
+    days = article_age_days(published_at, now=now)
+    if days is None:
+        return "unbekannt"
+    if days <= 2:
+        return "hoch"
+    if days <= 7:
+        return "mittel"
+    if days <= 30:
+        return "niedrig"
+    return "alt"

backend/app/repositories.py

@@ -0,0 +1,855 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+import json
+from datetime import datetime, timezone
+from typing import Any
+
+from .db import get_conn, rows_to_dicts
+
+
+@dataclass(frozen=True)
+class SourceCreate:
+    name: str
+    base_url: str | None
+    terms_url: str | None
+    license_name: str | None
+    risk_level: str
+    is_enabled: bool
+    notes: str | None
+    last_reviewed_at: str | None
+
+
+@dataclass(frozen=True)
+class FeedCreate:
+    name: str
+    url: str
+    source_id: int | None
+    is_enabled: bool
+
+
+@dataclass(frozen=True)
+class SourceUpdate:
+    name: str
+    base_url: str | None
+    terms_url: str | None
+    license_name: str | None
+    risk_level: str
+    is_enabled: bool
+    notes: str | None
+    last_reviewed_at: str | None
+
+
+@dataclass(frozen=True)
+class FeedUpdate:
+    name: str
+    url: str
+    source_id: int | None
+    is_enabled: bool
+
+
+@dataclass(frozen=True)
+class RunCreate:
+    run_type: str
+    status: str
+    details: str | None = None
+
+
+@dataclass(frozen=True)
+class ArticleUpsert:
+    feed_id: int | None
+    source_article_id: str | None
+    source_hash: str | None
+    title: str
+    source_url: str
+    canonical_url: str | None
+    published_at: str | None
+    author: str | None
+    summary: str | None
+    content_raw: str | None
+    content_rewritten: str | None
+    image_urls_json: str | None
+    press_contact: str | None
+    source_name_snapshot: str | None
+    source_terms_url_snapshot: str | None
+    source_license_name_snapshot: str | None
+    legal_checked: bool
+    legal_checked_at: str | None
+    legal_note: str | None
+    wp_post_id: int | None
+    wp_post_url: str | None
+    publish_attempts: int
+    publish_last_error: str | None
+    published_to_wp_at: str | None
+    word_count: int
+    status: str
+    meta_json: str | None
+
+
+@dataclass(frozen=True)
+class PublishJobCreate:
+    article_id: int
+    max_attempts: int = 3
+
+
+def create_source(payload: SourceCreate) -> int:
+    with get_conn() as conn:
+        cur = conn.execute(
+            """
+            INSERT INTO sources (name, base_url, terms_url, license_name, risk_level, is_enabled, notes, last_reviewed_at)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """,
+            (
+                payload.name.strip(),
+                payload.base_url,
+                payload.terms_url,
+                payload.license_name,
+                payload.risk_level,
+                1 if payload.is_enabled else 0,
+                payload.notes,
+                payload.last_reviewed_at,
+            ),
+        )
+        return int(cur.lastrowid)
+
+
+def list_sources() -> list[dict[str, Any]]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT id, name, base_url, terms_url, license_name, risk_level, is_enabled, notes, last_reviewed_at, created_at, updated_at
+            FROM sources
+            ORDER BY id DESC
+            """
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def get_source_by_id(source_id: int) -> dict[str, Any] | None:
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT id, name, base_url, terms_url, license_name, risk_level, is_enabled, notes, last_reviewed_at, created_at, updated_at
+            FROM sources
+            WHERE id = ?
+            """,
+            (source_id,),
+        ).fetchone()
+    return dict(row) if row else None
+
+
+def update_source(source_id: int, payload: SourceUpdate) -> bool:
+    with get_conn() as conn:
+        cur = conn.execute(
+            """
+            UPDATE sources
+            SET name = ?, base_url = ?, terms_url = ?, license_name = ?, risk_level = ?, is_enabled = ?, notes = ?, last_reviewed_at = ?
+            WHERE id = ?
+            """,
+            (
+                payload.name.strip(),
+                payload.base_url,
+                payload.terms_url,
+                payload.license_name,
+                payload.risk_level,
+                1 if payload.is_enabled else 0,
+                payload.notes,
+                payload.last_reviewed_at,
+                source_id,
+            ),
+        )
+    return cur.rowcount > 0
+
+
+def delete_source(source_id: int) -> bool:
+    with get_conn() as conn:
+        cur = conn.execute("DELETE FROM sources WHERE id = ?", (source_id,))
+    return cur.rowcount > 0
+
+
+def create_feed(payload: FeedCreate) -> int:
+    with get_conn() as conn:
+        cur = conn.execute(
+            "INSERT INTO feeds (name, url, source_id, is_enabled) VALUES (?, ?, ?, ?)",
+            (payload.name.strip(), payload.url.strip(), payload.source_id, 1 if payload.is_enabled else 0),
+        )
+        return int(cur.lastrowid)
+
+
+def list_feeds() -> list[dict[str, Any]]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT f.id, f.name, f.url, f.source_id, f.is_enabled, f.etag, f.last_modified, f.last_checked_at,
+                   f.created_at, f.updated_at, s.name AS source_name, s.license_name AS source_license_name,
+                   s.terms_url AS source_terms_url, s.risk_level AS source_risk_level, s.base_url AS source_base_url,
+                   s.last_reviewed_at AS source_last_reviewed_at, s.is_enabled AS source_is_enabled
+            FROM feeds f
+            LEFT JOIN sources s ON s.id = f.source_id
+            ORDER BY f.id DESC
+            """
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def list_enabled_feeds() -> list[dict[str, Any]]:
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT f.id, f.name, f.url, f.source_id, f.is_enabled, f.etag, f.last_modified, f.last_checked_at,
+                   s.name AS source_name, s.license_name AS source_license_name, s.terms_url AS source_terms_url,
+                   s.risk_level AS source_risk_level, s.base_url AS source_base_url,
+                   s.last_reviewed_at AS source_last_reviewed_at, s.is_enabled AS source_is_enabled
+            FROM feeds f
+            LEFT JOIN sources s ON s.id = f.source_id
+            WHERE f.is_enabled = 1
+            ORDER BY f.id ASC
+            """
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def get_feed_by_id(feed_id: int) -> dict[str, Any] | None:
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT f.id, f.name, f.url, f.source_id, f.is_enabled, f.etag, f.last_modified, f.last_checked_at,
+                   s.name AS source_name, s.license_name AS source_license_name, s.terms_url AS source_terms_url,
+                   s.risk_level AS source_risk_level, s.base_url AS source_base_url,
+                   s.last_reviewed_at AS source_last_reviewed_at, s.is_enabled AS source_is_enabled
+            FROM feeds f
+            LEFT JOIN sources s ON s.id = f.source_id
+            WHERE f.id = ?
+            """,
+            (feed_id,),
+        ).fetchone()
+    return dict(row) if row else None
+
+
+def update_feed(feed_id: int, payload: FeedUpdate) -> bool:
+    with get_conn() as conn:
+        cur = conn.execute(
+            """
+            UPDATE feeds
+            SET name = ?, url = ?, source_id = ?, is_enabled = ?
+            WHERE id = ?
+            """,
+            (
+                payload.name.strip(),
+                payload.url.strip(),
+                payload.source_id,
+                1 if payload.is_enabled else 0,
+                feed_id,
+            ),
+        )
+    return cur.rowcount > 0
+
+
+def delete_feed(feed_id: int) -> bool:
+    with get_conn() as conn:
+        cur = conn.execute("DELETE FROM feeds WHERE id = ?", (feed_id,))
+    return cur.rowcount > 0
+
+
+def update_feed_fetch_state(feed_id: int, etag: str | None, last_modified: str | None) -> None:
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE feeds
+            SET etag = ?, last_modified = ?, last_checked_at = datetime('now')
+            WHERE id = ?
+            """,
+            (etag, last_modified, feed_id),
+        )
+
+
+def create_run(payload: RunCreate) -> int:
+    with get_conn() as conn:
+        cur = conn.execute(
+            "INSERT INTO runs (run_type, status, details) VALUES (?, ?, ?)",
+            (payload.run_type, payload.status, payload.details),
+        )
+        return int(cur.lastrowid)
+
+
+def finish_run(run_id: int, status: str, details: str | None = None) -> None:
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE runs
+            SET status = ?, details = ?, finished_at = datetime('now')
+            WHERE id = ?
+            """,
+            (status, details, run_id),
+        )
+
+
+def list_runs(limit: int = 50) -> list[dict[str, Any]]:
+    safe_limit = max(1, min(limit, 500))
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT id, run_type, status, started_at, finished_at, details
+            FROM runs
+            ORDER BY id DESC
+            LIMIT ?
+            """,
+            (safe_limit,),
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def get_run_by_id(run_id: int) -> dict[str, Any] | None:
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT id, run_type, status, started_at, finished_at, details
+            FROM runs
+            WHERE id = ?
+            """,
+            (run_id,),
+        ).fetchone()
+    return dict(row) if row else None
+
+
+def get_article_by_id(article_id: int) -> dict[str, Any] | None:
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT a.id, a.feed_id, a.source_article_id, a.source_hash, a.title, a.source_url, a.canonical_url, a.published_at, a.author,
+                   a.summary, a.content_raw, a.content_rewritten, a.image_urls_json, a.press_contact,
+                   a.source_name_snapshot, a.source_terms_url_snapshot, a.source_license_name_snapshot,
+                   a.legal_checked, a.legal_checked_at, a.legal_note,
+                   a.wp_post_id, a.wp_post_url, a.publish_attempts, a.publish_last_error, a.published_to_wp_at,
+                   a.word_count, a.status, a.meta_json, a.created_at, a.updated_at,
+                   a.scheduled_publish_at
+            FROM articles a
+            WHERE a.id = ?
+            """,
+            (article_id,),
+        ).fetchone()
+    return dict(row) if row else None
+
+
+def _merge_review_event(meta_json: str | None, event: dict[str, Any]) -> str:
+    meta: dict[str, Any] = {}
+    if meta_json:
+        try:
+            meta = json.loads(meta_json)
+            if not isinstance(meta, dict):
+                meta = {}
+        except Exception:
+            meta = {}
+
+    events = meta.get("review_events")
+    if not isinstance(events, list):
+        events = []
+    events.append(event)
+    meta["review_events"] = events
+    return json.dumps(meta, ensure_ascii=False)
+
+
+def _load_meta(meta_json: str | None) -> dict[str, Any]:
+    if not meta_json:
+        return {}
+    try:
+        parsed = json.loads(meta_json)
+        return parsed if isinstance(parsed, dict) else {}
+    except Exception:
+        return {}
+
+
+def update_article_status(
+    article_id: int,
+    new_status: str,
+    *,
+    actor: str | None = None,
+    note: str | None = None,
+    decision: str | None = None,
+) -> bool:
+    article = get_article_by_id(article_id)
+    if not article:
+        return False
+
+    event = {
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "from_status": article.get("status"),
+        "to_status": new_status,
+        "actor": actor or "system",
+        "note": note,
+        "decision": decision,
+    }
+    merged_meta = _merge_review_event(article.get("meta_json"), event)
+
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE articles SET status = ?, meta_json = ? WHERE id = ?",
+            (new_status, merged_meta, article_id),
+        )
+    return True
+
+
+def set_article_legal_review(article_id: int, approved: bool, note: str | None, actor: str | None = None) -> bool:
+    article = get_article_by_id(article_id)
+    if not article:
+        return False
+
+    event = {
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "event": "legal_review",
+        "approved": approved,
+        "actor": actor or "system",
+        "note": note,
+    }
+    merged_meta = _merge_review_event(article.get("meta_json"), event)
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE articles
+            SET legal_checked = ?, legal_checked_at = datetime('now'), legal_note = ?, meta_json = ?
+            WHERE id = ?
+            """,
+            (1 if approved else 0, note, merged_meta, article_id),
+        )
+    return True
+
+
+def set_article_image_decision(article_id: int, image_url: str, action: str, actor: str | None = None) -> bool:
+    article = get_article_by_id(article_id)
+    if not article:
+        return False
+    url = (image_url or "").strip()
+    if not url:
+        return False
+    if action not in {"select", "exclude", "restore"}:
+        return False
+
+    meta = _load_meta(article.get("meta_json"))
+    image_review = meta.get("image_review")
+    if not isinstance(image_review, dict):
+        image_review = {}
+
+    excluded = image_review.get("excluded_urls")
+    if not isinstance(excluded, list):
+        excluded = []
+    excluded_set = {str(item) for item in excluded if item}
+
+    selected_url = image_review.get("selected_url")
+    if not isinstance(selected_url, str):
+        selected_url = None
+
+    if action == "select":
+        selected_url = url
+        excluded_set.discard(url)
+    elif action == "exclude":
+        excluded_set.add(url)
+        if selected_url == url:
+            selected_url = None
+    elif action == "restore":
+        excluded_set.discard(url)
+
+    image_review["selected_url"] = selected_url
+    image_review["excluded_urls"] = sorted(excluded_set)
+    image_review["updated_at"] = datetime.now(timezone.utc).isoformat()
+    image_review["updated_by"] = actor or "system"
+    meta["image_review"] = image_review
+
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE articles SET meta_json = ? WHERE id = ?",
+            (json.dumps(meta, ensure_ascii=False), article_id),
+        )
+    return True
+
+
+def create_publish_job(payload: PublishJobCreate) -> int:
+    with get_conn() as conn:
+        existing = conn.execute(
+            """
+            SELECT id FROM publish_jobs
+            WHERE article_id = ? AND status IN ('queued', 'running')
+            ORDER BY id DESC
+            LIMIT 1
+            """,
+            (payload.article_id,),
+        ).fetchone()
+        if existing:
+            return int(existing["id"])
+
+        cur = conn.execute(
+            """
+            INSERT INTO publish_jobs (article_id, status, attempts, max_attempts)
+            VALUES (?, 'queued', 0, ?)
+            """,
+            (payload.article_id, max(1, payload.max_attempts)),
+        )
+        return int(cur.lastrowid)
+
+
+def list_publish_jobs(limit: int = 100) -> list[dict[str, Any]]:
+    safe_limit = max(1, min(limit, 500))
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT j.id, j.article_id, j.status, j.attempts, j.max_attempts, j.error_message, j.wp_post_id, j.wp_post_url,
+                   j.created_at, j.started_at, j.finished_at, a.title AS article_title
+            FROM publish_jobs j
+            LEFT JOIN articles a ON a.id = j.article_id
+            ORDER BY j.id DESC
+            LIMIT ?
+            """,
+            (safe_limit,),
+        ).fetchall()
+    return rows_to_dicts(rows)
+
+
+def claim_next_publish_job() -> dict[str, Any] | None:
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT id, article_id, status, attempts, max_attempts, error_message, wp_post_id, wp_post_url
+            FROM publish_jobs
+            WHERE status = 'queued' AND attempts < max_attempts
+            ORDER BY id ASC
+            LIMIT 1
+            """
+        ).fetchone()
+        if not row:
+            return None
+        job_id = int(row["id"])
+        conn.execute(
+            """
+            UPDATE publish_jobs
+            SET status = 'running',
+                attempts = attempts + 1,
+                started_at = datetime('now'),
+                finished_at = NULL
+            WHERE id = ?
+            """,
+            (job_id,),
+        )
+        claimed = conn.execute(
+            """
+            SELECT id, article_id, status, attempts, max_attempts, error_message, wp_post_id, wp_post_url
+            FROM publish_jobs
+            WHERE id = ?
+            """,
+            (job_id,),
+        ).fetchone()
+    return dict(claimed) if claimed else None
+
+
+def complete_publish_job(job_id: int, wp_post_id: int | None, wp_post_url: str | None) -> None:
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE publish_jobs
+            SET status = 'success',
+                wp_post_id = ?,
+                wp_post_url = ?,
+                error_message = NULL,
+                finished_at = datetime('now')
+            WHERE id = ?
+            """,
+            (wp_post_id, wp_post_url, job_id),
+        )
+
+
+def fail_publish_job(job_id: int, error_message: str, requeue: bool) -> None:
+    next_status = "queued" if requeue else "failed"
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE publish_jobs
+            SET status = ?,
+                error_message = ?,
+                finished_at = datetime('now')
+            WHERE id = ?
+            """,
+            (next_status, error_message[:2000], job_id),
+        )
+
+
+def mark_article_publish_result(
+    article_id: int,
+    *,
+    wp_post_id: int | None,
+    wp_post_url: str | None,
+    error: str | None,
+    increment_attempts: bool,
+    set_published_status: bool,
+) -> None:
+    with get_conn() as conn:
+        conn.execute(
+            """
+            UPDATE articles
+            SET wp_post_id = ?,
+                wp_post_url = ?,
+                publish_attempts = CASE WHEN ? THEN publish_attempts + 1 ELSE publish_attempts END,
+                publish_last_error = ?,
+                published_to_wp_at = CASE WHEN ? IS NOT NULL THEN datetime('now') ELSE published_to_wp_at END,
+                status = CASE WHEN ? THEN 'published' ELSE status END
+            WHERE id = ?
+            """,
+            (
+                wp_post_id,
+                wp_post_url,
+                1 if increment_attempts else 0,
+                error[:2000] if error else None,
+                wp_post_id,
+                1 if set_published_status else 0,
+                article_id,
+            ),
+        )
+
+
+def _resolve_existing_article_id(payload: ArticleUpsert) -> int | None:
+    with get_conn() as conn:
+        # 1) strongest key: source_url
+        row = conn.execute(
+            "SELECT id FROM articles WHERE source_url = ?",
+            (payload.source_url.strip(),),
+        ).fetchone()
+        if row:
+            return int(row["id"])
+
+        # 2) stable feed+guid combo
+        if payload.feed_id is not None and payload.source_article_id:
+            row = conn.execute(
+                "SELECT id FROM articles WHERE feed_id = ? AND source_article_id = ?",
+                (payload.feed_id, payload.source_article_id),
+            ).fetchone()
+            if row:
+                return int(row["id"])
+
+        # 3) content hash fallback
+        if payload.source_hash:
+            row = conn.execute(
+                "SELECT id FROM articles WHERE source_hash = ?",
+                (payload.source_hash,),
+            ).fetchone()
+            if row:
+                return int(row["id"])
+
+    return None
+
+
+def find_existing_article_for_upsert(payload: ArticleUpsert) -> dict[str, Any] | None:
+    article_id = _resolve_existing_article_id(payload)
+    if article_id is None:
+        return None
+    return get_article_by_id(article_id)
+
+
+def upsert_article(payload: ArticleUpsert) -> int:
+    existing_id = _resolve_existing_article_id(payload)
+    with get_conn() as conn:
+        if existing_id is None:
+            conn.execute(
+                """
+                INSERT INTO articles (
+                    feed_id, source_article_id, source_hash, title, source_url, canonical_url, published_at, author,
+                    summary, content_raw, content_rewritten, image_urls_json, press_contact,
+                    source_name_snapshot, source_terms_url_snapshot, source_license_name_snapshot,
+                    legal_checked, legal_checked_at, legal_note,
+                    wp_post_id, wp_post_url, publish_attempts, publish_last_error, published_to_wp_at,
+                    word_count, status, meta_json
+                ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                """,
+                (
+                    payload.feed_id,
+                    payload.source_article_id,
+                    payload.source_hash,
+                    payload.title.strip(),
+                    payload.source_url.strip(),
+                    payload.canonical_url,
+                    payload.published_at,
+                    payload.author,
+                    payload.summary,
+                    payload.content_raw,
+                    payload.content_rewritten,
+                    payload.image_urls_json,
+                    payload.press_contact,
+                    payload.source_name_snapshot,
+                    payload.source_terms_url_snapshot,
+                    payload.source_license_name_snapshot,
+                    1 if payload.legal_checked else 0,
+                    payload.legal_checked_at,
+                    payload.legal_note,
+                    payload.wp_post_id,
+                    payload.wp_post_url,
+                    payload.publish_attempts,
+                    payload.publish_last_error,
+                    payload.published_to_wp_at,
+                    payload.word_count,
+                    payload.status,
+                    payload.meta_json,
+                ),
+            )
+        else:
+            conn.execute(
+                """
+                UPDATE articles
+                SET
+                    feed_id = ?,
+                    source_article_id = ?,
+                    source_hash = ?,
+                    title = ?,
+                    source_url = ?,
+                    canonical_url = ?,
+                    published_at = ?,
+                    author = ?,
+                    summary = ?,
+                    content_raw = ?,
+                    content_rewritten = ?,
+                    image_urls_json = ?,
+                    press_contact = ?,
+                    source_name_snapshot = ?,
+                    source_terms_url_snapshot = ?,
+                    source_license_name_snapshot = ?,
+                    legal_checked = ?,
+                    legal_checked_at = ?,
+                    legal_note = ?,
+                    wp_post_id = ?,
+                    wp_post_url = ?,
+                    publish_attempts = ?,
+                    publish_last_error = ?,
+                    published_to_wp_at = ?,
+                    word_count = ?,
+                    status = ?,
+                    meta_json = ?
+                WHERE id = ?
+                """,
+                (
+                    payload.feed_id,
+                    payload.source_article_id,
+                    payload.source_hash,
+                    payload.title.strip(),
+                    payload.source_url.strip(),
+                    payload.canonical_url,
+                    payload.published_at,
+                    payload.author,
+                    payload.summary,
+                    payload.content_raw,
+                    payload.content_rewritten,
+                    payload.image_urls_json,
+                    payload.press_contact,
+                    payload.source_name_snapshot,
+                    payload.source_terms_url_snapshot,
+                    payload.source_license_name_snapshot,
+                    1 if payload.legal_checked else 0,
+                    payload.legal_checked_at,
+                    payload.legal_note,
+                    payload.wp_post_id,
+                    payload.wp_post_url,
+                    payload.publish_attempts,
+                    payload.publish_last_error,
+                    payload.published_to_wp_at,
+                    payload.word_count,
+                    payload.status,
+                    payload.meta_json,
+                    existing_id,
+                ),
+            )
+        row = conn.execute("SELECT id FROM articles WHERE source_url = ?", (payload.source_url.strip(),)).fetchone()
+        if row:
+            return int(row["id"])
+    return int(existing_id) if existing_id else 0
+
+
+def list_articles_page(
+    limit: int = 50,
+    offset: int = 0,
+    status_filter: str | None = None,
+    search: str | None = None,
+) -> tuple[list[dict[str, Any]], int]:
+    """Return (articles, total_count) with optional status filter and title search."""
+    safe_limit = max(1, min(limit, 200))
+    safe_offset = max(0, offset)
+
+    conditions: list[str] = []
+    params: list[Any] = []
+    if status_filter:
+        conditions.append("a.status = ?")
+        params.append(status_filter)
+    if search:
+        conditions.append("(a.title LIKE ? OR a.id = ?)")
+        try:
+            params.extend([f"%{search}%", int(search)])
+        except ValueError:
+            params.extend([f"%{search}%", -1])
+
+    where = f"WHERE {' AND '.join(conditions)}" if conditions else ""
+    select = """
+        SELECT a.id, a.title, a.status, a.published_at, a.summary, a.content_raw,
+               a.meta_json, a.wp_post_id, a.wp_post_url, a.scheduled_publish_at,
+               a.word_count, f.name AS feed_name
+        FROM articles a
+        LEFT JOIN feeds f ON f.id = a.feed_id
+    """
+    with get_conn() as conn:
+        total = conn.execute(
+            f"SELECT COUNT(*) FROM articles a {where}", params
+        ).fetchone()[0]
+        rows = conn.execute(
+            f"{select} {where} ORDER BY a.id DESC LIMIT ? OFFSET ?",
+            params + [safe_limit, safe_offset],
+        ).fetchall()
+    return rows_to_dicts(rows), total
+
+
+def bulk_update_wp_post_ids(updates: list[tuple[int, int | None]]) -> int:
+    """Update wp_post_id (and clear stale wp_post_url) for multiple articles.
+
+    Returns the number of rows actually updated.
+    Call sync_db_from_wordpress() afterwards to repopulate wp_post_url and
+    scheduled_publish_at from the live WordPress data.
+    """
+    if not updates:
+        return 0
+    updated = 0
+    with get_conn() as conn:
+        for article_id, new_wp_id in updates:
+            conn.execute(
+                "UPDATE articles SET wp_post_id = ?, wp_post_url = NULL WHERE id = ?",
+                (new_wp_id, article_id),
+            )
+            updated += 1
+    return updated
+
+
+def list_articles(limit: int = 100, status_filter: str | None = None) -> list[dict[str, Any]]:
+    safe_limit = max(1, min(limit, 500))
+    with get_conn() as conn:
+        if status_filter:
+            rows = conn.execute(
+                """
+                SELECT a.id, a.feed_id, a.source_article_id, a.source_hash, a.title, a.source_url, a.canonical_url, a.published_at, a.author,
+                       a.summary, a.content_raw, a.word_count, a.status, a.meta_json, a.created_at, a.updated_at, f.name AS feed_name,
+                       a.image_urls_json, a.press_contact, a.source_name_snapshot, a.source_terms_url_snapshot,
+                       a.source_license_name_snapshot, a.legal_checked, a.legal_checked_at, a.legal_note,
+                       a.wp_post_id, a.wp_post_url, a.publish_attempts, a.publish_last_error, a.published_to_wp_at
+                FROM articles a
+                LEFT JOIN feeds f ON f.id = a.feed_id
+                WHERE a.status = ?
+                ORDER BY a.id DESC
+                LIMIT ?
+                """,
+                (status_filter, safe_limit),
+            ).fetchall()
+        else:
+            rows = conn.execute(
+                """
+                SELECT a.id, a.feed_id, a.source_article_id, a.source_hash, a.title, a.source_url, a.canonical_url, a.published_at, a.author,
+                       a.summary, a.content_raw, a.word_count, a.status, a.meta_json, a.created_at, a.updated_at, f.name AS feed_name,
+                       a.image_urls_json, a.press_contact, a.source_name_snapshot, a.source_terms_url_snapshot,
+                       a.source_license_name_snapshot, a.legal_checked, a.legal_checked_at, a.legal_note,
+                       a.wp_post_id, a.wp_post_url, a.publish_attempts, a.publish_last_error, a.published_to_wp_at
+                FROM articles a
+                LEFT JOIN feeds f ON f.id = a.feed_id
+                ORDER BY a.id DESC
+                LIMIT ?
+                """,
+                (safe_limit,),
+            ).fetchall()
+    return rows_to_dicts(rows)

backend/app/rewrite.py

@@ -0,0 +1,204 @@
+from __future__ import annotations
+
+import json
+import re
+from typing import Any
+from urllib.request import Request, urlopen
+
+from .config import get_settings
+
+
+def _sanitize_source_text(text: str) -> str:
+    raw = (text or "").strip()
+    if not raw:
+        return ""
+
+    lines = [ln.strip() for ln in raw.splitlines() if ln.strip()]
+    if len(lines) > 3:
+        lines = lines[3:]
+
+    joined = "\n".join(lines)
+    # Remove press contact block at end from "Pressekontakt" onward.
+    joined = re.sub(
+        r"\n?\s*Pressekontakt[\s\S]*$",
+        "",
+        joined,
+        flags=re.IGNORECASE,
+    ).strip()
+    return joined
+
+
+def _normalize_tags(tags: list[str], max_tags: int = 8) -> list[str]:
+    out: list[str] = []
+    seen: set[str] = set()
+    for raw in tags:
+        value = re.sub(r"\s+", " ", str(raw or "").strip())
+        value = re.sub(r"^[#\-•\s]+", "", value)
+        value = re.sub(r"[;,.:\s]+$", "", value)
+        if not value:
+            continue
+        if len(value) < 2 or len(value) > 40:
+            continue
+        key = value.casefold()
+        if key in seen:
+            continue
+        seen.add(key)
+        out.append(value)
+        if len(out) >= max_tags:
+            break
+    return out
+
+
+def _openai_chat(system: str, user: str, temperature: float = 0.4) -> str:
+    settings = get_settings()
+    api_key = settings.openai_api_key
+    if not api_key:
+        raise RuntimeError("OPENAI_API_KEY fehlt")
+
+    payload = {
+        "model": settings.openai_model,
+        "temperature": temperature,
+        "messages": [
+            {"role": "system", "content": system},
+            {"role": "user", "content": user},
+        ],
+    }
+    req = Request(
+        url="https://api.openai.com/v1/chat/completions",
+        method="POST",
+        data=json.dumps(payload).encode("utf-8"),
+        headers={
+            "Authorization": f"Bearer {api_key}",
+            "Content-Type": "application/json",
+            "Accept": "application/json",
+        },
+    )
+    with urlopen(req, timeout=60) as resp:
+        raw = resp.read().decode("utf-8", errors="replace")
+    data = json.loads(raw)
+    choices = data.get("choices")
+    if not isinstance(choices, list) or not choices:
+        raise RuntimeError(f"Ungültige OpenAI-Antwort: {data}")
+    message = choices[0].get("message", {})
+    content = message.get("content")
+    if not isinstance(content, str) or not content.strip():
+        raise RuntimeError("OpenAI lieferte keinen Inhalt")
+    return content.strip()
+
+
+def rewrite_article_text(article: dict[str, Any]) -> str:
+    source_text = _sanitize_source_text(article.get("content_raw") or "")
+    if not source_text:
+        source_text = (article.get("summary") or "").strip()
+    if not source_text:
+        raise RuntimeError("Kein Quelltext für Rewrite verfügbar")
+
+    title = (article.get("title") or "").strip()
+    source_name = (article.get("source_name_snapshot") or article.get("author") or "die Quelle").strip()
+    prompt = (
+        "Schreibe den folgenden News-Text neu auf Deutsch in persönlicher Du-Form. "
+        "Stil: ausführlich, gut lesbar, ohne Einleitung mit Datum/Uhrzeit/Firma/Ort, "
+        "ohne Pressekontakt, ohne Quellenblock. "
+        "Nutze klare Absätze und Zwischenüberschriften in HTML (<h2>, <p>, <ul><li> falls passend). "
+        "Inhaltlich korrekt bleiben, nichts erfinden. "
+        f"Wichtig: Der Artikel wurde von '{source_name}' veröffentlicht. "
+        "Verwende NIEMALS 'wir' oder 'ich' aus Sicht der Quelle – beziehe Aussagen stets auf die Quelle, "
+        f"z.B. 'laut {source_name}', '{source_name} hat ermittelt', 'die Auswertung zeigt'.\n\n"
+        f"Titel: {title}\n\n"
+        f"Originaltext:\n{source_text}"
+    )
+    return _openai_chat(
+        "Du bist ein deutscher News-Redakteur.",
+        prompt,
+        temperature=0.4,
+    )
+
+
+def generate_article_tags(article: dict[str, Any], rewritten_text: str | None = None, max_tags: int = 8) -> list[str]:
+    source_text = rewritten_text or _sanitize_source_text(article.get("content_raw") or "") or (article.get("summary") or "")
+    source_text = str(source_text).strip()
+    if not source_text:
+        return []
+    title = (article.get("title") or "").strip()
+    prompt = (
+        "Erzeuge präzise Schlagwörter für einen deutschen News-Artikel. "
+        f"Maximal {max_tags} Tags. Nur relevante Begriffe, keine allgemeinen Wörter wie News/Artikel. "
+        "Gib ausschließlich ein JSON-Array mit Strings zurück, ohne Erklärung.\n\n"
+        f"Titel: {title}\n\n"
+        f"Text:\n{source_text[:3500]}"
+    )
+    raw = _openai_chat(
+        "Du extrahierst präzise, kurze News-Tags auf Deutsch.",
+        prompt,
+        temperature=0.2,
+    )
+    try:
+        parsed = json.loads(raw)
+        if isinstance(parsed, list):
+            return _normalize_tags([str(x) for x in parsed], max_tags=max_tags)
+    except Exception:
+        pass
+    # fallback: extract first JSON-like array if model wrapped output
+    match = re.search(r"\[[\s\S]*\]", raw)
+    if match:
+        try:
+            parsed = json.loads(match.group(0))
+            if isinstance(parsed, list):
+                return _normalize_tags([str(x) for x in parsed], max_tags=max_tags)
+        except Exception:
+            return []
+    return []
+
+
+def score_article_relevance(article: dict[Any, Any]) -> dict[str, Any]:
+    """Score article relevance for VanLife/Camping/Outdoor blog (0-100).
+
+    Returns {"score": int, "reason": str, "topics": list[str]}.
+    Raises RuntimeError on OpenAI failure.
+    """
+    title = (article.get("title") or "").strip()
+    text = _sanitize_source_text(article.get("content_raw") or "")
+    if not text:
+        text = (article.get("summary") or "").strip()
+
+    prompt = (
+        "Bewerte die Relevanz des folgenden Artikels für einen deutschen VanLife-, Camping- und Outdoor-Blog. "
+        "Relevante Themen: Campingplätze, Stellplätze, Wohnmobil, Camper, Van, Roadtrip, "
+        "Outdoor-Ausrüstung, Wandern, Naturreisen, Reise-Tipps für Campende. "
+        "Nicht relevant: allgemeine Nachrichten, Politik, Wirtschaft, Sport (außer Outdoor), Unterhaltung.\n\n"
+        "Antworte NUR mit einem JSON-Objekt:\n"
+        '{"score": <0-100>, "reason": "<kurze Begründung auf Deutsch>", "topics": ["<Thema1>", "<Thema2>"]}\n\n'
+        f"Titel: {title}\n\n"
+        f"Text (Auszug):\n{text[:2000]}"
+    )
+    raw = _openai_chat(
+        "Du bist ein Redakteur für einen VanLife- und Camping-Blog und bewertest Artikelrelevanz.",
+        prompt,
+        temperature=0.1,
+    )
+    try:
+        match = re.search(r"\{[\s\S]*\}", raw)
+        if match:
+            parsed = json.loads(match.group(0))
+            score = max(0, min(100, int(parsed.get("score", 0))))
+            return {
+                "score": score,
+                "reason": str(parsed.get("reason", "")),
+                "topics": [str(t) for t in (parsed.get("topics") or [])],
+            }
+    except Exception:
+        pass
+    return {"score": 0, "reason": "Parsing-Fehler bei Relevanz-Score", "topics": []}
+
+
+def merge_generated_tags(meta_json: str | None, tags: list[str]) -> str:
+    meta: dict[str, Any] = {}
+    if meta_json:
+        try:
+            parsed = json.loads(meta_json)
+            if isinstance(parsed, dict):
+                meta = parsed
+        except Exception:
+            meta = {}
+    meta["generated_tags"] = _normalize_tags(tags)
+    return json.dumps(meta, ensure_ascii=False)

backend/app/scheduler.py

@@ -0,0 +1,336 @@
+"""Smart publishing scheduler.
+
+Calculates suggested publish slots for new WordPress drafts.
+Rules:
+- Maximum N drafts per day (configurable, default 2)
+- Preferred slots: configurable hours (default 09:00 and 14:00 CET)
+- New articles queue up after the last already-scheduled article
+- Checks both local DB AND WordPress future posts to avoid double-booking
+"""
+from __future__ import annotations
+
+import base64
+import json
+import threading
+import urllib.request
+from datetime import date, datetime, timedelta, timezone
+from typing import Any
+
+from .config import get_settings
+from .db import get_conn
+
+# Ensures that concurrent pipeline runs (two threads) never assign the same slot.
+_slot_lock = threading.Lock()
+
+
+# CET offset (UTC+1 winter / UTC+2 summer – fixed +1 for simplicity)
+_CET_OFFSET = timedelta(hours=1)
+
+
+def _today_cet() -> date:
+    return (datetime.now(timezone.utc) + _CET_OFFSET).date()
+
+
+def _preferred_hours() -> list[int]:
+    settings = get_settings()
+    try:
+        return [int(h.strip()) for h in settings.pipeline_publish_hours.split(",") if h.strip()]
+    except Exception:
+        return [9, 14]
+
+
+def _fetch_wp_occupied_slots() -> set[tuple[str, int]]:
+    """Fetch all future-scheduled WordPress posts and return occupied (date_iso, hour) pairs.
+
+    This prevents the scheduler from assigning a slot that is already taken
+    by a WP post that was not created via this pipeline (e.g. manually or via recovery scripts).
+    Returns an empty set on any error so the scheduler degrades gracefully.
+    """
+    settings = get_settings()
+    try:
+        auth = base64.b64encode(
+            f"{settings.wordpress_username}:{settings.wordpress_app_password}".encode()
+        ).decode()
+        url = (
+            f"{settings.wordpress_base_url}/wp-json/wp/v2/posts"
+            f"?status=future&per_page=100&orderby=date&order=asc&_fields=id,date"
+        )
+        req = urllib.request.Request(url, headers={"Authorization": f"Basic {auth}"})
+        with urllib.request.urlopen(req, timeout=10) as resp:
+            posts = json.loads(resp.read())
+        occupied: set[tuple[str, int]] = set()
+        for p in posts:
+            try:
+                dt = datetime.fromisoformat(p["date"])
+                occupied.add((dt.date().isoformat(), dt.hour))
+            except Exception:
+                pass
+        return occupied
+    except Exception:
+        return set()
+
+
+def _get_last_future_scheduled_date(wp_occupied: set[tuple[str, int]]) -> date | None:
+    """Return the date of the latest already-scheduled slot (DB + WP)."""
+    today = _today_cet()
+
+    # Latest from local DB
+    with get_conn() as conn:
+        row = conn.execute(
+            """
+            SELECT MAX(scheduled_publish_at) AS last_slot
+            FROM articles
+            WHERE scheduled_publish_at IS NOT NULL
+            AND scheduled_publish_at >= ?
+            AND status NOT IN ('error', 'no_image')
+            """,
+            (today.isoformat() + "T00:00:00",),
+        ).fetchone()
+    db_last: date | None = None
+    if row and row["last_slot"]:
+        try:
+            db_last = datetime.fromisoformat(row["last_slot"]).date()
+        except Exception:
+            pass
+
+    # Latest from WP
+    wp_last: date | None = None
+    for d_str, _ in wp_occupied:
+        try:
+            d = date.fromisoformat(d_str)
+            if d >= today and (wp_last is None or d > wp_last):
+                wp_last = d
+        except Exception:
+            pass
+
+    if db_last and wp_last:
+        return max(db_last, wp_last)
+    return db_last or wp_last
+
+
+def _next_free_hour(target_date: date, wp_occupied: set[tuple[str, int]]) -> int | None:
+    """Return first preferred hour not yet used on target_date (DB + WP), or None if day is full."""
+    hours = _preferred_hours()
+    date_str = target_date.isoformat()
+
+    # Hours used in local DB
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT scheduled_publish_at FROM articles
+            WHERE scheduled_publish_at >= ? AND scheduled_publish_at < ?
+            AND status NOT IN ('error', 'no_image')
+            """,
+            (date_str + "T00:00:00", date_str + "T23:59:59"),
+        ).fetchall()
+
+    used_hours: set[int] = set()
+    for row in rows:
+        ts = row["scheduled_publish_at"] or ""
+        try:
+            used_hours.add(datetime.fromisoformat(ts).hour)
+        except Exception:
+            pass
+
+    # Hours used in WordPress
+    for d_str, h in wp_occupied:
+        if d_str == date_str:
+            used_hours.add(h)
+
+    for h in hours:
+        if h not in used_hours:
+            return h
+    return None
+
+
+def _format_slot(d: date, hour: int) -> str:
+    weekday_names = ["Mo", "Di", "Mi", "Do", "Fr", "Sa", "So"]
+    wd = weekday_names[d.weekday()]
+    return f"{wd}, {d.strftime('%d.%m.%Y')} um {hour:02d}:00 Uhr"
+
+
+def _find_next_free_slot(
+    wp_occupied: set[tuple[str, int]], lookahead_days: int = 60
+) -> tuple[date, int] | None:
+    """Find the next free (date, hour) slot.
+
+    Starts from tomorrow and scans forward, filling any gaps in the schedule
+    rather than always appending after the last existing post.
+    """
+    today = _today_cet()
+    tomorrow = today + timedelta(days=1)
+
+    for offset in range(0, lookahead_days + 1):
+        candidate = tomorrow + timedelta(days=offset)
+        hour = _next_free_hour(candidate, wp_occupied)
+        if hour is not None:
+            return candidate, hour
+
+    return tomorrow, _preferred_hours()[0] if _preferred_hours() else 9
+
+
+def get_schedule_overview(lookahead_days: int = 60) -> list[dict]:
+    """Return all booked scheduling slots (DB + WP) for the next N days, sorted by date."""
+    today = _today_cet()
+    hours = _preferred_hours()
+
+    # Slots booked in local DB
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT id, title, status, wp_post_id, wp_post_url, scheduled_publish_at
+            FROM articles
+            WHERE scheduled_publish_at IS NOT NULL
+              AND scheduled_publish_at >= ?
+              AND status NOT IN ('error', 'no_image')
+            ORDER BY scheduled_publish_at
+            """,
+            (today.isoformat() + "T00:00:00",),
+        ).fetchall()
+
+    db_slots: dict[tuple[str, int], dict] = {}
+    for row in rows:
+        try:
+            dt = datetime.fromisoformat(row["scheduled_publish_at"])
+            key = (dt.date().isoformat(), dt.hour)
+            db_slots[key] = {
+                "date": dt.date().isoformat(),
+                "hour": dt.hour,
+                "formatted": _format_slot(dt.date(), dt.hour),
+                "source": "db",
+                "article_id": row["id"],
+                "article_title": row["title"],
+                "article_status": row["status"],
+                "wp_post_id": row["wp_post_id"],
+                "wp_post_url": row["wp_post_url"],
+            }
+        except Exception:
+            pass
+
+    # Slots occupied in WordPress but not in local DB
+    wp_occupied = _fetch_wp_occupied_slots()
+    wp_only: list[dict] = []
+    for d_str, h in sorted(wp_occupied):
+        if (d_str, h) in db_slots:
+            continue
+        try:
+            d = date.fromisoformat(d_str)
+            if d >= today:
+                wp_only.append({
+                    "date": d_str,
+                    "hour": h,
+                    "formatted": _format_slot(d, h),
+                    "source": "wordpress",
+                    "article_id": None,
+                    "article_title": "(WP-Beitrag außerhalb Pipeline)",
+                    "article_status": None,
+                    "wp_post_id": None,
+                    "wp_post_url": None,
+                })
+        except Exception:
+            pass
+
+    all_slots = list(db_slots.values()) + wp_only
+    all_slots.sort(key=lambda s: (s["date"], s["hour"]))
+    return all_slots
+
+
+def release_publish_slot(article_id: int) -> None:
+    """Clear a previously reserved slot (e.g. when article is rejected after slot assignment)."""
+    with get_conn() as conn:
+        conn.execute(
+            "UPDATE articles SET scheduled_publish_at = NULL WHERE id = ?",
+            (article_id,),
+        )
+
+
+def suggest_publish_slot() -> str:
+    """Return a suggested publish datetime string (CET) for the next free slot."""
+    wp_occupied = _fetch_wp_occupied_slots()
+    result = _find_next_free_slot(wp_occupied)
+    if result:
+        d, hour = result
+        return _format_slot(d, hour)
+    tomorrow = _today_cet() + timedelta(days=1)
+    return _format_slot(tomorrow, _preferred_hours()[0] if _preferred_hours() else 9)
+
+
+def reserve_publish_slot(article_id: int) -> str:
+    """Reserve a publish slot for an article and persist it in the DB.
+
+    If the article already has a scheduled_publish_at, keep it unchanged.
+    Returns the formatted publish datetime string.
+
+    Uses a module-level lock so that concurrent pipeline runs (two threads)
+    cannot read the same "free" slot and assign it twice.
+    """
+    # Fetch WP-occupied slots BEFORE acquiring the lock — the API call can be slow
+    # and must not block other threads unnecessarily.
+    wp_occupied = _fetch_wp_occupied_slots()
+
+    with _slot_lock:
+        # Single DB connection for the entire read-find-write cycle so the
+        # slot we pick is still free when we write it.
+        with get_conn() as conn:
+            row = conn.execute(
+                "SELECT scheduled_publish_at FROM articles WHERE id = ?",
+                (article_id,),
+            ).fetchone()
+            existing_slot = row["scheduled_publish_at"] if row else None
+            if existing_slot:
+                try:
+                    dt = datetime.fromisoformat(existing_slot)
+                    return _format_slot(dt.date(), dt.hour)
+                except Exception:
+                    pass  # invalid — fall through and assign a fresh slot
+
+            # Find the next free (date, hour) slot using THIS connection so we
+            # see all slots written during this lock window.
+            hours = _preferred_hours()
+            today = _today_cet()
+            tomorrow = today + timedelta(days=1)
+            candidate: date | None = None
+            chosen_hour: int | None = None
+
+            for offset in range(0, 61):
+                d = tomorrow + timedelta(days=offset)
+                date_str = d.isoformat()
+
+                rows = conn.execute(
+                    """
+                    SELECT scheduled_publish_at FROM articles
+                    WHERE scheduled_publish_at >= ? AND scheduled_publish_at < ?
+                    AND status NOT IN ('error', 'no_image')
+                    """,
+                    (date_str + "T00:00:00", date_str + "T23:59:59"),
+                ).fetchall()
+
+                used_hours: set[int] = set()
+                for r in rows:
+                    ts = r["scheduled_publish_at"] or ""
+                    try:
+                        used_hours.add(datetime.fromisoformat(ts).hour)
+                    except Exception:
+                        pass
+                for d_str, h in wp_occupied:
+                    if d_str == date_str:
+                        used_hours.add(h)
+
+                for h in hours:
+                    if h not in used_hours:
+                        candidate = d
+                        chosen_hour = h
+                        break
+                if candidate is not None:
+                    break
+
+            if candidate is None:
+                candidate = tomorrow
+                chosen_hour = hours[0] if hours else 9
+
+            iso_ts = f"{candidate.isoformat()}T{chosen_hour:02d}:00:00"
+            conn.execute(
+                "UPDATE articles SET scheduled_publish_at = ? WHERE id = ?",
+                (iso_ts, article_id),
+            )
+            return _format_slot(candidate, chosen_hour)

backend/app/source_extraction.py

@@ -0,0 +1,442 @@
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from html import unescape
+import re
+from typing import Any
+from urllib.parse import urljoin
+from urllib.request import Request, urlopen
+
+DEFAULT_TIMEOUT_SECONDS = 10
+DEFAULT_USER_AGENT = "rss-news-bot/1.0 (+https://news.vanityontour.de)"
+
+
+@dataclass(frozen=True)
+class ExtractedArticle:
+    title: str | None
+    author: str | None
+    canonical_url: str | None
+    summary: str | None
+    content_text: str | None
+    images: list[str]
+    press_contact: str | None
+    extraction_error: str | None = None
+    image_metadata: dict[str, dict] = field(default_factory=dict)
+
+
+def _clean_text(raw: str | None) -> str | None:
+    if not raw:
+        return None
+    text = unescape(raw)
+    text = re.sub(r"<[^>]+>", " ", text)
+    text = re.sub(r"\s+", " ", text).strip()
+    return text or None
+
+
+def _strip_noise(html: str) -> str:
+    html = re.sub(r"<script[\s\S]*?</script>", " ", html, flags=re.IGNORECASE)
+    html = re.sub(r"<style[\s\S]*?</style>", " ", html, flags=re.IGNORECASE)
+    html = re.sub(r"<noscript[\s\S]*?</noscript>", " ", html, flags=re.IGNORECASE)
+    return html
+
+
+def _meta_content(html: str, attr: str, value: str) -> str | None:
+    pattern = re.compile(
+        rf"<meta[^>]+{attr}\s*=\s*[\"']{re.escape(value)}[\"'][^>]*content\s*=\s*[\"']([^\"']+)[\"'][^>]*>",
+        re.IGNORECASE,
+    )
+    match = pattern.search(html)
+    if match:
+        return _clean_text(match.group(1))
+
+    # handle reversed attribute order
+    pattern_rev = re.compile(
+        rf"<meta[^>]+content\s*=\s*[\"']([^\"']+)[\"'][^>]*{attr}\s*=\s*[\"']{re.escape(value)}[\"'][^>]*>",
+        re.IGNORECASE,
+    )
+    match = pattern_rev.search(html)
+    if match:
+        return _clean_text(match.group(1))
+    return None
+
+
+def _extract_title(html: str) -> str | None:
+    title = _meta_content(html, "property", "og:title")
+    if title:
+        return title
+
+    match = re.search(r"<title[^>]*>([\s\S]*?)</title>", html, re.IGNORECASE)
+    if match:
+        cleaned = _clean_text(match.group(1))
+        if cleaned:
+            return cleaned
+
+    match = re.search(r"<h1[^>]*>([\s\S]*?)</h1>", html, re.IGNORECASE)
+    if match:
+        return _clean_text(match.group(1))
+    return None
+
+
+def _extract_canonical(html: str) -> str | None:
+    match = re.search(
+        r"<link[^>]+rel\s*=\s*[\"']canonical[\"'][^>]*href\s*=\s*[\"']([^\"']+)[\"'][^>]*>",
+        html,
+        re.IGNORECASE,
+    )
+    if match:
+        return _clean_text(match.group(1))
+
+    match = re.search(
+        r"<link[^>]+href\s*=\s*[\"']([^\"']+)[\"'][^>]*rel\s*=\s*[\"']canonical[\"'][^>]*>",
+        html,
+        re.IGNORECASE,
+    )
+    if match:
+        return _clean_text(match.group(1))
+    return None
+
+
+def _extract_author(html: str) -> str | None:
+    for attr, value in (("name", "author"), ("property", "article:author"), ("property", "og:article:author")):
+        author = _meta_content(html, attr, value)
+        if author:
+            return author
+
+    for pattern in (
+        r"(?:Von|Autor(?:in)?)\s*[:\-]\s*([^<\n\r]{3,120})",
+        r"class=[\"'][^\"']*(?:author|byline)[^\"']*[\"'][^>]*>([\s\S]{1,180})<",
+    ):
+        match = re.search(pattern, html, re.IGNORECASE)
+        if match:
+            author = _clean_text(match.group(1))
+            if author:
+                return author
+    return None
+
+
+def _extract_images(html: str, page_url: str) -> list[str]:
+    images: list[str] = []
+    seen: set[str] = set()
+
+    for prop in ("og:image", "twitter:image"):
+        pattern = re.compile(
+            rf"<meta[^>]+property\s*=\s*[\"']{re.escape(prop)}[\"'][^>]*content\s*=\s*[\"']([^\"']+)[\"'][^>]*>",
+            re.IGNORECASE,
+        )
+        for match in pattern.finditer(html):
+            src = match.group(1).strip()
+            abs_src = urljoin(page_url, src)
+            if abs_src not in seen:
+                seen.add(abs_src)
+                images.append(abs_src)
+
+    for match in re.finditer(r"<img[^>]+src\s*=\s*[\"']([^\"']+)[\"'][^>]*>", html, re.IGNORECASE):
+        src = match.group(1).strip()
+        abs_src = urljoin(page_url, src)
+        if abs_src not in seen:
+            seen.add(abs_src)
+            images.append(abs_src)
+
+    return images
+
+
+def _extract_content_text(html: str) -> str | None:
+    section = None
+    for pattern in (
+        r"<article[^>]*>([\s\S]*?)</article>",
+        r"<main[^>]*>([\s\S]*?)</main>",
+        r"<body[^>]*>([\s\S]*?)</body>",
+    ):
+        match = re.search(pattern, html, re.IGNORECASE)
+        if match:
+            section = match.group(1)
+            break
+
+    if not section:
+        section = html
+
+    paragraphs = []
+    for match in re.finditer(r"<h[2-4][^>]*>([\s\S]*?)</h[2-4]>", section, re.IGNORECASE):
+        text = _clean_text(match.group(1))
+        if text and re.search(r"\b(pressekontakt|press contact|kontakt|agentur)\b", text, re.IGNORECASE):
+            paragraphs.append(text)
+
+    for match in re.finditer(r"<p[^>]*>([\s\S]*?)</p>", section, re.IGNORECASE):
+        text = _clean_text(match.group(1))
+        if text and len(text) > 2:
+            paragraphs.append(text)
+
+    if paragraphs:
+        return "\n".join(paragraphs)
+
+    stripped = _clean_text(section)
+    return stripped
+
+
+def _extract_press_contact(content_text: str | None) -> str | None:
+    if not content_text:
+        return None
+
+    lines = [line.strip() for line in content_text.split("\n") if line.strip()]
+    marker_re = re.compile(r"\b(pressekontakt|press contact|presse\-kontakt|agentur)\b", re.IGNORECASE)
+    for idx, line in enumerate(lines):
+        if marker_re.search(line):
+            chunk = [line]
+            for nxt in lines[idx + 1 : idx + 6]:
+                if re.search(r"\b(original\-content von|ots:|newsroom:|alle meldungen|zum newsroom)\b", nxt, re.IGNORECASE):
+                    break
+                chunk.append(nxt)
+            return _clean_text("\n".join(chunk))
+
+    match = re.search(
+        r"((?:Pressekontakt|Agentur)[\s\S]{0,1200}?)(?:Original-Content von|OTS:|newsroom:|Alle Meldungen|Zum Newsroom|$)",
+        content_text,
+        re.IGNORECASE,
+    )
+    if match:
+        return _clean_text(match.group(1))
+    return None
+
+
+# CSS class keywords that indicate a copyright/credit element inside a figcaption
+_CREDIT_CLASS_RE = re.compile(
+    r"class\s*=\s*[\"'][^\"']*(?:copyright|credit|photographer|photo-credit|image-credit|bildrechte|fotocredit)[^\"']*[\"']",
+    re.IGNORECASE,
+)
+
+# Inline text patterns that signal a credit/copyright notice
+_CREDIT_TEXT_RE = re.compile(
+    r"(©[^<\n\r]{1,100}|(?:Foto|Bild|Credit|Fotograf|Fotografie)\s*:[^<\n\r]{1,100})",
+    re.IGNORECASE,
+)
+
+# data-* attribute names that carry credit/caption information directly on <img>
+_IMG_DATA_CREDIT_ATTRS = ("data-credit", "data-photographer", "data-copyright")
+_IMG_DATA_CAPTION_ATTRS = ("data-caption", "data-description")
+
+# Class keywords for adjacent sibling credit spans/divs after an <img>
+_ADJ_CREDIT_CLASS_RE = re.compile(
+    r"class\s*=\s*[\"'][^\"']*(?:copyright|credit|photographer|photo-credit|image-credit|bildrechte|fotocredit)[^\"']*[\"']",
+    re.IGNORECASE,
+)
+
+
+def _extract_image_metadata(html: str, page_url: str) -> dict[str, dict]:
+    """Return a mapping of absolute image URL → {"caption": ..., "credit": ...}.
+
+    Uses three progressive strategies:
+      1. <figure> with <img> + <figcaption>
+      2. data-* attributes on <img> tags not already covered
+      3. <img> tags whose immediately following HTML contains a credit element
+    """
+    result: dict[str, dict] = {}
+
+    try:
+        # ------------------------------------------------------------------
+        # Strategy 1: <figure> blocks containing <img> and <figcaption>
+        # ------------------------------------------------------------------
+        for fig_match in re.finditer(r"<figure[^>]*>([\s\S]*?)</figure>", html, re.IGNORECASE):
+            fig_html = fig_match.group(1)
+
+            # Locate image src (src or lazy-loaded data-src)
+            img_match = re.search(
+                r"<img[^>]+(?:src|data-src)\s*=\s*[\"']([^\"']+)[\"'][^>]*>",
+                fig_html,
+                re.IGNORECASE,
+            )
+            if not img_match:
+                continue
+            img_src = urljoin(page_url, img_match.group(1).strip())
+
+            # Locate figcaption
+            figcap_match = re.search(
+                r"<figcaption[^>]*>([\s\S]*?)</figcaption>",
+                fig_html,
+                re.IGNORECASE,
+            )
+            if not figcap_match:
+                continue
+            figcap_html = figcap_match.group(1)
+
+            # --- Extract credit ---
+            credit: str | None = None
+
+            # Try credit via class attribute on an inner element
+            credit_elem_match = re.search(
+                r"<(?:span|p|div)[^>]*"
+                + _CREDIT_CLASS_RE.pattern
+                + r"[^>]*>([\s\S]*?)</(?:span|p|div)>",
+                figcap_html,
+                re.IGNORECASE,
+            )
+            if credit_elem_match:
+                credit = _clean_text(credit_elem_match.group(1))
+
+            # Fallback: scan plain text of figcaption for credit patterns
+            if not credit:
+                figcap_text = unescape(re.sub(r"<[^>]+>", " ", figcap_html))
+                cred_text_match = _CREDIT_TEXT_RE.search(figcap_text)
+                if cred_text_match:
+                    credit = _clean_text(cred_text_match.group(1))
+
+            # --- Extract caption (full figcaption text) ---
+            caption = _clean_text(figcap_html)
+
+            # Only store entries that carry at least one piece of metadata
+            if caption or credit:
+                entry: dict[str, str] = {}
+                if caption:
+                    entry["caption"] = caption
+                if credit:
+                    entry["credit"] = credit
+                result[img_src] = entry
+
+        # ------------------------------------------------------------------
+        # Strategy 2: data-* attributes on <img> tags
+        # ------------------------------------------------------------------
+        for img_match in re.finditer(r"<img([^>]+)>", html, re.IGNORECASE):
+            img_attrs = img_match.group(1)
+
+            # Resolve image URL (prefer src over data-src)
+            src_match = re.search(r'(?:^|\s)src\s*=\s*["\']([^"\']+)["\']', img_attrs, re.IGNORECASE)
+            if not src_match:
+                src_match = re.search(r'data-src\s*=\s*["\']([^"\']+)["\']', img_attrs, re.IGNORECASE)
+            if not src_match:
+                continue
+            img_src = urljoin(page_url, src_match.group(1).strip())
+
+            # Skip images already handled by Strategy 1
+            if img_src in result:
+                continue
+
+            credit: str | None = None
+            caption: str | None = None
+
+            for attr in _IMG_DATA_CREDIT_ATTRS:
+                attr_match = re.search(
+                    rf'{re.escape(attr)}\s*=\s*["\']([^"\']+)["\']',
+                    img_attrs,
+                    re.IGNORECASE,
+                )
+                if attr_match:
+                    credit = _clean_text(attr_match.group(1))
+                    break
+
+            for attr in _IMG_DATA_CAPTION_ATTRS:
+                attr_match = re.search(
+                    rf'{re.escape(attr)}\s*=\s*["\']([^"\']+)["\']',
+                    img_attrs,
+                    re.IGNORECASE,
+                )
+                if attr_match:
+                    caption = _clean_text(attr_match.group(1))
+                    break
+
+            if caption or credit:
+                entry = {}
+                if caption:
+                    entry["caption"] = caption
+                if credit:
+                    entry["credit"] = credit
+                result[img_src] = entry
+
+        # ------------------------------------------------------------------
+        # Strategy 3: <img> followed within 200 chars by a credit element
+        # ------------------------------------------------------------------
+        for img_match in re.finditer(r"<img([^>]+)>", html, re.IGNORECASE):
+            img_attrs = img_match.group(1)
+
+            src_match = re.search(r'(?:^|\s)src\s*=\s*["\']([^"\']+)["\']', img_attrs, re.IGNORECASE)
+            if not src_match:
+                src_match = re.search(r'data-src\s*=\s*["\']([^"\']+)["\']', img_attrs, re.IGNORECASE)
+            if not src_match:
+                continue
+            img_src = urljoin(page_url, src_match.group(1).strip())
+
+            # Skip images already handled by earlier strategies
+            if img_src in result:
+                continue
+
+            # Look at the 200 characters of HTML immediately after the img tag
+            after_start = img_match.end()
+            after_html = html[after_start : after_start + 200]
+
+            adj_match = re.search(
+                r"<(?:span|p|div)[^>]*"
+                + _ADJ_CREDIT_CLASS_RE.pattern
+                + r"[^>]*>([\s\S]*?)</(?:span|p|div)>",
+                after_html,
+                re.IGNORECASE,
+            )
+            if adj_match:
+                credit = _clean_text(adj_match.group(1))
+                if credit:
+                    result[img_src] = {"credit": credit}
+
+    except Exception:
+        return {}
+
+    return result
+
+
+def extract_article(url: str, timeout_seconds: int = DEFAULT_TIMEOUT_SECONDS) -> ExtractedArticle:
+    try:
+        req = Request(
+            url=url,
+            headers={
+                "User-Agent": DEFAULT_USER_AGENT,
+                "Accept-Language": "de-DE,de;q=0.9,en;q=0.8",
+            },
+        )
+        with urlopen(req, timeout=timeout_seconds) as resp:
+            raw = resp.read()
+            charset = resp.headers.get_content_charset() or "utf-8"
+        html = raw.decode(charset, errors="replace")
+    except Exception as exc:
+        return ExtractedArticle(
+            title=None,
+            author=None,
+            canonical_url=None,
+            summary=None,
+            content_text=None,
+            images=[],
+            press_contact=None,
+            extraction_error=str(exc),
+        )
+
+    html = _strip_noise(html)
+    title = _extract_title(html)
+    author = _extract_author(html)
+    canonical_url = _extract_canonical(html)
+    summary = _meta_content(html, "name", "description")
+    content_text = _extract_content_text(html)
+    if not summary and content_text:
+        summary = _clean_text(content_text[:320])
+    images = _extract_images(html, url)
+    press_contact = _extract_press_contact(content_text)
+    image_metadata = _extract_image_metadata(html, url)
+
+    return ExtractedArticle(
+        title=title,
+        author=author,
+        canonical_url=canonical_url,
+        summary=summary,
+        content_text=content_text,
+        images=images,
+        press_contact=press_contact,
+        extraction_error=None,
+        image_metadata=image_metadata,
+    )
+
+
+def extracted_article_to_meta(article: ExtractedArticle) -> dict[str, Any]:
+    return {
+        "title": article.title,
+        "author": article.author,
+        "canonical_url": article.canonical_url,
+        "summary": article.summary,
+        "images": article.images,
+        "press_contact": article.press_contact,
+        "extraction_error": article.extraction_error,
+        "image_metadata": article.image_metadata,
+    }

backend/app/telegram_bot.py

@@ -0,0 +1,474 @@
+"""Telegram Bot integration for RSS-News pipeline notifications and controls."""
+from __future__ import annotations
+
+import json
+import logging
+from typing import Any
+from urllib.error import URLError
+from urllib.parse import urlencode
+from urllib.request import Request, urlopen
+
+from .config import get_settings
+
+logger = logging.getLogger(__name__)
+
+_BASE = "https://api.telegram.org/bot{token}/{method}"
+_N8N_APP_RELEASE_WEBHOOK = "https://n8n.vanityontour.de/webhook/tg-app-release-bot-v1/webhook"
+
+
+# ---------------------------------------------------------------------------
+# Low-level API helpers
+# ---------------------------------------------------------------------------
+
+def _call(method: str, payload: dict[str, Any]) -> dict[str, Any]:
+    settings = get_settings()
+    token = settings.telegram_bot_token
+    if not token:
+        raise RuntimeError("TELEGRAM_BOT_TOKEN nicht konfiguriert")
+    url = _BASE.format(token=token, method=method)
+    data = json.dumps(payload).encode("utf-8")
+    req = Request(
+        url=url,
+        data=data,
+        method="POST",
+        headers={"Content-Type": "application/json", "Accept": "application/json"},
+    )
+    try:
+        with urlopen(req, timeout=15) as resp:
+            raw = resp.read().decode("utf-8", errors="replace")
+        return json.loads(raw)
+    except URLError as exc:
+        logger.error("Telegram API Fehler (%s): %s", method, exc)
+        raise RuntimeError(f"Telegram API Fehler: {exc}") from exc
+
+
+def _chat_id() -> str:
+    settings = get_settings()
+    cid = settings.telegram_chat_id
+    if not cid:
+        raise RuntimeError("TELEGRAM_CHAT_ID nicht konfiguriert")
+    return cid
+
+
+def _inline_keyboard(buttons: list[list[dict[str, str]]]) -> dict:
+    return {"inline_keyboard": buttons}
+
+
+# ---------------------------------------------------------------------------
+# Public send functions
+# ---------------------------------------------------------------------------
+
+def send_message(text: str, reply_markup: dict | None = None, parse_mode: str = "HTML") -> dict:
+    payload: dict[str, Any] = {
+        "chat_id": _chat_id(),
+        "text": text,
+        "parse_mode": parse_mode,
+        "disable_web_page_preview": False,
+    }
+    if reply_markup:
+        payload["reply_markup"] = reply_markup
+    return _call("sendMessage", payload)
+
+
+def send_photo_message(
+    photo_url: str,
+    caption: str,
+    reply_markup: dict | None = None,
+    parse_mode: str = "HTML",
+) -> dict:
+    payload: dict[str, Any] = {
+        "chat_id": _chat_id(),
+        "photo": photo_url,
+        "caption": caption,
+        "parse_mode": parse_mode,
+    }
+    if reply_markup:
+        payload["reply_markup"] = reply_markup
+    try:
+        return _call("sendPhoto", payload)
+    except Exception:
+        # Fall back to text message if photo fails (e.g. image URL no longer valid)
+        return send_message(caption, reply_markup=reply_markup, parse_mode=parse_mode)
+
+
+def answer_callback_query(callback_query_id: str, text: str = "") -> None:
+    try:
+        _call("answerCallbackQuery", {"callback_query_id": callback_query_id, "text": text})
+    except Exception as exc:
+        logger.warning("answerCallbackQuery fehlgeschlagen: %s", exc)
+
+
+def edit_message_reply_markup(chat_id: str, message_id: int, reply_markup: dict | None = None) -> None:
+    payload: dict[str, Any] = {"chat_id": chat_id, "message_id": message_id}
+    if reply_markup:
+        payload["reply_markup"] = reply_markup
+    else:
+        payload["reply_markup"] = {"inline_keyboard": []}
+    try:
+        _call("editMessageReplyMarkup", payload)
+    except Exception as exc:
+        logger.warning("editMessageReplyMarkup fehlgeschlagen: %s", exc)
+
+
+def setup_webhook(webhook_url: str) -> dict:
+    settings = get_settings()
+    payload: dict[str, Any] = {"url": webhook_url, "allowed_updates": ["message", "callback_query"]}
+    if settings.telegram_webhook_secret:
+        payload["secret_token"] = settings.telegram_webhook_secret
+    return _call("setWebhook", payload)
+
+
+def delete_webhook() -> dict:
+    return _call("deleteWebhook", {})
+
+
+def _forward_to_n8n_app_release(update: dict[str, Any]) -> None:
+    """Forward a Telegram update to the N8N App Release webhook."""
+    try:
+        data = json.dumps(update).encode("utf-8")
+        req = Request(
+            url=_N8N_APP_RELEASE_WEBHOOK,
+            data=data,
+            method="POST",
+            headers={"Content-Type": "application/json"},
+        )
+        with urlopen(req, timeout=5) as _:
+            pass
+    except Exception as exc:
+        logger.debug("N8N App-Release-Forward fehlgeschlagen: %s", exc)
+
+
+# ---------------------------------------------------------------------------
+# Notification helpers
+# ---------------------------------------------------------------------------
+
+def _format_tags(meta_json: str | None) -> str:
+    if not meta_json:
+        return ""
+    try:
+        meta = json.loads(meta_json)
+        tags = meta.get("generated_tags") or []
+        if tags:
+            return " ".join(f"#{t.replace(' ', '_')}" for t in tags[:6])
+    except Exception:
+        pass
+    return ""
+
+
+def _score_emoji(score: int) -> str:
+    if score >= 85:
+        return "🟢"
+    if score >= 70:
+        return "🟡"
+    return "🔴"
+
+
+def notify_new_draft(
+    article: dict[str, Any],
+    score: int,
+    suggested_publish_at: str | None = None,
+) -> None:
+    """Send Telegram notification for a newly created WP draft."""
+    title = (article.get("title") or "Ohne Titel").strip()
+    wp_url = article.get("wp_post_url") or ""
+    tags_str = _format_tags(article.get("meta_json"))
+    art_id = article.get("id")
+
+    score_line = f"{_score_emoji(score)} Relevanz-Score: <b>{score}/100</b>"
+    publish_line = f"📅 Vorgeschlagene Veröffentlichung: <b>{suggested_publish_at}</b>" if suggested_publish_at else ""
+    link_line = f'🔗 <a href="{wp_url}">Draft in WordPress öffnen</a>' if wp_url else ""
+    tags_line = f"🏷 {tags_str}" if tags_str else ""
+
+    text_parts = [
+        f"✅ <b>Neuer Draft erstellt</b>",
+        f"📰 <b>{title}</b>",
+        score_line,
+    ]
+    if publish_line:
+        text_parts.append(publish_line)
+    if tags_line:
+        text_parts.append(tags_line)
+    if link_line:
+        text_parts.append(link_line)
+
+    text = "\n".join(text_parts)
+
+    keyboard = _inline_keyboard([
+        [
+            {"text": "✏️ Neu schreiben", "callback_data": f"rewrite:{art_id}"},
+            {"text": "❌ Verwerfen", "callback_data": f"discard:{art_id}"},
+        ]
+    ])
+
+    # Try with image first
+    meta = {}
+    try:
+        meta = json.loads(article.get("meta_json") or "{}")
+    except Exception:
+        pass
+    image_url = None
+    image_review = meta.get("image_review") or {}
+    if isinstance(image_review, dict):
+        image_url = image_review.get("selected_url")
+    if not image_url:
+        image_sel = (meta.get("extraction") or {}).get("image_selection") or {}
+        image_url = image_sel.get("primary")
+
+    if image_url:
+        send_photo_message(image_url, caption=text, reply_markup=keyboard)
+    else:
+        send_message(text, reply_markup=keyboard)
+
+
+def notify_relevance_warning(article: dict[str, Any], score: int, reason: str) -> None:
+    """Send Telegram warning for borderline articles (score between warn and auto thresholds)."""
+    title = (article.get("title") or "Ohne Titel").strip()
+    art_id = article.get("id")
+    source_url = article.get("source_url") or ""
+
+    text = (
+        f"⚠️ <b>Artikel mit niedrigem Relevanz-Score</b>\n"
+        f"📰 <b>{title}</b>\n"
+        f"{_score_emoji(score)} Score: <b>{score}/100</b>\n"
+        f"💬 {reason}\n"
+        f'🔗 <a href="{source_url}">Originalartikel</a>'
+    )
+    keyboard = _inline_keyboard([
+        [
+            {"text": "➕ Trotzdem verarbeiten", "callback_data": f"override:{art_id}"},
+            {"text": "❌ Ablehnen", "callback_data": f"reject:{art_id}"},
+        ]
+    ])
+    send_message(text, reply_markup=keyboard)
+
+
+def notify_rejected_summary(articles: list[dict[str, Any]]) -> None:
+    """Send summary of rejected articles for this pipeline run."""
+    if not articles:
+        return
+    lines = [f"🚫 <b>{len(articles)} Artikel abgelehnt (Score &lt; {get_settings().pipeline_relevance_warn})</b>\n"]
+    for art in articles[:10]:
+        title = (art.get("title") or "Ohne Titel")[:60]
+        score = _get_relevance_score(art)
+        reason = _get_rejection_reason(art)
+        art_id = art.get("id")
+        lines.append(f"• <b>{title}</b> (Score: {score}) — {reason}")
+    if len(articles) > 10:
+        lines.append(f"... und {len(articles) - 10} weitere")
+
+    text = "\n".join(lines)
+    # Build override buttons for first 5
+    rows = []
+    for art in articles[:5]:
+        art_id = art.get("id")
+        title = (art.get("title") or "")[:25]
+        rows.append([{"text": f"➕ {title}…", "callback_data": f"override:{art_id}"}])
+
+    keyboard = _inline_keyboard(rows) if rows else None
+    send_message(text, reply_markup=keyboard)
+
+
+def notify_error(message: str) -> None:
+    """Send error alert to Telegram."""
+    try:
+        send_message(f"🔴 <b>Fehler im RSS-Pipeline</b>\n{message}")
+    except Exception as exc:
+        logger.error("Telegram Fehler-Benachrichtigung fehlgeschlagen: %s", exc)
+
+
+def notify_pipeline_started(trigger: str = "auto") -> None:
+    icon = "🤖" if trigger == "auto" else "👤"
+    try:
+        send_message(f"{icon} Pipeline gestartet (Auslöser: {trigger})")
+    except Exception:
+        pass
+
+
+def notify_pipeline_done(stats: dict[str, Any]) -> None:
+    ingested = stats.get("ingested", 0)
+    processed = stats.get("processed", 0)
+    drafts = stats.get("drafts_created", 0)
+    rejected = stats.get("rejected", 0)
+    quality_gate_rejected = stats.get("quality_gate_rejected", 0)
+    no_image = stats.get("no_image", 0)
+    warnings = stats.get("warnings", 0)
+    errors = stats.get("errors", 0)
+
+    lines = [
+        "📊 <b>Pipeline abgeschlossen</b>",
+        f"📥 Neue Artikel importiert: {ingested}",
+        f"⚙️ Verarbeitet: {processed}",
+        f"📝 Drafts erstellt: {drafts}",
+    ]
+    if rejected:
+        lines.append(f"🚫 Abgelehnt (Score): {rejected}")
+    if quality_gate_rejected:
+        lines.append(f"✂️ Qualitätsprüfung: {quality_gate_rejected}")
+    if no_image:
+        lines.append(f"🖼️ Kein Bild: {no_image}")
+    if warnings:
+        lines.append(f"⚠️ Warnungen: {warnings}")
+    if errors:
+        lines.append(f"🔴 Fehler: {errors}")
+
+    try:
+        send_message("\n".join(lines))
+    except Exception:
+        pass
+
+
+# ---------------------------------------------------------------------------
+# Helper to read relevance info from meta_json
+# ---------------------------------------------------------------------------
+
+def _get_relevance_score(article: dict[str, Any]) -> int:
+    try:
+        meta = json.loads(article.get("meta_json") or "{}")
+        return int(meta.get("relevance", {}).get("score", 0))
+    except Exception:
+        return 0
+
+
+def _get_rejection_reason(article: dict[str, Any]) -> str:
+    try:
+        meta = json.loads(article.get("meta_json") or "{}")
+        return str(meta.get("relevance", {}).get("reason", ""))[:80]
+    except Exception:
+        return ""
+
+
+# ---------------------------------------------------------------------------
+# Incoming update handler (called by webhook endpoint)
+# ---------------------------------------------------------------------------
+
+def handle_update(update: dict[str, Any]) -> None:
+    """Process an incoming Telegram update."""
+    # Import here to avoid circular imports
+    from . import pipeline as _pipeline
+
+    if "callback_query" in update:
+        _handle_callback(update["callback_query"])
+    elif "message" in update:
+        _handle_message(update["message"])
+
+
+def _handle_message(message: dict[str, Any]) -> None:
+    from . import pipeline as _pipeline
+
+    text = (message.get("text") or "").strip()
+    if not text.startswith("/"):
+        return
+
+    cmd = text.split()[0].lower().lstrip("/")
+    if "@" in cmd:
+        cmd = cmd.split("@")[0]
+
+    if cmd == "run":
+        send_message("🤖 Pipeline wird manuell gestartet …")
+        try:
+            stats = _pipeline.run_auto_pipeline(trigger="manual")
+            notify_pipeline_done(stats)
+        except Exception as exc:
+            notify_error(f"/run fehlgeschlagen: {exc}")
+
+    elif cmd == "rejected":
+        try:
+            articles = _pipeline.get_recently_rejected(days=3)
+            if not articles:
+                send_message("✅ Keine abgelehnten Artikel in den letzten 3 Tagen.")
+            else:
+                notify_rejected_summary(articles)
+        except Exception as exc:
+            notify_error(f"/rejected fehlgeschlagen: {exc}")
+
+    elif cmd == "status":
+        try:
+            status_text = _pipeline.get_pipeline_status_text()
+            send_message(status_text)
+        except Exception as exc:
+            notify_error(f"/status fehlgeschlagen: {exc}")
+
+    elif cmd == "help":
+        send_message(
+            "📋 <b>Verfügbare Befehle</b>\n"
+            "/run — Pipeline manuell starten\n"
+            "/rejected — Abgelehnte Artikel der letzten 3 Tage\n"
+            "/status — Pipeline-Status\n"
+            "/help — Diese Hilfe"
+        )
+
+    else:
+        # Unbekannter Befehl → an N8N App-Release-Workflow weiterleiten
+        _forward_to_n8n_app_release({"message": message})
+
+
+def _handle_callback(callback_query: dict[str, Any]) -> None:
+    from . import pipeline as _pipeline
+    from .repositories import get_article_by_id, update_article_status
+
+    query_id = callback_query.get("id", "")
+    data = (callback_query.get("data") or "").strip()
+    chat_id = str(callback_query.get("message", {}).get("chat", {}).get("id", ""))
+    message_id = int(callback_query.get("message", {}).get("message_id", 0))
+
+    if ":" not in data:
+        answer_callback_query(query_id, "Ungültige Aktion")
+        return
+
+    action, _, raw_id = data.partition(":")
+    try:
+        article_id = int(raw_id)
+    except ValueError:
+        answer_callback_query(query_id, "Ungültige Artikel-ID")
+        return
+
+    article = get_article_by_id(article_id)
+    if not article:
+        answer_callback_query(query_id, "Artikel nicht gefunden")
+        return
+
+    # Answer Telegram immediately so the spinning indicator stops
+    action_labels = {
+        "rewrite": "✏️ Artikel wird neu geschrieben …",
+        "discard": "❌ Artikel verworfen",
+        "override": "➕ Artikel wird verarbeitet …",
+        "reject": "🚫 Abgelehnt",
+    }
+    answer_callback_query(query_id, action_labels.get(action, ""))
+    edit_message_reply_markup(chat_id, message_id)
+
+    logger.info("Callback: action=%s article_id=%s", action, article_id)
+
+    if action == "rewrite":
+        try:
+            logger.info("Rewrite #%d: starte rewrite_and_update_draft", article_id)
+            _pipeline.rewrite_and_update_draft(article_id)
+            logger.info("Rewrite #%d: abgeschlossen, sende Benachrichtigung", article_id)
+            updated = get_article_by_id(article_id)
+            if updated:
+                from .scheduler import suggest_publish_slot
+                slot = suggest_publish_slot()
+                notify_new_draft(updated, score=_get_relevance_score(updated), suggested_publish_at=slot)
+        except Exception as exc:
+            logger.error("Rewrite #%d fehlgeschlagen: %s", article_id, exc, exc_info=True)
+            notify_error(f"Rewrite #{article_id} fehlgeschlagen: {exc}")
+
+    elif action == "discard":
+        try:
+            _pipeline.discard_article(article_id)
+        except Exception as exc:
+            logger.error("Discard #%d fehlgeschlagen: %s", article_id, exc)
+            notify_error(f"Verwerfen #{article_id} fehlgeschlagen: {exc}")
+
+    elif action == "override":
+        try:
+            _pipeline.override_rejected_article(article_id)
+        except Exception as exc:
+            logger.error("Override #%d fehlgeschlagen: %s", article_id, exc)
+            notify_error(f"Override #{article_id} fehlgeschlagen: {exc}")
+
+    elif action == "reject":
+        update_article_status(article_id, "error", actor="telegram", note="Manuell abgelehnt via Telegram")
+
+    else:
+        logger.warning("Unbekannte Callback-Aktion: %s", action)

backend/app/wordpress.py

@@ -0,0 +1,689 @@
+from __future__ import annotations
+
+import base64
+from html import escape
+import logging
+import json
+import mimetypes
+from pathlib import Path
+import re
+from typing import Any
+from html import unescape as _html_unescape
+from urllib.parse import quote_plus, urlparse
+from urllib.request import Request, urlopen
+
+from .config import get_settings
+
+
+def _auth_header(username: str, app_password: str) -> str:
+    token = base64.b64encode(f"{username}:{app_password}".encode("utf-8")).decode("ascii")
+    return f"Basic {token}"
+
+
+def _wp_request(
+    *,
+    base_url: str,
+    auth_header: str,
+    method: str,
+    endpoint: str,
+    payload: dict[str, Any] | None = None,
+) -> Any:
+    url = f"{base_url.rstrip('/')}/wp-json/wp/v2/{endpoint.lstrip('/')}"
+    data = json.dumps(payload).encode("utf-8") if payload is not None else None
+    req = Request(
+        url=url,
+        data=data,
+        method=method,
+        headers={
+            "Authorization": auth_header,
+            "Content-Type": "application/json; charset=utf-8",
+            "Accept": "application/json",
+            "User-Agent": "rss-news-publisher/1.0",
+        },
+    )
+    with urlopen(req, timeout=20) as resp:
+        raw = resp.read().decode("utf-8", errors="replace")
+    return json.loads(raw) if raw else {}
+
+
+def _selected_image_url_from_meta(meta_json: str | None) -> str | None:
+    if not meta_json:
+        return None
+    try:
+        meta = json.loads(meta_json)
+    except Exception:
+        return None
+    if not isinstance(meta, dict):
+        return None
+    image_review = meta.get("image_review")
+    if not isinstance(image_review, dict):
+        return None
+    selected = image_review.get("selected_url")
+    return selected if isinstance(selected, str) and selected.strip() else None
+
+
+def _selected_tags_from_meta(meta_json: str | None) -> list[str]:
+    if not meta_json:
+        return []
+    try:
+        meta = json.loads(meta_json)
+    except Exception:
+        return []
+    if not isinstance(meta, dict):
+        return []
+    raw_tags = meta.get("generated_tags")
+    if not isinstance(raw_tags, list):
+        return []
+    tags: list[str] = []
+    seen: set[str] = set()
+    for item in raw_tags:
+        value = str(item or "").strip()
+        if not value:
+            continue
+        key = value.casefold()
+        if key in seen:
+            continue
+        seen.add(key)
+        tags.append(value)
+        if len(tags) >= 12:
+            break
+    return tags
+
+
+def _resolve_wp_tag_ids(*, base_url: str, auth_header: str, tags: list[str]) -> list[int]:
+    ids: list[int] = []
+    seen: set[int] = set()
+    for tag in tags:
+        name = tag.strip()
+        if not name:
+            continue
+        try:
+            endpoint = f"tags?search={quote_plus(name)}&per_page=20"
+            result = _wp_request(base_url=base_url, auth_header=auth_header, method="GET", endpoint=endpoint)
+            tag_id: int | None = None
+            if isinstance(result, list):
+                for row in result:
+                    if not isinstance(row, dict):
+                        continue
+                    row_name = str(row.get("name") or "")
+                    rid = int(row.get("id", 0) or 0)
+                    if rid <= 0:
+                        continue
+                    if row_name.casefold() == name.casefold():
+                        tag_id = rid
+                        break
+                if tag_id is None:
+                    for row in result:
+                        if isinstance(row, dict) and int(row.get("id", 0) or 0) > 0:
+                            tag_id = int(row.get("id", 0))
+                            break
+            if tag_id is None:
+                created = _wp_request(
+                    base_url=base_url,
+                    auth_header=auth_header,
+                    method="POST",
+                    endpoint="tags",
+                    payload={"name": name},
+                )
+                if isinstance(created, dict):
+                    rid = int(created.get("id", 0) or 0)
+                    if rid > 0:
+                        tag_id = rid
+            if tag_id is not None and tag_id > 0 and tag_id not in seen:
+                seen.add(tag_id)
+                ids.append(tag_id)
+        except Exception:
+            continue
+    return ids
+
+
+_BLOCKED_IMAGE_EXTS = {".svg", ".gif", ".ico", ".webp"}
+_logger = logging.getLogger(__name__)
+
+
+def _sanitize_image_url(url: str) -> str:
+    """Decode HTML entities (e.g. &amp; → &) in image URLs from RSS feeds."""
+    return _html_unescape(url)
+
+
+_PLACEHOLDER_PATTERNS = ("some-default.jpg", "default-image", "placeholder", "no-image", "noimage")
+
+def _is_usable_image_url(url: str) -> bool:
+    """Return False for URLs that are unlikely to work as WP featured images."""
+    if not url or url.startswith("data:"):
+        return False
+    try:
+        path = urlparse(url).path.lower()
+        _, ext = path.rsplit(".", 1) if "." in path else ("", "")
+        if f".{ext}" in _BLOCKED_IMAGE_EXTS:
+            return False
+        if any(p in path for p in _PLACEHOLDER_PATTERNS):
+            return False
+    except Exception:
+        pass
+    return True
+
+
+def _download_image_bytes(url: str, referer: str | None = None) -> tuple[bytes, str]:
+    url = _sanitize_image_url(url)
+    headers = {
+        "User-Agent": "Mozilla/5.0 (compatible; rss-news-publisher/1.0)",
+        "Accept": "image/avif,image/webp,image/apng,image/*,*/*;q=0.8",
+    }
+    if referer:
+        headers["Referer"] = referer
+    req = Request(url=url, headers=headers)
+    with urlopen(req, timeout=20) as resp:
+        raw = resp.read()
+        content_type = resp.headers.get("Content-Type", "application/octet-stream")
+    content_type = content_type.split(";")[0].strip() if content_type else "application/octet-stream"
+    if not content_type.lower().startswith("image/"):
+        raise RuntimeError(f"Ausgewählte Bild-URL liefert kein Bild ({content_type})")
+    return raw, content_type
+
+
+def _guess_filename(image_url: str, content_type: str) -> str:
+    parsed = urlparse(_sanitize_image_url(image_url))
+    stem = Path(parsed.path).name or "article-image"
+    if "." not in stem:
+        ext = mimetypes.guess_extension(content_type.split(";")[0].strip()) or ".jpg"
+        stem = f"{stem}{ext}"
+    # Sanitize to ASCII-safe characters for the HTTP Content-Disposition header
+    stem = stem.encode("ascii", errors="ignore").decode("ascii")
+    stem = re.sub(r"[^\w.\-]", "_", stem) or "article-image.jpg"
+    return stem
+
+
+def _get_image_meta_for_url(meta_json: str | None, image_url: str) -> dict:
+    """Return the caption/credit dict for a specific image URL from extraction metadata."""
+    if not meta_json or not image_url:
+        return {}
+    try:
+        from urllib.parse import urlparse
+        meta = json.loads(meta_json)
+        image_metadata = (meta.get("extraction") or {}).get("image_metadata") or {}
+        # Exact match first
+        if image_url in image_metadata:
+            return image_metadata[image_url]
+        # Fuzzy match: compare without query string (handles ?w=1200 variants)
+        base_url = urlparse(image_url)._replace(query="").geturl()
+        for key, val in image_metadata.items():
+            key_base = urlparse(key)._replace(query="").geturl()
+            if key_base == base_url:
+                return val
+        return {}
+    except Exception:
+        return {}
+
+
+def _build_image_caption(image_meta: dict, source_url: str) -> str:
+    """Build a WP caption string from image metadata and source URL."""
+    # caption from figcaption typically already contains the credit text
+    caption = (image_meta.get("caption") or "").strip()
+    if caption:
+        return caption
+    return f"Quelle: {source_url}"
+
+
+def _upload_featured_media(
+    *,
+    base_url: str,
+    auth_header: str,
+    image_url: str,
+    article_title: str,
+    source_url: str,
+    image_caption: str = "",
+) -> int:
+    image_bytes, content_type = _download_image_bytes(image_url, referer=source_url or None)
+    filename = _guess_filename(image_url, content_type)
+
+    media_url = f"{base_url.rstrip('/')}/wp-json/wp/v2/media"
+    media_req = Request(
+        url=media_url,
+        data=image_bytes,
+        method="POST",
+        headers={
+            "Authorization": auth_header,
+            "Content-Type": content_type,
+            "Content-Disposition": f'attachment; filename="{filename}"',
+            "Accept": "application/json",
+            "User-Agent": "rss-news-publisher/1.0",
+        },
+    )
+    with urlopen(media_req, timeout=30) as resp:
+        media_raw = resp.read().decode("utf-8", errors="replace")
+    media_payload = json.loads(media_raw) if media_raw else {}
+    media_id = int(media_payload.get("id", 0)) if isinstance(media_payload, dict) else 0
+    if media_id <= 0:
+        raise RuntimeError(f"WordPress Media-Upload fehlgeschlagen: {media_payload}")
+
+    _wp_request(
+        base_url=base_url,
+        auth_header=auth_header,
+        method="POST",
+        endpoint=f"media/{media_id}",
+        payload={
+            "title": f"{article_title[:120]} - Bild",
+            "caption": image_caption or f"Quelle: {source_url}",
+            "alt_text": article_title[:200],
+        },
+    )
+    return media_id
+
+
+def _as_paragraph_html(text: str) -> str:
+    chunks = [chunk.strip() for chunk in re.split(r"\n{2,}", text.strip()) if chunk.strip()]
+    if not chunks:
+        return ""
+    lines = []
+    for chunk in chunks:
+        compact = re.sub(r"\s*\n\s*", " ", chunk)
+        lines.append(f"<p>{escape(compact)}</p>")
+    return "\n".join(lines)
+
+
+def _as_block_paragraphs(text: str) -> str:
+    chunks = [chunk.strip() for chunk in re.split(r"\n{2,}", text.strip()) if chunk.strip()]
+    if not chunks:
+        return ""
+    lines = []
+    for chunk in chunks:
+        compact = re.sub(r"\s*\n\s*", " ", chunk)
+        lines.append(f"<!-- wp:paragraph --><p>{escape(compact)}</p><!-- /wp:paragraph -->")
+    return "\n".join(lines)
+
+
+def _strip_html_tags(raw: str) -> str:
+    text = re.sub(r"<[^>]+>", " ", raw or "")
+    return re.sub(r"\s+", " ", text).strip()
+
+
+def _html_to_wp_blocks(html: str) -> str:
+    src = (html or "").strip()
+    if not src:
+        return ""
+    pattern = re.compile(
+        r"<h([2-6])[^>]*>[\s\S]*?</h\1>|<p[^>]*>[\s\S]*?</p>|<ul[^>]*>[\s\S]*?</ul>|<ol[^>]*>[\s\S]*?</ol>",
+        re.IGNORECASE,
+    )
+    blocks: list[str] = []
+    for match in pattern.finditer(src):
+        block_html = match.group(0).strip()
+        if not block_html:
+            continue
+        tag_match = re.match(r"<([a-z0-9]+)", block_html, re.IGNORECASE)
+        tag = (tag_match.group(1).lower() if tag_match else "")
+        if tag == "p":
+            blocks.append(f"<!-- wp:paragraph -->{block_html}<!-- /wp:paragraph -->")
+        elif tag in {"ul", "ol"}:
+            ordered = tag == "ol"
+            if ordered:
+                blocks.append(f'<!-- wp:list {{"ordered":true}} -->{block_html}<!-- /wp:list -->')
+            else:
+                blocks.append(f"<!-- wp:list -->{block_html}<!-- /wp:list -->")
+        elif tag.startswith("h") and len(tag) == 2 and tag[1].isdigit():
+            level = int(tag[1])
+            blocks.append(f'<!-- wp:heading {{"level":{level}}} -->{block_html}<!-- /wp:heading -->')
+    if blocks:
+        return "\n".join(blocks)
+    return _as_block_paragraphs(_strip_html_tags(src))
+
+
+def _as_block_heading(level: int, text: str) -> str:
+    safe_level = min(6, max(1, int(level)))
+    return f'<!-- wp:heading {{"level":{safe_level}}} --><h{safe_level}>{escape(text)}</h{safe_level}><!-- /wp:heading -->'
+
+
+def _as_block_list(items: list[str]) -> str:
+    if not items:
+        return ""
+    content = "".join(f"<li>{item}</li>" for item in items)
+    return f"<!-- wp:list --><ul>{content}</ul><!-- /wp:list -->"
+
+
+def _sanitize_publish_text(text: str) -> str:
+    raw = (text or "").strip()
+    if not raw:
+        return ""
+    lines = [ln.strip() for ln in raw.splitlines() if ln.strip()]
+    if len(lines) > 3:
+        lines = lines[3:]
+    merged = "\n".join(lines)
+    merged = re.sub(r"\n?\s*Pressekontakt[\s\S]*$", "", merged, flags=re.IGNORECASE).strip()
+    return merged
+
+
+def _build_attribution_block(article: dict[str, Any]) -> str:
+    """Build a WP Gutenberg attribution block for the bottom of the article."""
+    from urllib.parse import urlparse
+    source_url = (article.get("canonical_url") or article.get("source_url") or "").strip()
+    source_name = (article.get("source_name_snapshot") or "").strip()
+    author = (article.get("author") or "").strip()
+
+    # If the feed name is "Google Alerts" (or similar generic names), derive the
+    # real source name from the hostname of the canonical URL.
+    if not source_name or source_name.lower() in ("google alerts", "google"):
+        try:
+            hostname = urlparse(source_url).hostname or ""
+            source_name = hostname.removeprefix("www.")
+        except Exception:
+            pass
+
+    # Get image credit from extraction metadata (uses fuzzy URL match)
+    meta_json = article.get("meta_json")
+    credit = ""
+    try:
+        meta = json.loads(meta_json or "{}")
+        selected_url = (meta.get("image_review") or {}).get("selected_url") or ""
+        if selected_url:
+            img_meta = _get_image_meta_for_url(meta_json, selected_url)
+            raw_credit = (img_meta.get("credit") or "").strip()
+            caption_text = (img_meta.get("caption") or "").strip()
+            # If credit is just a bare marker prefix (e.g. "Foto:", "Bild:"),
+            # clear it and extract the full credit from the caption text instead.
+            _BARE_MARKERS = {"foto", "bild", "credit", "fotograf", "fotografie", "photo", "bildnachweis"}
+            if raw_credit.endswith(":") and raw_credit[:-1].strip().lower() in _BARE_MARKERS:
+                raw_credit = ""
+            if raw_credit:
+                credit = raw_credit
+            elif caption_text:
+                # Extract credit markers like "Foto: IMAGO/…", "© Agentur", "Bild: …"
+                import re as _re
+                m = _re.search(
+                    r"(©[^\n]{1,120}|(?:Foto|Bild|Credit|Fotograf|Photo)\s*:[^\n]{1,120})",
+                    caption_text,
+                )
+                credit = m.group(1).strip() if m else ""
+    except Exception:
+        pass
+
+    parts: list[str] = []
+    if source_url:
+        label = source_name or source_url
+        parts.append(f'Originalartikel: <a href="{source_url}">{escape(label)}</a>')
+    if author:
+        parts.append(f"Autor: {escape(author)}")
+    if credit:
+        parts.append(f"Bildnachweis: {escape(credit)}")
+
+    if not parts:
+        return ""
+
+    inner = " &nbsp;|&nbsp; ".join(parts)
+    return (
+        "\n<!-- wp:separator {\"className\":\"is-style-wide\"} -->"
+        "<hr class=\"wp-block-separator is-style-wide\"/><!-- /wp:separator -->\n"
+        f'<!-- wp:paragraph {{\"className\":\"article-attribution\"}} -->'
+        f'<p class="article-attribution"><em>{inner}</em></p>'
+        "<!-- /wp:paragraph -->"
+    )
+
+
+def _build_post_content(article: dict[str, Any]) -> tuple[str, str | None]:
+    summary = (article.get("summary") or "").strip()
+    body_text = (article.get("content_rewritten") or article.get("content_raw") or "").strip()
+    body_text = _sanitize_publish_text(body_text)
+    if not body_text:
+        body_text = summary
+
+    has_html = bool(re.search(r"<[a-zA-Z][^>]*>", body_text))
+    body_html = _html_to_wp_blocks(body_text) if has_html else _as_block_paragraphs(body_text)
+    if not body_html:
+        body_html = "<!-- wp:paragraph --><p>Kein Inhalt verfügbar.</p><!-- /wp:paragraph -->"
+
+    attribution = _build_attribution_block(article)
+    content = (body_html + attribution).strip()
+    return content, None
+
+
+def publish_article_draft(article: dict[str, Any]) -> tuple[int, str | None]:
+    settings = get_settings()
+    if not settings.wordpress_base_url or not settings.wordpress_username or not settings.wordpress_app_password:
+        raise RuntimeError("WordPress Konfiguration fehlt (base_url, username, app_password)")
+
+    auth = _auth_header(settings.wordpress_username, settings.wordpress_app_password)
+
+    title = (article.get("title") or "Ohne Titel").strip()
+    content, excerpt = _build_post_content(article)
+    source_url = article.get("source_url") or ""
+
+    featured_media_id = None
+    selected_image_url = _selected_image_url_from_meta(article.get("meta_json"))
+
+    # Build candidate list: primary selected URL + fallbacks from image_urls_json
+    image_candidates: list[str] = []
+    if selected_image_url and _is_usable_image_url(selected_image_url):
+        image_candidates.append(selected_image_url)
+    try:
+        extra_urls = json.loads(article.get("image_urls_json") or "[]")
+        for u in extra_urls:
+            if u and u not in image_candidates and _is_usable_image_url(u):
+                image_candidates.append(u)
+    except Exception:
+        pass
+
+    for candidate_url in image_candidates:
+        image_meta = _get_image_meta_for_url(article.get("meta_json"), candidate_url)
+        image_caption = _build_image_caption(image_meta, source_url)
+        try:
+            featured_media_id = _upload_featured_media(
+                base_url=settings.wordpress_base_url,
+                auth_header=auth,
+                image_url=candidate_url,
+                article_title=title,
+                source_url=source_url,
+                image_caption=image_caption,
+            )
+            break  # success — stop trying further candidates
+        except Exception as img_exc:
+            _logger.warning(
+                "Bild-Upload fehlgeschlagen, versuche nächste URL: %s — %s", candidate_url, img_exc
+            )
+
+    if not featured_media_id and image_candidates:
+        _logger.warning(
+            "Alle %d Bild-Kandidaten fehlgeschlagen für Artikel #%s (%s)",
+            len(image_candidates), article.get("id"), title[:60],
+        )
+
+    payload = {
+        "title": title,
+        "content": content,
+        "status": settings.wordpress_default_status,
+    }
+    if excerpt:
+        payload["excerpt"] = excerpt
+    if featured_media_id:
+        payload["featured_media"] = featured_media_id
+    scheduled_at = article.get("scheduled_publish_at")
+    if scheduled_at:
+        payload["date"] = scheduled_at  # e.g. "2026-03-24T09:00:00"
+        # Use status "future" so WP schedules auto-publishing at the given date.
+        # WP ignores date for drafts and shows "Sofort veröffentlichen" instead.
+        try:
+            from datetime import datetime as _dt
+            if _dt.fromisoformat(scheduled_at) > _dt.now():
+                payload["status"] = "future"
+        except Exception:
+            pass
+
+    wp_post_id = article.get("wp_post_id")
+    tag_ids = _resolve_wp_tag_ids(
+        base_url=settings.wordpress_base_url,
+        auth_header=auth,
+        tags=_selected_tags_from_meta(article.get("meta_json")),
+    )
+    if tag_ids:
+        payload["tags"] = tag_ids
+
+    if wp_post_id:
+        result = _wp_request(
+            base_url=settings.wordpress_base_url,
+            auth_header=auth,
+            method="POST",
+            endpoint=f"posts/{int(wp_post_id)}",
+            payload=payload,
+        )
+    else:
+        result = _wp_request(
+            base_url=settings.wordpress_base_url,
+            auth_header=auth,
+            method="POST",
+            endpoint="posts",
+            payload=payload,
+        )
+
+    if not isinstance(result, dict):
+        raise RuntimeError(f"WordPress Antwort im unerwarteten Format: {result}")
+    post_id = int(result.get("id", 0))
+    if post_id <= 0:
+        raise RuntimeError(f"WordPress Antwort ohne Post-ID: {result}")
+    post_url = result.get("link")
+    return post_id, post_url if isinstance(post_url, str) else None
+
+
+def selected_image_exists(article: dict[str, Any]) -> bool:
+    return _selected_image_url_from_meta(article.get("meta_json")) is not None
+
+
+def delete_wp_post(wp_post_id: int) -> None:
+    """Permanently delete a WordPress post (moves to trash, then deletes)."""
+    settings = get_settings()
+    if not settings.wordpress_base_url or not settings.wordpress_username or not settings.wordpress_app_password:
+        raise RuntimeError("WordPress Konfiguration fehlt")
+    auth = _auth_header(settings.wordpress_username, settings.wordpress_app_password)
+    # force=true skips trash
+    _wp_request(
+        base_url=settings.wordpress_base_url,
+        auth_header=auth,
+        method="DELETE",
+        endpoint=f"posts/{wp_post_id}?force=true",
+    )
+
+
+def sync_db_from_wordpress() -> dict[str, Any]:
+    """Sync scheduled_publish_at and wp_post_url in the DB from WordPress.
+
+    WordPress is treated as the source of truth for scheduling.
+    For each DB article that has a wp_post_id:
+      - If WP post exists as 'future': update scheduled_publish_at to WP date.
+      - If WP post exists as 'draft': clear scheduled_publish_at (not yet scheduled).
+      - If WP post exists as 'publish': mark article as published in DB.
+      - If WP post is trashed/deleted (404 or trash status): clear wp_post_id,
+        wp_post_url, and scheduled_publish_at so the article can be re-processed.
+    Returns a stats dict with counts of each action taken.
+    """
+    from .db import get_conn
+
+    settings = get_settings()
+    if not settings.wordpress_base_url or not settings.wordpress_username or not settings.wordpress_app_password:
+        raise RuntimeError("WordPress Konfiguration fehlt")
+    auth = _auth_header(settings.wordpress_username, settings.wordpress_app_password)
+    base_url = settings.wordpress_base_url.rstrip("/")
+
+    # Fetch all future + draft + published WP posts in one pass (up to 300 per status)
+    wp_posts: dict[int, dict] = {}
+    for status in ("future", "draft", "publish"):
+        for page in range(1, 4):  # max 300 per status
+            try:
+                result = _wp_request(
+                    base_url=base_url,
+                    auth_header=auth,
+                    method="GET",
+                    endpoint=f"posts?status={status}&per_page=100&page={page}&_fields=id,date,status,link",
+                )
+            except Exception:
+                break
+            if not isinstance(result, list) or not result:
+                break
+            for post in result:
+                try:
+                    wp_posts[int(post["id"])] = post
+                except Exception:
+                    pass
+            if len(result) < 100:
+                break
+
+    # Load all DB articles that have a wp_post_id
+    with get_conn() as conn:
+        rows = conn.execute(
+            """
+            SELECT id, wp_post_id, wp_post_url, scheduled_publish_at, status
+            FROM articles
+            WHERE wp_post_id IS NOT NULL
+              AND status NOT IN ('no_image')
+            ORDER BY id
+            """
+        ).fetchall()
+
+    stats: dict[str, int] = {
+        "total_db_articles": len(rows),
+        "wp_posts_found": len(wp_posts),
+        "slot_updated": 0,
+        "slot_cleared_draft": 0,
+        "marked_published": 0,
+        "wp_reference_cleared": 0,
+        "already_in_sync": 0,
+    }
+
+    for row in rows:
+        article_id = row["id"]
+        wp_post_id = int(row["wp_post_id"])
+        wp_post = wp_posts.get(wp_post_id)
+
+        if wp_post is None:
+            # Post not found in future/draft/publish — likely trashed or deleted
+            # Clear wp reference so article can be re-processed if needed
+            with get_conn() as conn:
+                conn.execute(
+                    """UPDATE articles
+                       SET wp_post_id = NULL, wp_post_url = NULL, scheduled_publish_at = NULL
+                       WHERE id = ?""",
+                    (article_id,),
+                )
+            stats["wp_reference_cleared"] += 1
+            continue
+
+        wp_status = wp_post.get("status", "")
+        wp_date = wp_post.get("date", "")       # local CET datetime, e.g. "2026-05-05T09:00:00"
+        wp_link = wp_post.get("link") or row["wp_post_url"]
+
+        if wp_status == "publish":
+            # Already published in WP — mark as published in DB if not already
+            if row["status"] != "published":
+                with get_conn() as conn:
+                    conn.execute(
+                        "UPDATE articles SET status = 'published', wp_post_url = ? WHERE id = ?",
+                        (wp_link, article_id),
+                    )
+                stats["marked_published"] += 1
+            else:
+                stats["already_in_sync"] += 1
+
+        elif wp_status == "future":
+            # Scheduled — sync the date into scheduled_publish_at
+            current_slot = row["scheduled_publish_at"] or ""
+            # WP returns e.g. "2026-05-05T09:00:00" — compare ignoring seconds
+            if current_slot[:16] != wp_date[:16]:
+                with get_conn() as conn:
+                    conn.execute(
+                        "UPDATE articles SET scheduled_publish_at = ?, wp_post_url = ? WHERE id = ?",
+                        (wp_date, wp_link, article_id),
+                    )
+                stats["slot_updated"] += 1
+            else:
+                stats["already_in_sync"] += 1
+
+        elif wp_status == "draft":
+            # Draft without a schedule — clear scheduled_publish_at if set
+            if row["scheduled_publish_at"]:
+                with get_conn() as conn:
+                    conn.execute(
+                        "UPDATE articles SET scheduled_publish_at = NULL WHERE id = ?",
+                        (article_id,),
+                    )
+                stats["slot_cleared_draft"] += 1
+            else:
+                stats["already_in_sync"] += 1
+
+    return stats

backend/app/workflow.py

@@ -0,0 +1,39 @@
+from __future__ import annotations
+
+UI_STATUSES = ("new", "rewrite", "publish", "published", "close", "no_image")
+
+
+def internal_to_ui_status(status: str | None) -> str:
+    value = (status or "").strip()
+    if value == "approved":
+        return "publish"
+    if value == "error":
+        return "close"
+    if value == "review":
+        return "rewrite"
+    if value in {"new", "rewrite", "published", "no_image"}:
+        return value
+    return value or "new"
+
+
+def ui_to_internal_status(status: str | None) -> str:
+    value = (status or "").strip()
+    if value == "publish":
+        return "approved"
+    if value == "close":
+        return "error"
+    if value in {"new", "rewrite", "published", "no_image"}:
+        return value
+    if value in {"approved", "error", "review"}:
+        return value
+    return value
+
+
+ALLOWED_UI_TRANSITIONS: dict[str, set[str]] = {
+    "new": {"rewrite", "close"},
+    "rewrite": {"publish", "close"},
+    "publish": {"published", "close"},
+    "published": {"rewrite", "close"},
+    "close": {"rewrite"},
+    "no_image": {"rewrite", "close"},
+}

backend/requirements-test.txt

@@ -0,0 +1,3 @@
+pytest==8.3.5
+pytest-cov==6.0.0
+httpx==0.28.1

backend/requirements.txt

@@ -0,0 +1,8 @@
+fastapi==0.116.1
+uvicorn[standard]==0.35.0
+itsdangerous==2.2.0
+pydantic-settings==2.10.1
+python-dotenv==1.1.1
+feedparser==6.0.11
+jinja2==3.1.4
+python-multipart==0.0.20

backend/static/admin.css

@@ -0,0 +1,303 @@
+body {
+  margin: 0;
+  font-family: "Segoe UI", Tahoma, Geneva, Verdana, sans-serif;
+  background: #f4f6f8;
+  color: #1f2937;
+}
+
+.topbar {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  padding: 20px 28px;
+  background: #0f172a;
+  color: #f8fafc;
+}
+
+.container {
+  padding: 20px 28px 28px 28px;
+}
+
+.login {
+  max-width: 520px;
+  margin: 60px auto;
+}
+
+.card {
+  background: #ffffff;
+  border-radius: 10px;
+  box-shadow: 0 1px 3px rgba(0, 0, 0, 0.12);
+  padding: 16px;
+  margin-bottom: 16px;
+}
+
+.stats {
+  display: grid;
+  grid-template-columns: repeat(4, minmax(0, 1fr));
+  gap: 12px;
+  margin-bottom: 16px;
+}
+
+.stat {
+  background: #ffffff;
+  border-radius: 10px;
+  padding: 12px;
+  box-shadow: 0 1px 3px rgba(0, 0, 0, 0.12);
+}
+
+.stat .label {
+  font-size: 12px;
+  color: #64748b;
+}
+
+.stat .value {
+  font-size: 24px;
+  font-weight: 700;
+}
+
+.grid.two {
+  display: grid;
+  grid-template-columns: 1fr 1fr;
+  gap: 16px;
+}
+
+.stack {
+  display: grid;
+  gap: 10px;
+}
+
+.row {
+  display: flex;
+  gap: 8px;
+  align-items: center;
+}
+
+.filter-row {
+  margin-bottom: 10px;
+}
+
+.inline {
+  display: flex;
+  gap: 6px;
+  align-items: center;
+}
+
+table {
+  width: 100%;
+  border-collapse: collapse;
+}
+
+th, td {
+  text-align: left;
+  padding: 8px;
+  border-bottom: 1px solid #e5e7eb;
+  vertical-align: top;
+}
+
+input, select, button, textarea {
+  padding: 8px;
+  border-radius: 6px;
+  border: 1px solid #cbd5e1;
+  font: inherit;
+}
+
+button {
+  background: #0ea5e9;
+  border-color: #0ea5e9;
+  color: white;
+  cursor: pointer;
+}
+
+button.secondary {
+  background: #64748b;
+  border-color: #64748b;
+}
+
+.badge {
+  display: inline-block;
+  padding: 2px 8px;
+  border-radius: 999px;
+  background: #e2e8f0;
+  font-size: 12px;
+}
+
+.badge.ok {
+  background: #dcfce7;
+  color: #166534;
+}
+
+.badge.bad {
+  background: #fee2e2;
+  color: #991b1b;
+}
+
+.badge.errcat {
+  margin-bottom: 4px;
+}
+
+.badge.errcat-policy {
+  background: #fee2e2;
+  color: #991b1b;
+}
+
+.badge.errcat-auth {
+  background: #ffedd5;
+  color: #9a3412;
+}
+
+.badge.errcat-dns {
+  background: #dbeafe;
+  color: #1e40af;
+}
+
+.badge.errcat-media {
+  background: #fef9c3;
+  color: #854d0e;
+}
+
+.badge.errcat-api {
+  background: #ede9fe;
+  color: #5b21b6;
+}
+
+.badge.errcat-unknown {
+  background: #e2e8f0;
+  color: #334155;
+}
+
+.alert {
+  margin-bottom: 12px;
+  padding: 10px;
+  border-radius: 8px;
+  background: #fee2e2;
+  color: #991b1b;
+}
+
+.flash {
+  font-weight: 600;
+}
+
+.flash-success {
+  border-left: 4px solid #10b981;
+}
+
+.flash-error {
+  border-left: 4px solid #ef4444;
+}
+
+.subtle {
+  color: #64748b;
+  font-size: 12px;
+  margin-top: 4px;
+}
+
+.pre {
+  white-space: pre-wrap;
+  line-height: 1.35;
+  max-height: 220px;
+  overflow: auto;
+  background: #f8fafc;
+  border: 1px solid #e2e8f0;
+  border-radius: 8px;
+  padding: 8px;
+  margin-top: 6px;
+}
+
+.linkbtn {
+  display: inline-block;
+  padding: 8px 10px;
+  border-radius: 6px;
+  text-decoration: none;
+  border: 1px solid #cbd5e1;
+  color: #334155;
+  background: #f8fafc;
+}
+
+.detail-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+  gap: 8px 12px;
+  margin-bottom: 10px;
+}
+
+.detail-item {
+  background: #f8fafc;
+  border: 1px solid #e2e8f0;
+  border-radius: 8px;
+  padding: 8px;
+  display: grid;
+  gap: 4px;
+}
+
+.detail-item .k {
+  font-size: 12px;
+  color: #64748b;
+}
+
+.thumb {
+  width: 72px;
+  height: 72px;
+  object-fit: cover;
+  border-radius: 8px;
+  border: 1px solid #cbd5e1;
+  margin-top: 6px;
+}
+
+.image-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fill, minmax(180px, 1fr));
+  gap: 10px;
+}
+
+.image-card {
+  border: 1px solid #e2e8f0;
+  border-radius: 8px;
+  padding: 8px;
+  background: #fff;
+}
+
+.image-card img {
+  width: 100%;
+  height: 120px;
+  object-fit: cover;
+  border-radius: 6px;
+  border: 1px solid #e2e8f0;
+  background: #f8fafc;
+}
+
+.img-failed {
+  opacity: 0.3;
+  filter: grayscale(1);
+}
+
+.image-meta {
+  margin-top: 6px;
+  display: flex;
+  gap: 6px;
+  flex-wrap: wrap;
+}
+
+.image-actions {
+  margin-top: 8px;
+  display: flex;
+  gap: 6px;
+  flex-wrap: wrap;
+}
+
+.image-selected {
+  border-color: #10b981;
+  box-shadow: 0 0 0 1px rgba(16, 185, 129, 0.25);
+}
+
+.image-excluded {
+  opacity: 0.65;
+}
+
+@media (max-width: 920px) {
+  .stats {
+    grid-template-columns: repeat(2, minmax(0, 1fr));
+  }
+  .grid.two {
+    grid-template-columns: 1fr;
+  }
+}

backend/templates/admin_article_detail.html

@@ -0,0 +1,224 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+</head>
+<body>
+  <header class="topbar">
+    <div>
+      <h1>Artikel-Detail #{{ article.id }}</h1>
+      <p>Angemeldet als <strong>{{ user }}</strong></p>
+    </div>
+    <div class="row">
+      <a class="linkbtn" href="/admin/dashboard">Zurück</a>
+      <form method="post" action="/admin/logout">
+        <button type="submit" class="secondary">Logout</button>
+      </form>
+    </div>
+  </header>
+
+  <main class="container">
+    {% if flash_msg %}
+    <section class="card flash {{ 'flash-error' if flash_type == 'error' else 'flash-success' }}">
+      {{ flash_msg }}
+    </section>
+    {% endif %}
+
+    <section class="card">
+      <h2>{{ article.title }}</h2>
+      <div class="detail-grid">
+        <div class="detail-item"><span class="k">Status</span><span><span class="badge">{{ article.status_ui }}</span></span></div>
+        <div class="detail-item"><span class="k">Artikel-Datum</span><span>{{ article.published_at or "-" }}</span></div>
+        <div class="detail-item"><span class="k">Alter</span><span>{{ article.days_old if article.days_old is not none else "-" }} Tage</span></div>
+        <div class="detail-item"><span class="k">Relevanz</span><span>{{ article.relevance }}</span></div>
+        <div class="detail-item"><span class="k">Autor</span><span>{{ article.author or "-" }}</span></div>
+        <div class="detail-item"><span class="k">Feed</span><span>{{ feed.name if feed else "-" }}</span></div>
+        <div class="detail-item"><span class="k">Quelle Snapshot</span><span>{{ article.source_name_snapshot or "-" }}</span></div>
+        <div class="detail-item"><span class="k">Lizenz Snapshot</span><span>{{ article.source_license_name_snapshot or "-" }}</span></div>
+        <div class="detail-item"><span class="k">Terms Snapshot</span><span>{{ article.source_terms_url_snapshot or "-" }}</span></div>
+      </div>
+      <p><strong>Quelle:</strong> <a href="{{ article.source_url }}" target="_blank" rel="noopener">{{ article.source_url }}</a></p>
+      {% if article.canonical_url %}<p><strong>Canonical:</strong> <a href="{{ article.canonical_url }}" target="_blank" rel="noopener">{{ article.canonical_url }}</a></p>{% endif %}
+      {% if article.summary %}
+      <p><strong>Summary:</strong> {{ article.summary }}</p>
+      {% endif %}
+      <p><strong>WordPress Post:</strong>
+        {% if article.wp_post_url %}
+        <a href="{{ article.wp_post_url }}" target="_blank" rel="noopener">#{{ article.wp_post_id }}</a>
+        {% elif article.wp_post_id %}
+        #{{ article.wp_post_id }}
+        {% else %}
+        -
+        {% endif %}
+      </p>
+      <p><strong>Publish Attempts:</strong> {{ article.publish_attempts or 0 }} | <strong>Letzter Fehler:</strong> {{ article.publish_last_error or "-" }}</p>
+    </section>
+
+    <section class="card">
+      <h2>Checkliste</h2>
+      <table>
+        <thead>
+          <tr><th>Kriterium</th><th>Status</th><th>Wert</th></tr>
+        </thead>
+        <tbody>
+          {% for c in checklist %}
+          <tr>
+            <td>{{ c.label }}</td>
+            <td>
+              {% if c.status == "ok" %}
+              <span class="badge ok">OK</span>
+              {% else %}
+              <span class="badge bad">Fehlt</span>
+              {% endif %}
+            </td>
+            <td>{{ c.value }}</td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Extrahierte Daten</h2>
+      <p><strong>Bilder:</strong> {{ article.image_entries|length if article.image_entries else 0 }}</p>
+      {% if article.selected_image_url %}
+      <p><strong>Ausgewähltes Hauptbild:</strong> <a href="{{ article.selected_image_url }}" target="_blank" rel="noopener">{{ article.selected_image_url }}</a></p>
+      {% if article.selected_image_proxy_url %}
+      <img src="{{ article.selected_image_proxy_url }}" alt="Ausgewähltes Hauptbild" class="thumb" loading="lazy" />
+      {% endif %}
+      {% endif %}
+      {% if article.image_entries %}
+      {% if article.image_selection %}
+      <details>
+        <summary>Automatische Bildauswahl (Score + Gründe)</summary>
+        <div class="subtle">Primärbild (Auto): {{ article.image_selection.primary or "-" }}</div>
+        <div class="subtle">Ausgewählt: {{ article.image_selection.selected_count or 0 }} / Kandidaten: {{ article.image_selection.total_candidates or 0 }}</div>
+        {% if article.image_selection.ranked %}
+        <table>
+          <thead>
+            <tr><th>Bild</th><th>Score</th><th>Gründe</th></tr>
+          </thead>
+          <tbody>
+            {% for r in article.image_selection.ranked %}
+            <tr>
+              <td><a href="{{ r.url }}" target="_blank" rel="noopener">{{ r.url }}</a></td>
+              <td>{{ r.score }}</td>
+              <td>{{ r.reasons|join(", ") if r.reasons else "-" }}</td>
+            </tr>
+            {% endfor %}
+          </tbody>
+        </table>
+        {% endif %}
+      </details>
+      {% endif %}
+      <div class="image-grid">
+        {% for image in article.image_entries %}
+        <article class="image-card {{ 'image-selected' if image.is_selected else '' }} {{ 'image-excluded' if image.is_excluded else '' }}">
+          <a href="{{ image.url }}" target="_blank" rel="noopener">
+            <img src="{{ image.proxy_url }}" data-fallback-src="{{ image.url }}" alt="Artikelbild" loading="lazy" onerror="if(!this.dataset.fallbackUsed){this.dataset.fallbackUsed='1';this.src=this.dataset.fallbackSrc;}else{this.classList.add('img-failed');}" />
+          </a>
+          <div class="image-meta">
+            {% if image.is_selected %}<span class="badge ok">Ausgewählt</span>{% endif %}
+            {% if image.is_excluded %}<span class="badge bad">Ausgeblendet</span>{% endif %}
+            {% if image.is_irrelevant_hint %}<span class="badge">evtl. irrelevant</span>{% endif %}
+          </div>
+          <div class="image-actions">
+            <form method="post" action="/admin/articles/{{ article.id }}/images/decision">
+              <input type="hidden" name="image_url" value="{{ image.url }}" />
+              <input type="hidden" name="action" value="select" />
+              <button type="submit">Als Hauptbild</button>
+            </form>
+            {% if not image.is_excluded %}
+            <form method="post" action="/admin/articles/{{ article.id }}/images/decision">
+              <input type="hidden" name="image_url" value="{{ image.url }}" />
+              <input type="hidden" name="action" value="exclude" />
+              <button type="submit" class="secondary">Ausblenden</button>
+            </form>
+            {% else %}
+            <form method="post" action="/admin/articles/{{ article.id }}/images/decision">
+              <input type="hidden" name="image_url" value="{{ image.url }}" />
+              <input type="hidden" name="action" value="restore" />
+              <button type="submit" class="secondary">Einblenden</button>
+            </form>
+            {% endif %}
+          </div>
+          <div class="subtle"><a href="{{ image.url }}" target="_blank" rel="noopener">{{ image.url }}</a></div>
+        </article>
+        {% endfor %}
+      </div>
+      {% endif %}
+      {% if article.press_contact or article.extraction.press_contact %}
+      <p><strong>Pressekontakt</strong></p>
+      <div class="pre">{{ article.press_contact or article.extraction.press_contact }}</div>
+      {% endif %}
+      {% if article.extraction.extraction_error %}
+      <p class="subtle">Extraktionsfehler: {{ article.extraction.extraction_error }}</p>
+      {% endif %}
+    </section>
+
+    <section class="card">
+      <h2>Volltext</h2>
+      <div class="pre">{{ article.content_raw or "-" }}</div>
+    </section>
+
+    <section class="card">
+      <h2>Rewrite-Text (editierbar)</h2>
+      <form method="post" action="/admin/articles/{{ article.id }}/rewrite-save" class="stack">
+        <textarea name="content_rewritten" rows="14" style="width:100%;">{{ article.content_rewritten or "" }}</textarea>
+        <button type="submit">Rewrite-Text speichern</button>
+      </form>
+      {% if article.meta.generated_tags %}
+      <p><strong>Generierte Tags:</strong> {{ article.meta.generated_tags|join("; ") }}</p>
+      {% endif %}
+      <p class="subtle">Dieser Text wird für den WordPress-Entwurf verwendet, falls vorhanden.</p>
+    </section>
+
+    <section class="card">
+      <h2>Status ändern</h2>
+      {% if article.status_ui in ["new", "rewrite"] %}
+      <form method="post" action="/admin/articles/{{ article.id }}/rewrite-run" class="row" style="margin-bottom:8px;">
+        <button type="submit">Rewrite ausführen (OpenAI)</button>
+      </form>
+      {% endif %}
+      {% if article.status_ui == "published" %}
+      <form method="post" action="/admin/articles/{{ article.id }}/reopen" class="row" style="margin-bottom:8px;">
+        <button type="submit">Zurück in Rewrite-Workflow</button>
+      </form>
+      {% endif %}
+      <form method="post" action="/admin/articles/{{ article.id }}/transition" class="row">
+        <select name="target_status">
+          {% for s in allowed_transitions %}
+          <option value="{{ s }}">{{ s }}</option>
+          {% endfor %}
+        </select>
+        <input name="note" placeholder="Notiz" />
+        <button type="submit" class="secondary">Setzen</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>WordPress Publish Queue</h2>
+      {% if article.publish_ready %}
+      <p><span class="badge ok">Publish bereit</span></p>
+      {% else %}
+      <p><span class="badge bad">Publish blockiert</span></p>
+      {% if article.publish_blockers %}
+      <ul>
+        {% for reason in article.publish_blockers %}
+        <li class="subtle">{{ reason }}</li>
+        {% endfor %}
+      </ul>
+      {% endif %}
+      {% endif %}
+      <p class="subtle">Voraussetzungen: Status `publish` und Hauptbild gesetzt.</p>
+      <form method="post" action="/admin/articles/{{ article.id }}/publish-enqueue" class="row">
+        <input name="max_attempts" value="3" />
+        <button type="submit" {% if not article.publish_ready %}disabled{% endif %}>In Queue einreihen</button>
+      </form>
+    </section>
+  </main>
+</body>
+</html>

backend/templates/admin_article_list.html

@@ -0,0 +1,221 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+  <style>
+    .al-table { width: 100%; border-collapse: collapse; }
+    .al-table th, .al-table td { padding: 8px 10px; border-bottom: 1px solid #e5e7eb; vertical-align: middle; }
+    .al-table th { background: #f3f4f6; font-size: 0.85em; text-transform: uppercase; letter-spacing: .04em; }
+    .al-table tr:hover td { background: #fafafa; }
+    .al-thumb { width: 72px; height: 52px; object-fit: cover; border-radius: 4px; display: block; }
+    .al-thumb-placeholder { width: 72px; height: 52px; background: #e5e7eb; border-radius: 4px; display: flex; align-items: center; justify-content: center; color: #9ca3af; font-size: 1.4em; }
+    .al-title { font-weight: 600; font-size: 0.95em; }
+    .al-excerpt { font-size: 0.82em; color: #6b7280; margin-top: 3px; }
+    .wp-id-input { width: 90px; font-family: monospace; font-size: 0.9em; padding: 4px 6px; border: 1px solid #d1d5db; border-radius: 4px; }
+    .wp-id-input.changed { border-color: #f59e0b; background: #fffbeb; font-weight: bold; }
+    .wp-link { font-size: 0.8em; margin-top: 3px; display: block; }
+    .sticky-bar { position: sticky; top: 0; z-index: 100; background: #1e3a5f; color: #fff; padding: 10px 20px; display: flex; align-items: center; gap: 1.5rem; box-shadow: 0 2px 8px rgba(0,0,0,.2); }
+    .sticky-bar button { background: #f59e0b; color: #000; border: none; padding: 8px 20px; border-radius: 6px; font-weight: bold; cursor: pointer; font-size: 0.95em; }
+    .sticky-bar button:disabled { background: #9ca3af; color: #fff; cursor: default; }
+    .change-badge { background: #f59e0b; color: #000; border-radius: 12px; padding: 2px 10px; font-weight: bold; font-size: 0.85em; display: none; }
+    .change-badge.visible { display: inline; }
+    .filter-bar { display: flex; gap: 1rem; align-items: flex-end; flex-wrap: wrap; margin-bottom: 1rem; }
+    .filter-bar label { font-size: 0.85em; color: #6b7280; display: block; margin-bottom: 3px; }
+    .filter-bar input, .filter-bar select { padding: 6px 10px; border: 1px solid #d1d5db; border-radius: 4px; font-size: 0.9em; }
+    .pagination { display: flex; gap: 8px; align-items: center; justify-content: center; margin-top: 1.5rem; flex-wrap: wrap; }
+    .pagination a, .pagination span { padding: 6px 12px; border: 1px solid #d1d5db; border-radius: 4px; font-size: 0.9em; text-decoration: none; color: #374151; }
+    .pagination .current { background: #1e3a5f; color: #fff; border-color: #1e3a5f; font-weight: bold; }
+    .pagination a:hover { background: #f3f4f6; }
+    .badge-sm { padding: 2px 7px; border-radius: 10px; font-size: 0.75em; font-weight: 600; }
+    .badge-new { background: #dbeafe; color: #1e40af; }
+    .badge-approved { background: #d1fae5; color: #065f46; }
+    .badge-error { background: #fee2e2; color: #991b1b; }
+    .badge-published { background: #ede9fe; color: #5b21b6; }
+    .badge-review { background: #fef3c7; color: #92400e; }
+  </style>
+</head>
+<body>
+  <header class="topbar">
+    <div>
+      <h1>Artikelliste</h1>
+      <p>Angemeldet als <strong>{{ user }}</strong></p>
+    </div>
+    <div class="row">
+      <a class="linkbtn" href="/admin/dashboard">Dashboard</a>
+      <a class="linkbtn" href="/admin/schedule">Veröffentlichungsplan</a>
+      <form method="post" action="/admin/logout">
+        <button type="submit" class="secondary">Logout</button>
+      </form>
+    </div>
+  </header>
+
+  <main class="container">
+    {% if flash_msg %}
+    <section class="card flash {{ 'flash-error' if flash_type == 'error' else 'flash-success' }}">
+      {{ flash_msg }}
+    </section>
+    {% endif %}
+
+    <!-- Filter bar (outside main form so it doesn't submit with bulk save) -->
+    <section class="card" style="padding-bottom: 0.5rem;">
+      <form method="get" action="/admin/article-list">
+        <div class="filter-bar">
+          <div>
+            <label>Suche (Titel / ID)</label>
+            <input type="text" name="search" value="{{ search }}" placeholder="z.B. Camping …" />
+          </div>
+          <div>
+            <label>Status</label>
+            <select name="status_filter">
+              <option value="">Alle</option>
+              {% for s in ["new","review","approved","published","error","no_image"] %}
+              <option value="{{ s }}" {% if status_filter == s %}selected{% endif %}>{{ s }}</option>
+              {% endfor %}
+            </select>
+          </div>
+          <div style="padding-bottom:1px;">
+            <button type="submit">Filtern</button>
+            <a href="/admin/article-list" class="linkbtn" style="margin-left:4px;">Reset</a>
+          </div>
+        </div>
+      </form>
+      <p class="subtle" style="margin: 4px 0 0;">{{ total }} Artikel gesamt · Seite {{ page }} / {{ total_pages }} · {{ page_size }} pro Seite</p>
+    </section>
+
+    <!-- Main form for bulk WP ID editing -->
+    <form method="post" action="/admin/article-list/update" id="bulk-form">
+      <!-- Pass filter/page state so redirect goes back to same view -->
+      <input type="hidden" name="page" value="{{ page }}">
+      <input type="hidden" name="status_filter" value="{{ status_filter }}">
+      <input type="hidden" name="search" value="{{ search }}">
+
+      <div class="sticky-bar">
+        <button type="submit" id="save-btn" disabled>💾 Änderungen speichern</button>
+        <span class="change-badge" id="change-badge">0 Änderungen</span>
+        <span style="font-size:0.85em;opacity:.8;">Nur geänderte WP-IDs werden gespeichert. Danach WP-Sync ausführen.</span>
+      </div>
+
+      <section class="card" style="padding: 0; overflow-x: auto;">
+        <table class="al-table">
+          <thead>
+            <tr>
+              <th style="width:80px;">Bild</th>
+              <th>Titel &amp; Kurztext</th>
+              <th style="width:90px;">Status</th>
+              <th style="width:110px;">Datum</th>
+              <th style="width:140px;">WP ID</th>
+            </tr>
+          </thead>
+          <tbody>
+            {% for a in articles %}
+            <tr>
+              <td>
+                {% if a.thumb_proxy %}
+                <a href="{{ a.thumb_url }}" target="_blank" rel="noopener">
+                  <img src="{{ a.thumb_proxy }}"
+                       class="al-thumb"
+                       alt="Vorschau"
+                       loading="lazy"
+                       onerror="this.style.display='none';this.nextElementSibling.style.display='flex';" />
+                  <div class="al-thumb-placeholder" style="display:none;">🖼</div>
+                </a>
+                {% else %}
+                <div class="al-thumb-placeholder">🖼</div>
+                {% endif %}
+              </td>
+              <td>
+                <div class="al-title">
+                  <a href="/admin/articles/{{ a.id }}">#{{ a.id }} {{ a.title }}</a>
+                </div>
+                {% if a.excerpt %}
+                <div class="al-excerpt">{{ a.excerpt }}</div>
+                {% endif %}
+                {% if a.feed_name %}
+                <div class="al-excerpt" style="margin-top:4px;">📡 {{ a.feed_name }}</div>
+                {% endif %}
+              </td>
+              <td>
+                <span class="badge-sm badge-{{ a.status }}">{{ a.status }}</span>
+              </td>
+              <td style="font-size:0.82em;">
+                {% if a.scheduled_publish_at %}
+                📅 {{ a.scheduled_publish_at[:16] }}
+                {% elif a.published_at %}
+                {{ a.published_at[:10] }}
+                {% else %}
+                —
+                {% endif %}
+              </td>
+              <td>
+                <!-- Hidden original value for change detection -->
+                <input type="hidden" name="orig_{{ a.id }}" value="{{ a.wp_post_id or '' }}">
+                <input
+                  type="text"
+                  name="wp_{{ a.id }}"
+                  value="{{ a.wp_post_id or '' }}"
+                  data-orig="{{ a.wp_post_id or '' }}"
+                  class="wp-id-input"
+                  placeholder="—"
+                  inputmode="numeric"
+                  pattern="[0-9]*"
+                />
+                {% if a.wp_post_url %}
+                <a href="{{ a.wp_post_url }}" target="_blank" rel="noopener" class="wp-link">↗ WP öffnen</a>
+                {% endif %}
+              </td>
+            </tr>
+            {% endfor %}
+          </tbody>
+        </table>
+      </section>
+    </form>
+
+    <!-- Pagination (outside form) -->
+    <div class="pagination">
+      {% if page > 1 %}
+      <a href="?page=1&status_filter={{ status_filter }}&search={{ search }}">«</a>
+      <a href="?page={{ page - 1 }}&status_filter={{ status_filter }}&search={{ search }}">‹ Zurück</a>
+      {% endif %}
+
+      {% for p in range([1, page - 2]|max, [total_pages + 1, page + 3]|min) %}
+      {% if p == page %}
+      <span class="current">{{ p }}</span>
+      {% else %}
+      <a href="?page={{ p }}&status_filter={{ status_filter }}&search={{ search }}">{{ p }}</a>
+      {% endif %}
+      {% endfor %}
+
+      {% if page < total_pages %}
+      <a href="?page={{ page + 1 }}&status_filter={{ status_filter }}&search={{ search }}">Weiter ›</a>
+      <a href="?page={{ total_pages }}&status_filter={{ status_filter }}&search={{ search }}">»</a>
+      {% endif %}
+    </div>
+  </main>
+
+  <script>
+    (function () {
+      const inputs = document.querySelectorAll('.wp-id-input');
+      const btn = document.getElementById('save-btn');
+      const badge = document.getElementById('change-badge');
+
+      function countChanges() {
+        let n = 0;
+        inputs.forEach(inp => {
+          const changed = inp.value.trim() !== inp.dataset.orig.trim();
+          inp.classList.toggle('changed', changed);
+          if (changed) n++;
+        });
+        btn.disabled = n === 0;
+        badge.textContent = n + (n === 1 ? ' Änderung' : ' Änderungen');
+        badge.classList.toggle('visible', n > 0);
+      }
+
+      inputs.forEach(inp => inp.addEventListener('input', countChanges));
+      countChanges();
+    })();
+  </script>
+</body>
+</html>

backend/templates/admin_connectivity.html

@@ -0,0 +1,84 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+</head>
+<body>
+  <header class="topbar">
+    <div>
+      <h1>Connectivity Check</h1>
+      <p>Angemeldet als <strong>{{ user }}</strong></p>
+    </div>
+    <div class="row">
+      <a class="linkbtn" href="/admin/dashboard">Zurück</a>
+      <form method="post" action="/admin/logout">
+        <button type="submit" class="secondary">Logout</button>
+      </form>
+    </div>
+  </header>
+
+  <main class="container">
+    <section class="stats">
+      <article class="stat">
+        <div class="label">Checks</div>
+        <div class="value">{{ checks|length }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">OK</div>
+        <div class="value">{{ ok_count }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">Fehler</div>
+        <div class="value">{{ error_count }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">Zeitpunkt</div>
+        <div class="value">Live</div>
+      </article>
+    </section>
+
+    <section class="card">
+      <h2>Ziele</h2>
+      <p class="subtle">Geprüft werden DNS-Auflösung, TCP-Erreichbarkeit und bei URLs ein HTTP-Request.</p>
+      <form method="get" action="/admin/connectivity" class="row">
+        <button type="submit">Checks neu ausführen</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>Ergebnis</h2>
+      <table>
+        <thead>
+          <tr><th>Status</th><th>Name</th><th>Typ</th><th>Ziel</th><th>DNS</th><th>TCP</th><th>HTTP</th><th>Dauer</th></tr>
+        </thead>
+        <tbody>
+          {% for c in checks %}
+          <tr>
+            <td>{% if c.ok %}<span class="badge ok">OK</span>{% else %}<span class="badge bad">Fehler</span>{% endif %}</td>
+            <td>{{ c.label }}</td>
+            <td>{{ c.kind }}</td>
+            <td><code>{{ c.target }}</code></td>
+            <td>
+              {% if c.dns_ok %}<span class="badge ok">OK</span>{% else %}<span class="badge bad">FAIL</span>{% endif %}
+              <div class="subtle">{{ c.dns_info }}</div>
+            </td>
+            <td>
+              {% if c.tcp_ok %}<span class="badge ok">OK</span>{% else %}<span class="badge bad">FAIL</span>{% endif %}
+              <div class="subtle">{{ c.tcp_info }}</div>
+            </td>
+            <td>
+              {% if c.http_ok %}<span class="badge ok">OK</span>{% else %}<span class="badge bad">FAIL</span>{% endif %}
+              <div class="subtle">{{ c.http_info }}</div>
+            </td>
+            <td>{{ c.duration_ms }} ms</td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+  </main>
+</body>
+</html>

backend/templates/admin_dashboard.html

@@ -0,0 +1,405 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+</head>
+<body>
+  <header class="topbar">
+    <div>
+      <h1>rss-news Admin Dashboard</h1>
+      <p>Angemeldet als <strong>{{ user }}</strong></p>
+    </div>
+    <div class="row">
+      <a class="linkbtn" href="/admin/article-list">Artikelliste</a>
+      <a class="linkbtn" href="/admin/schedule">Veröffentlichungsplan</a>
+      <a class="linkbtn" href="/admin/connectivity">Connectivity Check</a>
+      <form method="post" action="/admin/logout">
+        <button type="submit" class="secondary">Logout</button>
+      </form>
+    </div>
+  </header>
+
+  <main class="container">
+    {% if flash_msg %}
+    <section class="card flash {{ 'flash-error' if flash_type == 'error' else 'flash-success' }}">
+      {{ flash_msg }}
+    </section>
+    {% endif %}
+
+    <section class="stats">
+      <article class="stat">
+        <div class="label">Quellen</div>
+        <div class="value">{{ sources|length }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">Feeds</div>
+        <div class="value">{{ feeds|length }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">Artikel</div>
+        <div class="value">{{ articles|length }}</div>
+      </article>
+      <article class="stat">
+        <div class="label">Runs</div>
+        <div class="value">{{ runs|length }}</div>
+      </article>
+    </section>
+
+    <section class="grid two">
+      <article class="card">
+        <h2>Quelle anlegen</h2>
+        <form method="post" action="/admin/sources/create" class="stack">
+          <input name="name" placeholder="Name" required />
+          <input name="base_url" placeholder="Base URL" />
+          <input name="terms_url" placeholder="Terms URL" />
+          <input name="license_name" placeholder="Lizenzname" />
+          <select name="risk_level">
+            <option value="green">green</option>
+            <option value="yellow" selected>yellow</option>
+            <option value="red">red</option>
+          </select>
+          <input name="last_reviewed_at" placeholder="last_reviewed_at (ISO)" />
+          <button type="submit">Quelle speichern</button>
+        </form>
+      </article>
+
+      <article class="card">
+        <h2>Feed anlegen</h2>
+        <form method="post" action="/admin/feeds/create" class="stack">
+          <input name="name" placeholder="Feed Name" required />
+          <input name="url" placeholder="https://..." required />
+          <label>Quelle</label>
+          <select name="source_id">
+            <option value="">-- keine --</option>
+            {% for s in sources %}
+            <option value="{{ s.id }}">{{ s.name }} (#{{ s.id }})</option>
+            {% endfor %}
+          </select>
+          <button type="submit">Feed speichern</button>
+        </form>
+      </article>
+    </section>
+
+    <section class="card">
+      <h2>Ingestion starten</h2>
+      <form method="post" action="/admin/ingestion/run" class="row">
+        <select name="feed_id">
+          <option value="">Alle aktivierten Feeds</option>
+          {% for f in feeds %}
+          <option value="{{ f.id }}">{{ f.name }} (#{{ f.id }})</option>
+          {% endfor %}
+        </select>
+        <button type="submit">Ingestion starten</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>Publisher ausführen</h2>
+      <form method="post" action="/admin/publisher/run" class="row">
+        <input name="max_jobs" value="10" />
+        <button type="submit">Publisher Run starten</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>Rewrite Run (geplante Artikel)</h2>
+      <p class="subtle">Verarbeitet alle Artikel im Status <code>rewrite</code> und setzt sie auf <code>publish</code>.</p>
+      <form method="post" action="/admin/rewrite/run" class="row">
+        <input name="max_jobs" value="10" />
+        <button type="submit">Rewrite Run starten</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>Quellen + Policy</h2>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Name</th><th>Risk</th><th>Lizenz</th><th>Terms</th><th>Policy</th></tr>
+        </thead>
+        <tbody>
+          {% for s in sources %}
+          <tr>
+            <td>{{ s.id }}</td>
+            <td>{{ s.name }}</td>
+            <td>{{ s.risk_level }}</td>
+            <td>{{ s.license_name or "-" }}</td>
+            <td>{{ s.terms_url or "-" }}</td>
+            <td>
+              {% if source_policy[s.id] %}
+                <span class="badge bad">BLOCKED ({{ source_policy[s.id]|length }})</span>
+                <div class="subtle">{{ source_policy[s.id]|join(", ") }}</div>
+              {% else %}
+                <span class="badge ok">OK</span>
+              {% endif %}
+            </td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Quellen verwalten</h2>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Name</th><th>URLs</th><th>Meta</th><th>Aktionen</th></tr>
+        </thead>
+        <tbody>
+          {% for s in sources %}
+          {% set source_form_id = 'source-update-' ~ s.id %}
+          <tr>
+            <td>#{{ s.id }}</td>
+            <td>
+              <input form="{{ source_form_id }}" name="name" value="{{ s.name }}" required />
+            </td>
+            <td>
+              <input form="{{ source_form_id }}" name="base_url" value="{{ s.base_url or '' }}" placeholder="Base URL" />
+              <input form="{{ source_form_id }}" name="terms_url" value="{{ s.terms_url or '' }}" placeholder="Terms URL" />
+              <input form="{{ source_form_id }}" name="license_name" value="{{ s.license_name or '' }}" placeholder="Lizenz" />
+            </td>
+            <td>
+              <select form="{{ source_form_id }}" name="risk_level">
+                <option value="green" {% if s.risk_level == 'green' %}selected{% endif %}>green</option>
+                <option value="yellow" {% if s.risk_level == 'yellow' %}selected{% endif %}>yellow</option>
+                <option value="red" {% if s.risk_level == 'red' %}selected{% endif %}>red</option>
+              </select>
+              <select form="{{ source_form_id }}" name="is_enabled">
+                <option value="1" {% if s.is_enabled %}selected{% endif %}>aktiv</option>
+                <option value="0" {% if not s.is_enabled %}selected{% endif %}>inaktiv</option>
+              </select>
+              <input form="{{ source_form_id }}" name="last_reviewed_at" value="{{ s.last_reviewed_at or '' }}" placeholder="last_reviewed_at" />
+              <input form="{{ source_form_id }}" name="notes" value="{{ s.notes or '' }}" placeholder="Notiz" />
+            </td>
+            <td>
+              <div class="inline">
+              <form method="post" action="/admin/sources/{{ s.id }}/update" id="{{ source_form_id }}" class="inline">
+                <button type="submit" class="secondary">Speichern</button>
+              </form>
+              <form method="post" action="/admin/sources/{{ s.id }}/delete" class="inline" onsubmit="return confirm('Quelle wirklich löschen?');">
+                <button type="submit">Löschen</button>
+              </form>
+              </div>
+            </td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Feeds verwalten</h2>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Name</th><th>URL</th><th>Quelle</th><th>Status</th><th>Aktionen</th></tr>
+        </thead>
+        <tbody>
+          {% for f in feeds %}
+          {% set feed_form_id = 'feed-update-' ~ f.id %}
+          <tr>
+            <td>#{{ f.id }}</td>
+            <td>
+              <input form="{{ feed_form_id }}" name="name" value="{{ f.name }}" required />
+            </td>
+            <td><input form="{{ feed_form_id }}" name="url" value="{{ f.url }}" required /></td>
+            <td>
+              <select form="{{ feed_form_id }}" name="source_id">
+                <option value="">-- keine --</option>
+                {% for s in sources %}
+                <option value="{{ s.id }}" {% if f.source_id == s.id %}selected{% endif %}>{{ s.name }} (#{{ s.id }})</option>
+                {% endfor %}
+              </select>
+            </td>
+            <td>
+              <select form="{{ feed_form_id }}" name="is_enabled">
+                <option value="1" {% if f.is_enabled %}selected{% endif %}>aktiv</option>
+                <option value="0" {% if not f.is_enabled %}selected{% endif %}>inaktiv</option>
+              </select>
+            </td>
+            <td>
+              <div class="inline">
+              <form method="post" action="/admin/feeds/{{ f.id }}/update" id="{{ feed_form_id }}" class="inline">
+                <button type="submit" class="secondary">Speichern</button>
+              </form>
+              <form method="post" action="/admin/feeds/{{ f.id }}/delete" class="inline" onsubmit="return confirm('Feed wirklich löschen?');">
+                <button type="submit">Löschen</button>
+              </form>
+              </div>
+            </td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Artikel (Review)</h2>
+      <form method="get" action="/admin/dashboard" class="row filter-row">
+        <label>Status-Filter</label>
+        <select name="status_filter">
+          <option value="" {% if not status_filter %}selected{% endif %}>alle</option>
+          {% for s in status_options %}
+          <option value="{{ s }}" {% if status_filter == s %}selected{% endif %}>{{ s }}</option>
+          {% endfor %}
+        </select>
+        <button type="submit" class="secondary">Filtern</button>
+        <a href="/admin/dashboard" class="linkbtn">Reset</a>
+        <a href="/api/articles/export?format=json{% if status_filter %}&status_filter={{ status_filter }}{% endif %}" class="linkbtn">Export JSON</a>
+        <a href="/api/articles/export?format=csv{% if status_filter %}&status_filter={{ status_filter }}{% endif %}" class="linkbtn">Export CSV</a>
+      </form>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Artikel</th><th>Status</th><th>Details</th><th>Rewrite</th><th>Transition</th></tr>
+        </thead>
+        <tbody>
+          {% for a in articles %}
+          <tr>
+            <td>{{ a.id }}</td>
+            <td>
+              <strong>{{ a.title }}</strong><br />
+              <span class="subtle">Autor: {{ a.author or "-" }}</span><br />
+              <span class="subtle">Datum: {{ a.published_at or "-" }} | Alter: {{ a.days_old if a.days_old is not none else "-" }} Tage | Relevanz: {{ a.relevance }}</span><br />
+              <a href="{{ a.source_url }}" target="_blank" rel="noopener">Original öffnen</a>
+              <br /><a href="/admin/articles/{{ a.id }}">Details anzeigen</a>
+              {% if a.canonical_url and a.canonical_url != a.source_url %}
+              <br /><a href="{{ a.canonical_url }}" target="_blank" rel="noopener">Canonical öffnen</a>
+              {% endif %}
+            </td>
+            <td><span class="badge">{{ a.status_ui }}</span></td>
+            <td>
+              <div class="subtle">Publish: {{ "bereit" if a.publish_ready else "blockiert" }}</div>
+              {% if not a.publish_ready and a.publish_blockers %}
+              <div class="subtle">{{ a.publish_blockers|join(", ") }}</div>
+              {% endif %}
+              {% if a.selected_image_url %}
+              <div class="subtle">Hauptbild gesetzt</div>
+              <a href="{{ a.selected_image_url }}" target="_blank" rel="noopener"><img src="{{ a.selected_image_proxy_url }}" data-fallback-src="{{ a.selected_image_url }}" alt="Hauptbild" class="thumb" loading="lazy" onerror="if(!this.dataset.fallbackUsed){this.dataset.fallbackUsed='1';this.src=this.dataset.fallbackSrc;}else{this.classList.add('img-failed');}" /></a>
+              {% endif %}
+              {% if a.summary %}
+              <div><strong>Summary:</strong> {{ a.summary }}</div>
+              {% endif %}
+              {% if a.generated_tags %}
+              <div><strong>Tags:</strong> {{ a.generated_tags|join("; ") }}</div>
+              {% endif %}
+              {% if a.content_raw %}
+              <details>
+                <summary>Volltext anzeigen</summary>
+                <div class="pre">{{ a.content_raw }}</div>
+              </details>
+              {% endif %}
+              <div class="subtle">Bilder: {{ a.extracted_images|length }}</div>
+              {% if a.extracted_images %}
+              <details>
+                <summary>Bild-URLs</summary>
+                <ul>
+                  {% for img in a.extracted_images %}
+                  <li><a href="{{ img }}" target="_blank" rel="noopener">{{ img }}</a></li>
+                  {% endfor %}
+                </ul>
+              </details>
+              {% endif %}
+              {% if a.press_contact %}
+              <details>
+                <summary>Pressekontakt</summary>
+                <div class="pre">{{ a.press_contact }}</div>
+              </details>
+              {% endif %}
+              {% if a.extraction_error %}
+              <div class="subtle">Extraktionsfehler: {{ a.extraction_error }}</div>
+              {% endif %}
+            </td>
+            <td>
+              {% if a.status_ui in ["new", "rewrite"] %}
+              <form method="post" action="/admin/articles/{{ a.id }}/rewrite-run" class="inline">
+                <button type="submit">Rewrite ausführen</button>
+              </form>
+              {% else %}
+              -
+              {% endif %}
+            </td>
+            <td>
+              <form method="post" action="/admin/articles/{{ a.id }}/transition" class="inline">
+                <select name="target_status">
+                  {% for s in allowed_transitions.get(a.status_ui, []) %}
+                  <option value="{{ s }}">{{ s }}</option>
+                  {% endfor %}
+                </select>
+                {% if allowed_transitions.get(a.status_ui, []) %}
+                <button type="submit" class="secondary">Setzen</button>
+                {% else %}
+                <span class="subtle">keine Aktion</span>
+                {% endif %}
+              </form>
+              {% if a.status_ui == 'close' %}
+              <form method="post" action="/admin/articles/{{ a.id }}/retry" class="inline" style="margin-top:4px;">
+                <button type="submit" title="Artikel auf 'neu' zurücksetzen – wird beim nächsten Pipeline-Lauf erneut verarbeitet">🔄 Wiederholen</button>
+              </form>
+              {% endif %}
+            </td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Runs</h2>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Typ</th><th>Status</th><th>Start</th><th>Ende</th></tr>
+        </thead>
+        <tbody>
+          {% for r in runs %}
+          <tr>
+            <td>{{ r.id }}</td>
+            <td>{{ r.run_type }}</td>
+            <td>{{ r.status }}</td>
+            <td>{{ r.started_at }}</td>
+            <td>{{ r.finished_at or "-" }}</td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+
+    <section class="card">
+      <h2>Publish Jobs</h2>
+      <table>
+        <thead>
+          <tr><th>ID</th><th>Artikel</th><th>Status</th><th>Attempts</th><th>WP Post</th><th>Fehler</th><th>Hinweis</th></tr>
+        </thead>
+        <tbody>
+          {% for j in publish_jobs %}
+          <tr>
+            <td>{{ j.id }}</td>
+            <td>#{{ j.article_id }} {{ j.article_title or "-" }}</td>
+            <td>{{ j.status }}</td>
+            <td>{{ j.attempts }}/{{ j.max_attempts }}</td>
+            <td>
+              {% if j.wp_post_url %}
+              <a href="{{ j.wp_post_url }}" target="_blank" rel="noopener">#{{ j.wp_post_id }}</a>
+              {% elif j.wp_post_id %}
+              #{{ j.wp_post_id }}
+              {% else %}
+              -
+              {% endif %}
+            </td>
+            <td>
+              {% if j.error_message %}
+              <span class="badge errcat errcat-{{ j.error_category }}">{{ j.error_category }}</span>
+              <div>{{ j.error_message }}</div>
+              {% else %}
+              -
+              {% endif %}
+            </td>
+            <td>{{ j.error_hint or "-" }}</td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+  </main>
+</body>
+</html>

backend/templates/admin_login.html

@@ -0,0 +1,27 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+</head>
+<body>
+  <main class="container login">
+    <h1>rss-news Admin</h1>
+    <p>Bitte anmelden, um das Tool zu verwalten.</p>
+    {% if error %}
+    <div class="alert">Login fehlgeschlagen. Bitte pruefen.</div>
+    {% endif %}
+    <form method="post" action="/admin/login" class="card">
+      <label>Benutzername
+        <input type="text" name="username" required />
+      </label>
+      <label>Passwort
+        <input type="password" name="password" required />
+      </label>
+      <button type="submit">Anmelden</button>
+    </form>
+  </main>
+</body>
+</html>

backend/templates/admin_schedule.html

@@ -0,0 +1,143 @@
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{{ title }}</title>
+  <link rel="stylesheet" href="/admin/static/admin.css" />
+  <style>
+    .schedule-table td, .schedule-table th { padding: 6px 10px; }
+    .slot-free { color: #aaa; font-style: italic; }
+    .slot-booked-db { color: #1a7a1a; font-weight: bold; }
+    .slot-booked-wp { color: #b35a00; font-weight: bold; }
+    .badge-db { background: #d4edda; color: #155724; padding: 2px 6px; border-radius: 4px; font-size: 0.75em; }
+    .badge-wp { background: #fff3cd; color: #856404; padding: 2px 6px; border-radius: 4px; font-size: 0.75em; }
+    .summary-bar { display: flex; gap: 1.5rem; margin-bottom: 1rem; font-size: 0.95em; }
+  </style>
+</head>
+<body>
+  <header class="topbar">
+    <div>
+      <h1>rss-news Veröffentlichungsplan</h1>
+      <p>Angemeldet als <strong>{{ user }}</strong></p>
+    </div>
+    <div class="row">
+      <a class="linkbtn" href="/admin/dashboard">Dashboard</a>
+      <a class="linkbtn" href="/admin/connectivity">Connectivity</a>
+      <form method="post" action="/admin/logout">
+        <button type="submit" class="secondary">Logout</button>
+      </form>
+    </div>
+  </header>
+
+  <main class="container">
+    {% if flash_msg %}
+    <section class="card flash {{ 'flash-error' if flash_type == 'error' else 'flash-success' }}">
+      {{ flash_msg }}
+    </section>
+    {% endif %}
+
+    <section class="card" style="display:flex;align-items:center;justify-content:space-between;flex-wrap:wrap;gap:1rem;">
+      <div>
+        <h2 style="margin:0;">WordPress → DB Synchronisieren</h2>
+        <p class="subtle" style="margin:4px 0 0;">Liest alle geplanten WP-Beiträge und aktualisiert die Slots in der lokalen DB.<br>Nutze dies nach manuellen Änderungen in WordPress.</p>
+      </div>
+      <form method="post" action="/admin/wp-sync">
+        <button type="submit">🔄 Jetzt synchronisieren</button>
+      </form>
+    </section>
+
+    <section class="card">
+      <h2>Slot-Übersicht (nächste 60 Tage)</h2>
+      <div class="summary-bar">
+        <span>📅 Belegte Slots gesamt: <strong>{{ slots|length }}</strong></span>
+        <span>🗄️ Aus Pipeline-DB: <strong>{{ slots|selectattr('source', 'eq', 'db')|list|length }}</strong></span>
+        <span>🌐 Nur in WordPress: <strong>{{ slots|selectattr('source', 'eq', 'wordpress')|list|length }}</strong></span>
+      </div>
+      <table class="schedule-table">
+        <thead>
+          <tr>
+            <th>Tag</th>
+            {% for h in hours %}
+            <th>{{ "%02d:00 Uhr"|format(h) }}</th>
+            {% endfor %}
+          </tr>
+        </thead>
+        <tbody>
+          {% for day in calendar_days %}
+          {% if day.any_booked %}
+          <tr>
+            <td><strong>{{ day.weekday }}</strong> {{ day.date_fmt }}</td>
+            {% for s in day.slots %}
+            <td>
+              {% if s.booked %}
+                {% set info = s.slot %}
+                {% if info.source == 'db' %}
+                <span class="slot-booked-db">✅</span>
+                <span class="badge-db">DB</span>
+                <div style="font-size:0.85em;">
+                  {% if info.article_id %}
+                  <a href="/admin/articles/{{ info.article_id }}">
+                    {{ (info.article_title or "Artikel")[:50] }}{% if (info.article_title or "")|length > 50 %}…{% endif %}
+                  </a>
+                  {% endif %}
+                  <br /><span class="subtle">Status: {{ info.article_status }}</span>
+                  {% if info.wp_post_url %}
+                  <br /><a href="{{ info.wp_post_url }}" target="_blank" rel="noopener">WP öffnen</a>
+                  {% endif %}
+                </div>
+                {% else %}
+                <span class="slot-booked-wp">⚠️</span>
+                <span class="badge-wp">WP</span>
+                <div style="font-size:0.85em;">{{ info.article_title }}</div>
+                {% endif %}
+              {% else %}
+                <span class="slot-free">frei</span>
+              {% endif %}
+            </td>
+            {% endfor %}
+          </tr>
+          {% endif %}
+          {% endfor %}
+        </tbody>
+      </table>
+      {% if not slots %}
+      <p class="subtle">Keine geplanten Beiträge in den nächsten 60 Tagen.</p>
+      {% endif %}
+    </section>
+
+    <section class="card">
+      <h2>Alle belegten Slots (Liste)</h2>
+      <table>
+        <thead>
+          <tr><th>Datum/Zeit</th><th>Quelle</th><th>Artikel</th><th>Status</th><th>WordPress</th></tr>
+        </thead>
+        <tbody>
+          {% for s in slots %}
+          <tr>
+            <td>{{ s.formatted }}</td>
+            <td>
+              {% if s.source == 'db' %}<span class="badge-db">Pipeline-DB</span>
+              {% else %}<span class="badge-wp">WordPress</span>{% endif %}
+            </td>
+            <td>
+              {% if s.article_id %}
+              <a href="/admin/articles/{{ s.article_id }}">{{ (s.article_title or "")[:60] }}</a>
+              {% else %}
+              {{ s.article_title or "-" }}
+              {% endif %}
+            </td>
+            <td>{{ s.article_status or "-" }}</td>
+            <td>
+              {% if s.wp_post_url %}
+              <a href="{{ s.wp_post_url }}" target="_blank" rel="noopener">Draft öffnen</a>
+              {% else %}-{% endif %}
+            </td>
+          </tr>
+          {% endfor %}
+        </tbody>
+      </table>
+    </section>
+  </main>
+</body>
+</html>

backend/tests/__init__.py

@@ -0,0 +1 @@
+"""Tests package."""

backend/tests/test_admin_ui.py

@@ -0,0 +1,419 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+
+from fastapi.testclient import TestClient
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.main import app
+from backend.app.repositories import (
+    ArticleUpsert,
+    FeedCreate,
+    SourceCreate,
+    create_feed,
+    create_source,
+    get_article_by_id,
+    upsert_article,
+)
+
+
+class TestAdminUi(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        os.environ["APP_DB_PATH"] = str(Path(self.tmp_dir.name) / "admin_ui.db")
+        os.environ["APP_ADMIN_USERNAME"] = "admin"
+        os.environ["APP_ADMIN_PASSWORD"] = "secret"
+        config_module.get_settings.cache_clear()
+        init_db()
+        self.client = TestClient(app)
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        os.environ.pop("APP_DB_PATH", None)
+        os.environ.pop("APP_ADMIN_USERNAME", None)
+        os.environ.pop("APP_ADMIN_PASSWORD", None)
+        self.tmp_dir.cleanup()
+
+    def test_admin_login_and_dashboard(self) -> None:
+        login_page = self.client.get("/admin/login")
+        self.assertEqual(login_page.status_code, 200)
+        self.assertIn("rss-news Admin", login_page.text)
+
+        login = self.client.post(
+            "/admin/login",
+            data={"username": "admin", "password": "secret"},
+            follow_redirects=True,
+        )
+        self.assertEqual(login.status_code, 200)
+        self.assertIn("Admin Dashboard", login.text)
+
+    def test_dashboard_redirects_if_not_logged_in(self) -> None:
+        res = self.client.get("/admin/dashboard", follow_redirects=False)
+        self.assertEqual(res.status_code, 303)
+        self.assertEqual(res.headers.get("location"), "/admin/login")
+
+    def test_create_feed_with_empty_source_id_does_not_error(self) -> None:
+        self.client.post(
+            "/admin/login",
+            data={"username": "admin", "password": "secret"},
+            follow_redirects=True,
+        )
+        # empty source_id used to cause validation issues in form parsing
+        res = self.client.post(
+            "/admin/feeds/create",
+            data={"name": "Feed X", "url": "https://example.org/feed.xml", "source_id": ""},
+            follow_redirects=False,
+        )
+        self.assertEqual(res.status_code, 303)
+        self.assertTrue(res.headers.get("location", "").startswith("/admin/dashboard"))
+
+    def test_article_detail_page_renders(self) -> None:
+        source_id = create_source(
+            SourceCreate(
+                name="Test Source",
+                base_url="https://example.org",
+                terms_url="https://example.org/terms",
+                license_name="cc-by",
+                risk_level="green",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at="2026-02-18T00:00:00Z",
+            )
+        )
+        feed_id = create_feed(
+            FeedCreate(
+                name="Test Feed",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+        article_id = upsert_article(
+            ArticleUpsert(
+                feed_id=feed_id,
+                source_article_id="id-1",
+                source_hash="hash-1",
+                title="Titel A",
+                source_url="https://example.org/a",
+                canonical_url="https://example.org/a",
+                published_at=None,
+                author="Autor A",
+                summary="Summary A",
+                content_raw="Volltext A",
+                content_rewritten=None,
+                image_urls_json='["https://example.org/img.jpg"]',
+                press_contact="Kontakt",
+                source_name_snapshot="Test Source",
+                source_terms_url_snapshot="https://example.org/terms",
+                source_license_name_snapshot="cc-by",
+                legal_checked=False,
+                legal_checked_at=None,
+                legal_note=None,
+                wp_post_id=None,
+                wp_post_url=None,
+                publish_attempts=0,
+                publish_last_error=None,
+                published_to_wp_at=None,
+                word_count=2,
+                status="new",
+                meta_json='{"extraction":{"images":["https://example.org/img.jpg"],"press_contact":"Kontakt"}}',
+            )
+        )
+
+        self.client.post(
+            "/admin/login",
+            data={"username": "admin", "password": "secret"},
+            follow_redirects=True,
+        )
+        res = self.client.get(f"/admin/articles/{article_id}", follow_redirects=True)
+        self.assertEqual(res.status_code, 200)
+        self.assertIn("Artikel-Detail", res.text)
+        self.assertIn("Checkliste", res.text)
+
+        decision = self.client.post(
+            f"/admin/articles/{article_id}/images/decision",
+            data={"image_url": "https://example.org/img.jpg", "action": "select"},
+            follow_redirects=True,
+        )
+        self.assertEqual(decision.status_code, 200)
+        self.assertIn("Ausgewähltes Hauptbild", decision.text)
+
+        article = get_article_by_id(article_id)
+        self.assertIsNotNone(article)
+        self.assertIn("selected_url", article.get("meta_json", ""))
+
+    def test_manage_source_and_feed(self) -> None:
+        source_id = create_source(
+            SourceCreate(
+                name="Edit Source",
+                base_url="https://example.org",
+                terms_url="https://example.org/terms",
+                license_name="cc-by",
+                risk_level="yellow",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at=None,
+            )
+        )
+        feed_id = create_feed(
+            FeedCreate(
+                name="Edit Feed",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+        self.client.post("/admin/login", data={"username": "admin", "password": "secret"}, follow_redirects=True)
+
+        update_source_res = self.client.post(
+            f"/admin/sources/{source_id}/update",
+            data={
+                "name": "Edit Source 2",
+                "base_url": "https://example.org/new",
+                "terms_url": "https://example.org/new-terms",
+                "license_name": "cc0",
+                "risk_level": "green",
+                "is_enabled": "1",
+                "notes": "ok",
+                "last_reviewed_at": "2026-02-21T12:00:00Z",
+            },
+            follow_redirects=False,
+        )
+        self.assertEqual(update_source_res.status_code, 303)
+
+        update_feed_res = self.client.post(
+            f"/admin/feeds/{feed_id}/update",
+            data={
+                "name": "Edit Feed 2",
+                "url": "https://example.org/feed2.xml",
+                "source_id": str(source_id),
+                "is_enabled": "0",
+            },
+            follow_redirects=False,
+        )
+        self.assertEqual(update_feed_res.status_code, 303)
+
+        delete_feed_res = self.client.post(f"/admin/feeds/{feed_id}/delete", follow_redirects=False)
+        self.assertEqual(delete_feed_res.status_code, 303)
+        delete_source_res = self.client.post(f"/admin/sources/{source_id}/delete", follow_redirects=False)
+        self.assertEqual(delete_source_res.status_code, 303)
+
+    def test_rewrite_save_and_reopen(self) -> None:
+        source_id = create_source(
+            SourceCreate(
+                name="Test Source",
+                base_url="https://example.org",
+                terms_url="https://example.org/terms",
+                license_name="cc-by",
+                risk_level="green",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at="2026-02-18T00:00:00Z",
+            )
+        )
+        feed_id = create_feed(
+            FeedCreate(
+                name="Test Feed",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+        article_id = upsert_article(
+            ArticleUpsert(
+                feed_id=feed_id,
+                source_article_id="id-published",
+                source_hash="hash-published",
+                title="Titel Published",
+                source_url="https://example.org/published",
+                canonical_url="https://example.org/published",
+                published_at=None,
+                author="Autor A",
+                summary="Summary",
+                content_raw="Raw",
+                content_rewritten="<p>Alt</p>",
+                image_urls_json=None,
+                press_contact=None,
+                source_name_snapshot="Test Source",
+                source_terms_url_snapshot="https://example.org/terms",
+                source_license_name_snapshot="cc-by",
+                legal_checked=True,
+                legal_checked_at="2026-02-21T10:00:00Z",
+                legal_note=None,
+                wp_post_id=123,
+                wp_post_url="https://example.org/?p=123",
+                publish_attempts=2,
+                publish_last_error=None,
+                published_to_wp_at="2026-02-21T10:10:00Z",
+                word_count=1,
+                status="published",
+                meta_json="{}",
+            )
+        )
+        self.client.post("/admin/login", data={"username": "admin", "password": "secret"}, follow_redirects=True)
+
+        save_res = self.client.post(
+            f"/admin/articles/{article_id}/rewrite-save",
+            data={"content_rewritten": "<h2>Neu</h2><p>Text</p>"},
+            follow_redirects=False,
+        )
+        self.assertEqual(save_res.status_code, 303)
+
+        reopen_res = self.client.post(f"/admin/articles/{article_id}/reopen", follow_redirects=False)
+        self.assertEqual(reopen_res.status_code, 303)
+
+        article = get_article_by_id(article_id)
+        self.assertIsNotNone(article)
+        self.assertEqual(article.get("status"), "rewrite")
+        self.assertIn("Neu", article.get("content_rewritten") or "")
+        self.assertIsNone(article.get("wp_post_id"))
+
+    @patch("backend.app.admin_ui.generate_article_tags")
+    @patch("backend.app.admin_ui.rewrite_article_text")
+    def test_batch_rewrite_run_processes_planned_articles(self, mock_rewrite_text, mock_tags) -> None:
+        mock_rewrite_text.return_value = "<h2>Neu</h2><p>Text</p>"
+        mock_tags.return_value = ["Rheingas", "Monheim"]
+
+        source_id = create_source(
+            SourceCreate(
+                name="Batch Source",
+                base_url="https://example.org",
+                terms_url="https://example.org/terms",
+                license_name="cc-by",
+                risk_level="green",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at=None,
+            )
+        )
+        feed_id = create_feed(
+            FeedCreate(
+                name="Batch Feed",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+        article_id = upsert_article(
+            ArticleUpsert(
+                feed_id=feed_id,
+                source_article_id="batch-1",
+                source_hash="batch-hash-1",
+                title="Batch Titel",
+                source_url="https://example.org/batch",
+                canonical_url="https://example.org/batch",
+                published_at=None,
+                author="Autor",
+                summary="Summary",
+                content_raw="Raw",
+                content_rewritten=None,
+                image_urls_json=None,
+                press_contact=None,
+                source_name_snapshot="Batch Source",
+                source_terms_url_snapshot="https://example.org/terms",
+                source_license_name_snapshot="cc-by",
+                legal_checked=False,
+                legal_checked_at=None,
+                legal_note=None,
+                wp_post_id=None,
+                wp_post_url=None,
+                publish_attempts=0,
+                publish_last_error=None,
+                published_to_wp_at=None,
+                word_count=1,
+                status="rewrite",
+                meta_json="{}",
+            )
+        )
+        self.client.post("/admin/login", data={"username": "admin", "password": "secret"}, follow_redirects=True)
+        res = self.client.post("/admin/rewrite/run", data={"max_jobs": "10"}, follow_redirects=False)
+        self.assertEqual(res.status_code, 303)
+        article = get_article_by_id(article_id)
+        self.assertIsNotNone(article)
+        self.assertEqual(article.get("status"), "approved")
+        self.assertIn("generated_tags", article.get("meta_json", ""))
+
+    @patch("backend.app.admin_ui.urlopen")
+    def test_image_proxy_returns_image_data(self, mock_urlopen) -> None:
+        class _FakeHeaders:
+            def get(self, key: str, default=None):
+                if key.lower() == "content-type":
+                    return "image/jpeg"
+                return default
+
+        class _FakeResponse:
+            headers = _FakeHeaders()
+
+            def read(self):
+                return b"\xff\xd8\xff\xd9"
+
+            def __enter__(self):
+                return self
+
+            def __exit__(self, exc_type, exc_val, exc_tb):
+                return False
+
+        mock_urlopen.return_value = _FakeResponse()
+
+        self.client.post(
+            "/admin/login",
+            data={"username": "admin", "password": "secret"},
+            follow_redirects=True,
+        )
+        res = self.client.get("/admin/images/proxy?url=https%3A%2F%2Fexample.org%2Fimg.jpg")
+        self.assertEqual(res.status_code, 200)
+        self.assertIn("image/jpeg", res.headers.get("content-type", ""))
+
+    @patch("backend.app.admin_ui._run_connectivity_check")
+    @patch("backend.app.admin_ui._build_connectivity_targets")
+    def test_connectivity_page_renders(self, mock_targets, mock_check) -> None:
+        mock_targets.return_value = [
+            {"label": "OpenAI API", "kind": "host", "value": "api.openai.com"},
+            {"label": "WordPress REST", "kind": "url", "value": "https://example.org/wp-json/wp/v2"},
+        ]
+        mock_check.side_effect = [
+            {
+                "label": "OpenAI API",
+                "kind": "host",
+                "target": "api.openai.com",
+                "dns_ok": True,
+                "dns_info": "1.2.3.4",
+                "tcp_ok": True,
+                "tcp_info": "port 443 erreichbar",
+                "http_ok": True,
+                "http_info": "n/a (host-only)",
+                "duration_ms": 12,
+                "ok": True,
+            },
+            {
+                "label": "WordPress REST",
+                "kind": "url",
+                "target": "https://example.org/wp-json/wp/v2",
+                "dns_ok": False,
+                "dns_info": "Name or service not known",
+                "tcp_ok": False,
+                "tcp_info": "-",
+                "http_ok": False,
+                "http_info": "-",
+                "duration_ms": 10,
+                "ok": False,
+            },
+        ]
+
+        self.client.post(
+            "/admin/login",
+            data={"username": "admin", "password": "secret"},
+            follow_redirects=True,
+        )
+        res = self.client.get("/admin/connectivity", follow_redirects=True)
+        self.assertEqual(res.status_code, 200)
+        self.assertIn("Connectivity Check", res.text)
+        self.assertIn("OpenAI API", res.text)
+        self.assertIn("WordPress REST", res.text)
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_api_auth.py

@@ -0,0 +1,144 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+
+from fastapi.testclient import TestClient
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.main import app
+
+
+class TestApiAuth(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        os.environ["APP_DB_PATH"] = str(Path(self.tmp_dir.name) / "api.db")
+        os.environ["APP_ADMIN_USERNAME"] = "admin"
+        os.environ["APP_ADMIN_PASSWORD"] = "secret"
+        config_module.get_settings.cache_clear()
+        init_db()
+        self.client = TestClient(app)
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        os.environ.pop("APP_DB_PATH", None)
+        os.environ.pop("APP_ADMIN_USERNAME", None)
+        os.environ.pop("APP_ADMIN_PASSWORD", None)
+        self.tmp_dir.cleanup()
+
+    def test_login_and_protected_endpoint(self) -> None:
+        r = self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+        self.assertEqual(r.status_code, 200)
+
+        p = self.client.get("/api/protected")
+        self.assertEqual(p.status_code, 200)
+        self.assertTrue(p.json().get("ok"))
+
+    def test_protected_requires_auth(self) -> None:
+        r = self.client.get("/api/protected")
+        self.assertEqual(r.status_code, 401)
+
+    def test_run_detail_endpoint(self) -> None:
+        login = self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+        self.assertEqual(login.status_code, 200)
+
+        created = self.client.post("/api/runs", json={"run_type": "ingestion", "status": "running"})
+        self.assertEqual(created.status_code, 200)
+        run_id = created.json()["id"]
+
+        detail = self.client.get(f"/api/runs/{run_id}")
+        self.assertEqual(detail.status_code, 200)
+        self.assertEqual(detail.json()["item"]["id"], run_id)
+
+    def test_source_policy_check_endpoint(self) -> None:
+        login = self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+        self.assertEqual(login.status_code, 200)
+
+        created = self.client.post(
+            "/api/sources",
+            json={
+                "name": "Policy Source",
+                "risk_level": "yellow",
+                "is_enabled": True,
+            },
+        )
+        self.assertEqual(created.status_code, 200)
+        source_id = created.json()["id"]
+
+        check = self.client.get(f"/api/sources/{source_id}/policy-check")
+        self.assertEqual(check.status_code, 200)
+        body = check.json()
+        self.assertFalse(body["allowed"])
+        self.assertGreaterEqual(len(body["issues"]), 1)
+
+    def test_articles_export_json_and_csv_contains_relevance(self) -> None:
+        login = self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+        self.assertEqual(login.status_code, 200)
+
+        source = self.client.post(
+            "/api/sources",
+            json={
+                "name": "Export Source",
+                "base_url": "https://example.org",
+                "terms_url": "https://example.org/terms",
+                "license_name": "cc-by",
+                "risk_level": "green",
+                "is_enabled": True,
+                "last_reviewed_at": "2026-02-18T00:00:00Z",
+            },
+        )
+        self.assertEqual(source.status_code, 200)
+        source_id = source.json()["id"]
+
+        feed = self.client.post(
+            "/api/feeds",
+            json={"name": "Export Feed", "url": "https://example.org/feed.xml", "source_id": source_id, "is_enabled": True},
+        )
+        self.assertEqual(feed.status_code, 200)
+        feed_id = feed.json()["id"]
+
+        article = self.client.post(
+            "/api/articles/upsert",
+            json={
+                "feed_id": feed_id,
+                "source_article_id": "exp-1",
+                "source_hash": "exp-hash-1",
+                "title": "Export Artikel",
+                "source_url": "https://example.org/article/1",
+                "canonical_url": "https://example.org/article/1",
+                "published_at": "2026-02-18T00:00:00Z",
+                "author": "Autor",
+                "summary": "Kurz",
+                "content_raw": "Langtext",
+                "image_urls_json": "[\"https://example.org/img.jpg\"]",
+                "press_contact": "Kontakt",
+                "source_name_snapshot": "Export Source",
+                "source_terms_url_snapshot": "https://example.org/terms",
+                "source_license_name_snapshot": "cc-by",
+                "status": "review",
+            },
+        )
+        self.assertEqual(article.status_code, 200)
+
+        export_json = self.client.get("/api/articles/export?format=json")
+        self.assertEqual(export_json.status_code, 200)
+        body = export_json.json()
+        self.assertTrue(body.get("ok"))
+        self.assertGreaterEqual(body.get("count", 0), 1)
+        first = body["items"][0]
+        self.assertIn("published_at", first)
+        self.assertIn("days_old", first)
+        self.assertIn("relevance", first)
+
+        export_csv = self.client.get("/api/articles/export?format=csv")
+        self.assertEqual(export_csv.status_code, 200)
+        self.assertIn("text/csv", export_csv.headers.get("content-type", ""))
+        csv_text = export_csv.text
+        self.assertIn("published_at", csv_text)
+        self.assertIn("days_old", csv_text)
+        self.assertIn("relevance", csv_text)
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_article_workflow.py

@@ -0,0 +1,110 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+
+from fastapi.testclient import TestClient
+from unittest.mock import patch
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.main import app
+
+
+class TestArticleWorkflow(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        os.environ["APP_DB_PATH"] = str(Path(self.tmp_dir.name) / "workflow.db")
+        os.environ["APP_ADMIN_USERNAME"] = "admin"
+        os.environ["APP_ADMIN_PASSWORD"] = "secret"
+        config_module.get_settings.cache_clear()
+        init_db()
+        self.client = TestClient(app)
+        self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        os.environ.pop("APP_DB_PATH", None)
+        os.environ.pop("APP_ADMIN_USERNAME", None)
+        os.environ.pop("APP_ADMIN_PASSWORD", None)
+        self.tmp_dir.cleanup()
+
+    def _create_article(self) -> int:
+        source = self.client.post(
+            "/api/sources",
+            json={
+                "name": "Workflow Source",
+                "base_url": "https://example.org",
+                "terms_url": "https://example.org/terms",
+                "license_name": "cc-by",
+                "risk_level": "green",
+                "is_enabled": True,
+                "last_reviewed_at": "2026-02-18T00:00:00Z",
+            },
+        )
+        source_id = source.json()["id"]
+
+        feed = self.client.post(
+            "/api/feeds",
+            json={"name": "Workflow Feed", "url": "https://example.org/feed.xml", "source_id": source_id, "is_enabled": True},
+        )
+        feed_id = feed.json()["id"]
+
+        article = self.client.post(
+            "/api/articles/upsert",
+            json={
+                "feed_id": feed_id,
+                "source_article_id": "wf-1",
+                "source_url": "https://example.org/a1",
+                "title": "Workflow Artikel",
+                "summary": "s",
+                "content_raw": "c",
+                "status": "new",
+            },
+        )
+        return article.json()["id"]
+
+    def test_valid_transition_chain(self) -> None:
+        article_id = self._create_article()
+
+        t1 = self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "rewrite"})
+        self.assertEqual(t1.status_code, 200)
+
+        t2 = self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "publish"})
+        self.assertEqual(t2.status_code, 200)
+
+        t3 = self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "published"})
+        self.assertEqual(t3.status_code, 200)
+
+        t4 = self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "rewrite"})
+        self.assertEqual(t4.status_code, 200)
+
+        final = self.client.get(f"/api/articles/{article_id}")
+        self.assertEqual(final.status_code, 200)
+        self.assertEqual(final.json()["item"]["status"], "rewrite")
+        self.assertEqual(final.json()["item"]["status_ui"], "rewrite")
+
+    def test_invalid_transition_rejected(self) -> None:
+        article_id = self._create_article()
+        bad = self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "published"})
+        self.assertEqual(bad.status_code, 400)
+
+    def test_legacy_review_endpoint_is_gone(self) -> None:
+        article_id = self._create_article()
+        bad = self.client.post(f"/api/articles/{article_id}/review", json={"decision": "approve"})
+        self.assertEqual(bad.status_code, 410)
+
+    @patch("backend.app.main.rewrite_article_text")
+    def test_rewrite_run_sets_publish_status(self, mock_rewrite) -> None:
+        mock_rewrite.return_value = "<h2>Neu</h2><p>Umschreibung</p>"
+        article_id = self._create_article()
+        self.client.post(f"/api/articles/{article_id}/transition", json={"target_status": "rewrite"})
+        r = self.client.post(f"/api/articles/{article_id}/rewrite-run")
+        self.assertEqual(r.status_code, 200)
+        self.assertEqual(r.json()["status"], "publish")
+        final = self.client.get(f"/api/articles/{article_id}")
+        self.assertEqual(final.json()["item"]["status_ui"], "publish")
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_db_repositories.py

@@ -0,0 +1,145 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.repositories import (
+    ArticleUpsert,
+    FeedCreate,
+    RunCreate,
+    SourceCreate,
+    create_feed,
+    create_run,
+    create_source,
+    finish_run,
+    list_articles,
+    list_feeds,
+    list_runs,
+    list_sources,
+    upsert_article,
+)
+
+
+class TestSQLiteRepositories(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        self.db_path = str(Path(self.tmp_dir.name) / "test.db")
+        os.environ["APP_DB_PATH"] = self.db_path
+        config_module.get_settings.cache_clear()
+        init_db()
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        os.environ.pop("APP_DB_PATH", None)
+        self.tmp_dir.cleanup()
+
+    def test_end_to_end_basic_crud(self) -> None:
+        source_id = create_source(
+            SourceCreate(
+                name="GovData",
+                base_url="https://data.gov.de",
+                terms_url="https://www.govdata.de/dl-de/by-2-0",
+                license_name="dl-de/by-2-0",
+                risk_level="green",
+                is_enabled=True,
+                notes="test source",
+                last_reviewed_at="2026-02-18T00:00:00Z",
+            )
+        )
+        self.assertGreater(source_id, 0)
+
+        feed_id = create_feed(
+            FeedCreate(
+                name="GovData RSS",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+        self.assertGreater(feed_id, 0)
+
+        run_id = create_run(RunCreate(run_type="ingest", status="running", details="start"))
+        self.assertGreater(run_id, 0)
+        finish_run(run_id=run_id, status="success", details="ok")
+
+        article_id = upsert_article(
+            ArticleUpsert(
+                feed_id=feed_id,
+                source_article_id="abc-1",
+                source_hash="hash-abc-1",
+                title="Beispielartikel",
+                source_url="https://example.org/articles/1",
+                canonical_url="https://example.org/articles/1",
+                published_at="2026-02-18T00:00:00Z",
+                author="Max Mustermann",
+                summary="Kurzfassung",
+                content_raw="Originaltext",
+                content_rewritten="Umschreibung",
+                image_urls_json='["https://example.org/img.jpg"]',
+                press_contact="Pressekontakt X",
+                source_name_snapshot="GovData",
+                source_terms_url_snapshot="https://www.govdata.de/dl-de/by-2-0",
+                source_license_name_snapshot="dl-de/by-2-0",
+                legal_checked=False,
+                legal_checked_at=None,
+                legal_note=None,
+                wp_post_id=None,
+                wp_post_url=None,
+                publish_attempts=0,
+                publish_last_error=None,
+                published_to_wp_at=None,
+                word_count=120,
+                status="review",
+                meta_json='{"lang":"de"}',
+            )
+        )
+        self.assertGreater(article_id, 0)
+
+        # Upsert with same source_url updates same row
+        article_id_2 = upsert_article(
+            ArticleUpsert(
+                feed_id=feed_id,
+                source_article_id="abc-1",
+                source_hash="hash-abc-1",
+                title="Beispielartikel aktualisiert",
+                source_url="https://example.org/articles/1",
+                canonical_url="https://example.org/articles/1",
+                published_at="2026-02-18T00:00:00Z",
+                author="Max Mustermann",
+                summary="Kurzfassung 2",
+                content_raw="Originaltext 2",
+                content_rewritten="Umschreibung 2",
+                image_urls_json='["https://example.org/img2.jpg"]',
+                press_contact="Pressekontakt Y",
+                source_name_snapshot="GovData",
+                source_terms_url_snapshot="https://www.govdata.de/dl-de/by-2-0",
+                source_license_name_snapshot="dl-de/by-2-0",
+                legal_checked=True,
+                legal_checked_at="2026-02-18T00:10:00Z",
+                legal_note="ok",
+                wp_post_id=123,
+                wp_post_url="https://example.org/wp/123",
+                publish_attempts=1,
+                publish_last_error=None,
+                published_to_wp_at="2026-02-18T00:12:00Z",
+                word_count=140,
+                status="approved",
+                meta_json='{"lang":"de","v":2}',
+            )
+        )
+        self.assertEqual(article_id, article_id_2)
+
+        self.assertEqual(len(list_sources()), 1)
+        self.assertEqual(len(list_feeds()), 1)
+        self.assertEqual(len(list_runs()), 1)
+
+        articles = list_articles()
+        self.assertEqual(len(articles), 1)
+        self.assertEqual(articles[0]["title"], "Beispielartikel aktualisiert")
+        self.assertEqual(articles[0]["status"], "approved")
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_ingestion.py

@@ -0,0 +1,245 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.ingestion import run_ingestion
+from backend.app.repositories import (
+    ArticleUpsert,
+    FeedCreate,
+    SourceCreate,
+    create_feed,
+    create_source,
+    get_article_by_id,
+    list_articles,
+    upsert_article,
+)
+from backend.app.source_extraction import ExtractedArticle
+
+
+class TestIngestion(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        os.environ["APP_DB_PATH"] = str(Path(self.tmp_dir.name) / "ingestion.db")
+        config_module.get_settings.cache_clear()
+        init_db()
+
+        source_id = create_source(
+            SourceCreate(
+                name="Test Source",
+                base_url="https://example.org",
+                terms_url="https://example.org/terms",
+                license_name="cc-by",
+                risk_level="green",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at="2026-02-18T00:00:00Z",
+            )
+        )
+        self.feed_id = create_feed(
+            FeedCreate(
+                name="Test Feed",
+                url="https://example.org/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        os.environ.pop("APP_DB_PATH", None)
+        self.tmp_dir.cleanup()
+
+    @patch("backend.app.ingestion.extract_article")
+    @patch("backend.app.ingestion.feedparser.parse")
+    def test_ingestion_deduplicates_by_feed_and_guid(self, mock_parse, mock_extract_article) -> None:
+        mock_extract_article.return_value = ExtractedArticle(
+            title="Artikel 1 original",
+            author="Autorin A",
+            canonical_url="https://example.org/article/1",
+            summary="Original Summary",
+            content_text="Original Volltext",
+            images=["https://example.org/a.jpg"],
+            press_contact="Pressekontakt: Team A",
+            extraction_error=None,
+        )
+        mock_parse.return_value = {
+            "etag": "etag-1",
+            "modified": "Tue, 18 Feb 2026 10:00:00 GMT",
+            "entries": [
+                {
+                    "id": "item-1",
+                    "title": "Artikel 1",
+                    "link": "https://example.org/article/1",
+                    "summary": "A",
+                },
+                {
+                    "id": "item-1",
+                    "title": "Artikel 1 aktualisiert",
+                    "link": "https://example.org/article/1-neu",
+                    "summary": "B",
+                },
+            ],
+        }
+
+        stats = run_ingestion(feed_id=self.feed_id)
+        self.assertEqual(stats.status, "success")
+        self.assertEqual(stats.entries_seen, 2)
+        self.assertEqual(len(list_articles()), 1)
+        article = list_articles()[0]
+        self.assertEqual(article["title"], "Artikel 1 original")
+        self.assertEqual(article["author"], "Autorin A")
+        self.assertIn("Original Volltext", article["content_raw"] or "")
+        self.assertIn("Pressekontakt", article["meta_json"] or "")
+        self.assertIsNotNone(article["image_urls_json"])
+
+    @patch("backend.app.ingestion.extract_article")
+    @patch("backend.app.ingestion.feedparser.parse")
+    def test_ingestion_processes_any_enabled_source(self, mock_parse, mock_extract_article) -> None:
+        # Ampel/risk-level system removed – all enabled feeds are processed regardless of risk_level
+        source_id = create_source(
+            SourceCreate(
+                name="Any Risk Source",
+                base_url="https://example.net",
+                terms_url="https://example.net/terms",
+                license_name="custom",
+                risk_level="yellow",
+                is_enabled=True,
+                notes=None,
+                last_reviewed_at="2026-02-18T00:00:00Z",
+            )
+        )
+        feed_id = create_feed(
+            FeedCreate(
+                name="Any Risk Feed",
+                url="https://example.net/feed.xml",
+                source_id=source_id,
+                is_enabled=True,
+            )
+        )
+
+        mock_parse.return_value = type("FP", (), {"entries": [], "etag": None, "modified": None})()
+        mock_extract_article.return_value = type("E", (), {
+            "title": None, "author": None, "summary": None, "content_text": None,
+            "canonical_url": None, "images": [], "press_contact": None,
+        })()
+
+        stats = run_ingestion(feed_id=feed_id)
+        self.assertEqual(stats.status, "success")
+        # Feed was processed (feedparser was called), even with yellow risk_level
+        mock_parse.assert_called_once()
+
+    @patch("backend.app.ingestion.extract_article")
+    @patch("backend.app.ingestion.feedparser.parse")
+    def test_ingestion_preserves_existing_work_and_skips_closed(self, mock_parse, mock_extract_article) -> None:
+        existing_closed_id = upsert_article(
+            ArticleUpsert(
+                feed_id=self.feed_id,
+                source_article_id="closed-1",
+                source_hash="closed-hash-1",
+                title="Alt Closed",
+                source_url="https://example.org/closed-article",
+                canonical_url="https://example.org/closed-article",
+                published_at=None,
+                author="Autor",
+                summary="Alt",
+                content_raw="Alt Raw",
+                content_rewritten="<p>Alt Rewrite Closed</p>",
+                image_urls_json=None,
+                press_contact="Kontakt Alt",
+                source_name_snapshot="Test Source",
+                source_terms_url_snapshot="https://example.org/terms",
+                source_license_name_snapshot="cc-by",
+                legal_checked=False,
+                legal_checked_at=None,
+                legal_note=None,
+                wp_post_id=42,
+                wp_post_url="https://wp.local/?p=42",
+                publish_attempts=2,
+                publish_last_error=None,
+                published_to_wp_at="2026-02-21T12:00:00Z",
+                word_count=3,
+                status="error",  # UI: close
+                meta_json='{"generated_tags":["AltTag"]}',
+            )
+        )
+        existing_published_id = upsert_article(
+            ArticleUpsert(
+                feed_id=self.feed_id,
+                source_article_id="published-1",
+                source_hash="published-hash-1",
+                title="Alt Published",
+                source_url="https://example.org/published-article",
+                canonical_url="https://example.org/published-article",
+                published_at=None,
+                author="Autor",
+                summary="Alt",
+                content_raw="Alt Raw",
+                content_rewritten="<p>Alt Rewrite Published</p>",
+                image_urls_json=None,
+                press_contact="Kontakt Alt",
+                source_name_snapshot="Test Source",
+                source_terms_url_snapshot="https://example.org/terms",
+                source_license_name_snapshot="cc-by",
+                legal_checked=False,
+                legal_checked_at=None,
+                legal_note=None,
+                wp_post_id=77,
+                wp_post_url="https://wp.local/?p=77",
+                publish_attempts=3,
+                publish_last_error=None,
+                published_to_wp_at="2026-02-21T12:10:00Z",
+                word_count=3,
+                status="published",
+                meta_json='{"generated_tags":["Rheingas"],"image_review":{"selected_url":"https://img.local/1.jpg"}}',
+            )
+        )
+
+        mock_extract_article.return_value = ExtractedArticle(
+            title="Neu Titel",
+            author="Neu Autor",
+            canonical_url=None,
+            summary="Neu Summary",
+            content_text="Neu Volltext",
+            images=["https://example.org/a.jpg"],
+            press_contact=None,
+            extraction_error=None,
+        )
+        mock_parse.return_value = {
+            "etag": "etag-2",
+            "modified": "Tue, 18 Feb 2026 11:00:00 GMT",
+            "entries": [
+                {
+                    "id": "closed-1",
+                    "title": "Closed Entry",
+                    "link": "https://example.org/closed-article",
+                    "summary": "X",
+                },
+                {
+                    "id": "published-1",
+                    "title": "Published Entry",
+                    "link": "https://example.org/published-article",
+                    "summary": "Y",
+                },
+            ],
+        }
+
+        stats = run_ingestion(feed_id=self.feed_id)
+        self.assertEqual(stats.status, "success")
+        closed_row = get_article_by_id(existing_closed_id) or {}
+        self.assertEqual(closed_row["status"], "error")
+        self.assertIn("Alt Rewrite Closed", closed_row.get("content_rewritten") or "")
+        self.assertEqual(closed_row.get("wp_post_id"), 42)
+
+        published_row = get_article_by_id(existing_published_id) or {}
+        self.assertEqual(published_row["status"], "published")
+        self.assertIn("Alt Rewrite Published", published_row.get("content_rewritten") or "")
+        self.assertEqual(published_row.get("wp_post_id"), 77)
+        self.assertIn("generated_tags", published_row.get("meta_json") or "")
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_publisher.py

@@ -0,0 +1,112 @@
+import os
+import tempfile
+import unittest
+from pathlib import Path
+from unittest.mock import patch
+
+from fastapi.testclient import TestClient
+
+from backend.app import config as config_module
+from backend.app.db import init_db
+from backend.app.main import app
+
+
+class TestPublisher(unittest.TestCase):
+    def setUp(self) -> None:
+        self.tmp_dir = tempfile.TemporaryDirectory()
+        os.environ["APP_DB_PATH"] = str(Path(self.tmp_dir.name) / "publisher.db")
+        os.environ["APP_ADMIN_USERNAME"] = "admin"
+        os.environ["APP_ADMIN_PASSWORD"] = "secret"
+        os.environ["WORDPRESS_BASE_URL"] = "https://example.org"
+        os.environ["WORDPRESS_USERNAME"] = "wp-user"
+        os.environ["WORDPRESS_APP_PASSWORD"] = "wp-pass"
+        config_module.get_settings.cache_clear()
+        init_db()
+        self.client = TestClient(app)
+        self.client.post("/auth/login", json={"username": "admin", "password": "secret"})
+
+    def tearDown(self) -> None:
+        config_module.get_settings.cache_clear()
+        for key in (
+            "APP_DB_PATH",
+            "APP_ADMIN_USERNAME",
+            "APP_ADMIN_PASSWORD",
+            "WORDPRESS_BASE_URL",
+            "WORDPRESS_USERNAME",
+            "WORDPRESS_APP_PASSWORD",
+        ):
+            os.environ.pop(key, None)
+        self.tmp_dir.cleanup()
+
+    def _create_publishable_article(self) -> int:
+        source = self.client.post(
+            "/api/sources",
+            json={
+                "name": "WP Source",
+                "base_url": "https://example.org",
+                "terms_url": "https://example.org/terms",
+                "license_name": "cc-by",
+                "risk_level": "green",
+                "is_enabled": True,
+                "last_reviewed_at": "2026-02-18T00:00:00Z",
+            },
+        )
+        source_id = source.json()["id"]
+        feed = self.client.post(
+            "/api/feeds",
+            json={"name": "WP Feed", "url": "https://example.org/feed.xml", "source_id": source_id, "is_enabled": True},
+        )
+        feed_id = feed.json()["id"]
+
+        article = self.client.post(
+            "/api/articles/upsert",
+            json={
+                "feed_id": feed_id,
+                "source_article_id": "pub-1",
+                "source_hash": "pub-hash-1",
+                "title": "Publish Artikel",
+                "source_url": "https://example.org/article/1",
+                "canonical_url": "https://example.org/article/1",
+                "published_at": "2026-02-18T00:00:00Z",
+                "author": "Autor",
+                "summary": "Kurz",
+                "content_raw": "Langtext",
+                "image_urls_json": "[\"https://example.org/img.jpg\"]",
+                "press_contact": "Kontakt",
+                "source_name_snapshot": "WP Source",
+                "source_terms_url_snapshot": "https://example.org/terms",
+                "source_license_name_snapshot": "cc-by",
+                "legal_checked": True,
+                "status": "approved",
+                "meta_json": "{\"image_review\":{\"selected_url\":\"https://example.org/img.jpg\"}}",
+            },
+        )
+        return article.json()["id"]
+
+    @patch("backend.app.publisher.publish_article_draft")
+    def test_enqueue_and_run_publisher(self, mock_publish) -> None:
+        mock_publish.return_value = (777, "https://example.org/?p=777")
+        article_id = self._create_publishable_article()
+
+        enqueue = self.client.post("/api/publisher/enqueue", json={"article_id": article_id, "max_attempts": 3})
+        self.assertEqual(enqueue.status_code, 200)
+
+        run = self.client.post("/api/publisher/run", json={"max_jobs": 5})
+        self.assertEqual(run.status_code, 200)
+        stats = run.json()["stats"]
+        self.assertEqual(stats["success"], 1)
+
+        article = self.client.get(f"/api/articles/{article_id}")
+        self.assertEqual(article.status_code, 200)
+        item = article.json()["item"]
+        self.assertEqual(item["status"], "published")
+        self.assertEqual(item["wp_post_id"], 777)
+        self.assertIn("?p=777", item["wp_post_url"] or "")
+
+        jobs = self.client.get("/api/publisher/jobs")
+        self.assertEqual(jobs.status_code, 200)
+        self.assertGreaterEqual(len(jobs.json()["items"]), 1)
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_relevance.py

@@ -0,0 +1,21 @@
+from datetime import datetime, timezone
+import unittest
+
+from backend.app.relevance import article_age_days, article_relevance
+
+
+class TestRelevance(unittest.TestCase):
+    def test_article_age_and_relevance(self) -> None:
+        now = datetime(2026, 2, 18, 12, 0, 0, tzinfo=timezone.utc)
+        self.assertEqual(article_age_days("2026-02-18T10:00:00Z", now=now), 0)
+        self.assertEqual(article_relevance("2026-02-18T10:00:00Z", now=now), "hoch")
+
+        self.assertEqual(article_age_days("2026-02-14T12:00:00Z", now=now), 4)
+        self.assertEqual(article_relevance("2026-02-14T12:00:00Z", now=now), "mittel")
+
+        self.assertEqual(article_relevance("2025-12-01T00:00:00Z", now=now), "alt")
+        self.assertEqual(article_relevance(None, now=now), "unbekannt")
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_source_extraction.py

@@ -0,0 +1,96 @@
+import unittest
+from unittest.mock import patch
+
+from backend.app.source_extraction import extract_article
+
+
+SAMPLE_HTML = """
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta property="og:title" content="Demo Meldung von Presseportal" />
+  <meta name="author" content="Max Mustermann" />
+  <meta name="description" content="Kurzbeschreibung aus der Originalseite" />
+  <meta property="og:image" content="/images/demo.jpg" />
+  <link rel="canonical" href="https://www.presseportal.de/pm/118273/6158137" />
+</head>
+<body>
+  <article>
+    <p>Dies ist der vollstaendige Inhalt des Artikels.</p>
+    <p>Weitere relevante Informationen fuer die Meldung.</p>
+    <h3>Pressekontakt</h3>
+    <p>Musterfirma GmbH, Kontakt: presse@example.org</p>
+  </article>
+</body>
+</html>
+"""
+
+SAMPLE_HTML_AGENTUR = """
+<!doctype html>
+<html lang="de">
+<head>
+  <meta charset="utf-8" />
+  <meta property="og:title" content="Demo Meldung Agentur" />
+</head>
+<body>
+  <article>
+    <p>Inhalt der Meldung.</p>
+    <h3>Agentur</h3>
+    <p>Agenturname GmbH</p>
+    <p>presse@agentur.example</p>
+    <p>Original-Content von Beispiel</p>
+  </article>
+</body>
+</html>
+"""
+
+
+class _FakeHeaders:
+    @staticmethod
+    def get_content_charset():
+        return "utf-8"
+
+
+class _FakeResponse:
+    headers = _FakeHeaders()
+
+    def __init__(self, body: str):
+        self._body = body.encode("utf-8")
+
+    def read(self):
+        return self._body
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        return False
+
+
+class TestSourceExtraction(unittest.TestCase):
+    @patch("backend.app.source_extraction.urlopen")
+    def test_extract_article_parses_author_images_and_press_contact(self, mock_urlopen) -> None:
+        mock_urlopen.return_value = _FakeResponse(SAMPLE_HTML)
+
+        extracted = extract_article("https://www.presseportal.de/pm/118273/6158137")
+        self.assertEqual(extracted.title, "Demo Meldung von Presseportal")
+        self.assertEqual(extracted.author, "Max Mustermann")
+        self.assertEqual(extracted.canonical_url, "https://www.presseportal.de/pm/118273/6158137")
+        self.assertIn("vollstaendige Inhalt", extracted.content_text or "")
+        self.assertIn("Kurzbeschreibung", extracted.summary or "")
+        self.assertIn("https://www.presseportal.de/images/demo.jpg", extracted.images)
+        self.assertIn("Pressekontakt", extracted.press_contact or "")
+        self.assertIsNone(extracted.extraction_error)
+
+    @patch("backend.app.source_extraction.urlopen")
+    def test_extract_article_detects_agentur_block_as_press_contact(self, mock_urlopen) -> None:
+        mock_urlopen.return_value = _FakeResponse(SAMPLE_HTML_AGENTUR)
+        extracted = extract_article("https://www.presseportal.de/pm/155103/6210401")
+        self.assertIn("Agentur", extracted.press_contact or "")
+        self.assertIn("Agenturname", extracted.press_contact or "")
+        self.assertIn("presse@agentur.example", extracted.press_contact or "")
+
+
+if __name__ == "__main__":
+    unittest.main()

backend/tests/test_wordpress.py

@@ -0,0 +1,139 @@
+import os
+import unittest
+from unittest.mock import patch
+
+from backend.app import config as config_module
+from backend.app.wordpress import publish_article_draft
+
+
+class TestWordpressPublish(unittest.TestCase):
+    def setUp(self) -> None:
+        os.environ["WORDPRESS_BASE_URL"] = "https://example.org"
+        os.environ["WORDPRESS_USERNAME"] = "wp-user"
+        os.environ["WORDPRESS_APP_PASSWORD"] = "wp-pass"
+        config_module.get_settings.cache_clear()
+
+    def tearDown(self) -> None:
+        for key in ("WORDPRESS_BASE_URL", "WORDPRESS_USERNAME", "WORDPRESS_APP_PASSWORD"):
+            os.environ.pop(key, None)
+        config_module.get_settings.cache_clear()
+
+    @patch("backend.app.wordpress._upload_featured_media")
+    @patch("backend.app.wordpress._wp_request")
+    def test_publish_sets_featured_media_when_selected_image_exists(self, mock_wp_request, mock_upload_media) -> None:
+        mock_upload_media.return_value = 456
+        mock_wp_request.return_value = {"id": 321, "link": "https://example.org/?p=321"}
+
+        article = {
+            "title": "Testartikel",
+            "content_raw": "Inhalt",
+            "source_url": "https://example.com/source",
+            "canonical_url": "https://example.com/source",
+            "meta_json": '{"image_review":{"selected_url":"https://example.com/image.jpg"}}',
+        }
+        post_id, post_url = publish_article_draft(article)
+
+        self.assertEqual(post_id, 321)
+        self.assertIn("?p=321", post_url or "")
+        self.assertTrue(mock_upload_media.called)
+        payload = mock_wp_request.call_args.kwargs["payload"]
+        self.assertEqual(payload.get("featured_media"), 456)
+        self.assertIn("<!-- wp:paragraph -->", payload.get("content", ""))
+        self.assertIn("<p>Inhalt</p>", payload.get("content", ""))
+        self.assertNotIn("excerpt", payload)
+
+    @patch("backend.app.wordpress._upload_featured_media")
+    @patch("backend.app.wordpress._wp_request")
+    def test_publish_without_selected_image_has_no_featured_media(self, mock_wp_request, mock_upload_media) -> None:
+        mock_wp_request.return_value = {"id": 654, "link": "https://example.org/?p=654"}
+
+        article = {
+            "title": "Testartikel",
+            "content_raw": "Inhalt",
+            "source_url": "https://example.com/source",
+            "canonical_url": "https://example.com/source",
+            "meta_json": "{}",
+        }
+        post_id, _ = publish_article_draft(article)
+
+        self.assertEqual(post_id, 654)
+        self.assertFalse(mock_upload_media.called)
+        payload = mock_wp_request.call_args.kwargs["payload"]
+        self.assertNotIn("featured_media", payload)
+        self.assertIn("<p>Inhalt</p>", payload.get("content", ""))
+
+    @patch("backend.app.wordpress._upload_featured_media")
+    @patch("backend.app.wordpress._wp_request")
+    def test_publish_strips_feed_header_and_press_contact(self, mock_wp_request, mock_upload_media) -> None:
+        mock_wp_request.return_value = {"id": 100, "link": "https://example.org/?p=100"}
+        article = {
+            "title": "Header Test",
+            "content_raw": "21.02.2026 10:00\nFirma GmbH\n(ots)\nDas ist der eigentliche Text.\nPressekontakt: Test Person",
+            "source_url": "https://example.com/source",
+            "canonical_url": "https://example.com/source",
+            "meta_json": "{}",
+        }
+        publish_article_draft(article)
+        payload = mock_wp_request.call_args.kwargs["payload"]
+        content = payload.get("content", "")
+        self.assertNotIn("Firma GmbH", content)
+        self.assertNotIn("Pressekontakt", content)
+        self.assertIn("eigentliche Text", content)
+        self.assertNotIn("Artikeldetails", content)
+
+    @patch("backend.app.wordpress._upload_featured_media")
+    @patch("backend.app.wordpress._wp_request")
+    def test_publish_resolves_and_sets_tags(self, mock_wp_request, mock_upload_media) -> None:
+        def _fake_wp_request(**kwargs):
+            endpoint = kwargs.get("endpoint", "")
+            method = kwargs.get("method", "")
+            if method == "GET" and endpoint.startswith("tags?search="):
+                if "Rheingas" in endpoint:
+                    return [{"id": 11, "name": "Rheingas"}]
+                return []
+            if method == "POST" and endpoint == "tags":
+                name = (kwargs.get("payload") or {}).get("name")
+                if name == "Gasflasche":
+                    return {"id": 12, "name": "Gasflasche"}
+                return {"id": 13, "name": str(name)}
+            if method == "POST" and endpoint == "posts":
+                return {"id": 900, "link": "https://example.org/?p=900"}
+            return {}
+
+        mock_wp_request.side_effect = _fake_wp_request
+        article = {
+            "title": "Tag Test",
+            "content_raw": "Inhalt",
+            "source_url": "https://example.com/source",
+            "canonical_url": "https://example.com/source",
+            "meta_json": '{"generated_tags":["Rheingas","Gasflasche"]}',
+        }
+        post_id, _ = publish_article_draft(article)
+        self.assertEqual(post_id, 900)
+        post_calls = [call for call in mock_wp_request.call_args_list if call.kwargs.get("endpoint") == "posts"]
+        self.assertEqual(len(post_calls), 1)
+        payload = post_calls[0].kwargs.get("payload", {})
+        self.assertEqual(payload.get("tags"), [11, 12])
+
+    @patch("backend.app.wordpress._upload_featured_media")
+    @patch("backend.app.wordpress._wp_request")
+    def test_publish_converts_html_to_wp_blocks_without_html_block(self, mock_wp_request, mock_upload_media) -> None:
+        mock_wp_request.return_value = {"id": 111, "link": "https://example.org/?p=111"}
+        article = {
+            "title": "Block Test",
+            "content_rewritten": "<h2>Überschrift</h2><p>Absatz 1</p><ul><li>A</li><li>B</li></ul>",
+            "source_url": "https://example.com/source",
+            "canonical_url": "https://example.com/source",
+            "meta_json": "{}",
+        }
+        publish_article_draft(article)
+        payload = mock_wp_request.call_args.kwargs["payload"]
+        content = payload.get("content", "")
+        self.assertIn("<!-- wp:heading", content)
+        self.assertIn("<!-- wp:paragraph -->", content)
+        self.assertIn("<!-- wp:list -->", content)
+        self.assertNotIn("<!-- wp:html -->", content)
+
+
+if __name__ == "__main__":
+    unittest.main()

data/feeds.json

@@ -10,5 +10,33 @@
   {
     "url": "https://www.promobil.de/rss/ratgeber",
     "name": "Promobil Ratgeber"
+  },
+  {
+    "url": "https://www.presseportal.de/rss/rss2_vts.htx?q=camping&langid=1 ",
+    "name": "Presseportal Camping"
+  },
+  {
+    "url": "https://www.presseportal.de/rss/rss2_vts.htx?q=wohnmobil&langid=1 ",
+    "name": "Presseportal Wohnmobil"
+  },
+  {
+    "url": "https://caravan-news.de/rss/schlagzeilen.php",
+    "name": "Caravan News"
+  },
+  {
+    "url": "https://www.google.de/alerts/feeds/03077836356662926441/16793724126187652294",
+    "name": "Google Campingplatz"
+  },
+  {
+    "url": "https://www.google.de/alerts/feeds/03077836356662926441/987500860911797305",
+    "name": "Google VanLife"
+  },
+  {
+    "url": "https://www.google.de/alerts/feeds/03077836356662926441/4770194054838089856",
+    "name": "Google Camping Termine"
+  },
+  {
+    "url": "https://www.google.com/alerts/feeds/03077836356662926441/14685692393152596493",
+    "name": "Google Camping Messe 2025"
   }
 ]

docs/AUTOMATION.md

@@ -0,0 +1,190 @@
+# Automatischer Pipeline-Betrieb
+
+## Überblick
+
+Das System läuft vollautomatisch und benötigt nur noch gelegentliche Telegram-Interaktion.
+
+```
+N8N (2× täglich, 08:00 + 16:00 Uhr)
+  └─► POST /api/n8n/pipeline  (X-API-Key Header)
+        ├── RSS Ingestion (alle aktivierten Feeds)
+        ├── Relevanz-Score per GPT (0–100)
+        │     ├── Score ≥ 80 → Rewrite + WP-Draft + Telegram
+        │     ├── Score 60–79 → Telegram-Warnung + manueller Override möglich
+        │     └── Score < 60 → Abgelehnt + tägliche Telegram-Liste
+        └── Pipeline-Zusammenfassung via Telegram
+```
+
+---
+
+## Einrichtung
+
+### 1. Umgebungsvariablen setzen
+
+Kopiere `backend/.env.example` nach `backend/.env` und fülle alle Felder aus:
+
+```bash
+cp backend/.env.example backend/.env
+nano backend/.env
+```
+
+Wichtige Variablen:
+
+| Variable | Beschreibung |
+|----------|-------------|
+| `TELEGRAM_BOT_TOKEN` | Bot-Token von @BotFather |
+| `TELEGRAM_CHAT_ID` | Deine persönliche Chat-ID |
+| `TELEGRAM_WEBHOOK_SECRET` | Zufälliger String (≥ 20 Zeichen) |
+| `N8N_API_KEY` | Starker zufälliger API-Key |
+| `OPENAI_API_KEY` | OpenAI API-Key |
+| `WP_BASE_URL` | WordPress-URL |
+| `WP_USERNAME` | WordPress-Benutzername |
+| `WP_PASSWORD` | WordPress App-Passwort |
+
+### 2. Telegram-Webhook registrieren
+
+Nach dem Deployment einmalig aufrufen:
+
+```bash
+curl -X POST https://news.vanityontour.de/api/telegram/setup-webhook \
+  -H "Cookie: rss_news_session=<dein-session-token>"
+```
+
+Oder über die Admin-UI: Settings → Telegram Webhook einrichten.
+
+### 3. N8N Workflow einrichten
+
+In N8N einen neuen Workflow erstellen:
+
+**Trigger:** Cron
+- Zeitplan 1: `0 8 * * *` (täglich 08:00)
+- Zeitplan 2: `0 16 * * *` (täglich 16:00)
+
+**Aktion:** HTTP Request
+- Method: `POST`
+- URL: `https://news.vanityontour.de/api/n8n/pipeline`
+- Header: `X-API-Key: <dein-n8n-api-key>`
+
+**Fehlerbehandlung:** Bei HTTP-Fehler → E-Mail/Telegram-Alert
+
+---
+
+## Telegram-Befehle
+
+| Befehl | Funktion |
+|--------|----------|
+| `/run` | Pipeline manuell starten |
+| `/rejected` | Abgelehnte Artikel der letzten 3 Tage anzeigen |
+| `/status` | Aktuellen Pipeline-Status |
+| `/help` | Alle Befehle anzeigen |
+
+---
+
+## Telegram-Benachrichtigungen
+
+### Neuer Draft erstellt
+Wenn ein Artikel erfolgreich verarbeitet wurde:
+
+```
+✅ Neuer Draft erstellt
+📰 [Artikel-Titel]
+🟢 Relevanz-Score: 87/100
+📅 Vorgeschlagene Veröffentlichung: Mo, 24.03.2026 um 09:00 Uhr
+🏷 #VanLife #Camping #Wohnmobil
+🔗 Draft in WordPress öffnen
+
+  [✏️ Neu schreiben]  [❌ Verwerfen]
+```
+
+### Relevanz-Warnung (Score 60–79)
+```
+⚠️ Artikel mit niedrigem Relevanz-Score
+📰 [Artikel-Titel]
+🟡 Score: 72/100
+💬 Artikel behandelt hauptsächlich...
+🔗 Originalartikel
+
+  [➕ Trotzdem verarbeiten]  [❌ Ablehnen]
+```
+
+### Abgelehnte Artikel (Ende jedes Runs)
+Liste aller abgelehnten Artikel mit Override-Buttons für jeden einzelnen.
+
+---
+
+## Relevanz-Score
+
+Der GPT-basierte Score bewertet die Themenrelevanz für den VanLife/Camping-Blog:
+
+| Score | Aktion |
+|-------|--------|
+| 80–100 | Automatisch verarbeiten |
+| 60–79 | Telegram-Warnung, manueller Override |
+| 0–59 | Automatisch abgelehnt |
+
+Themen die hoch scored werden: Campingplätze, Stellplätze, Wohnmobile, Van-Ausbau,
+Outdoor-Equipment, Wandern, Naturreisen, Roadtrips, Camping-Tipps.
+
+Schwellwerte sind in `.env` konfigurierbar:
+```
+PIPELINE_RELEVANCE_AUTO=80
+PIPELINE_RELEVANCE_WARN=60
+```
+
+---
+
+## Veröffentlichungsplan
+
+- Maximal **2 Beiträge pro Tag**
+- Bevorzugte Zeiten: **09:00 und 14:00 Uhr** (CET)
+- Gleichmäßig über die Woche verteilt
+- Der Vorschlag erscheint in der Telegram-Nachricht
+- Manuell in WordPress setzen oder über WP Scheduling-Plugin automatisieren
+
+Einstellbar via:
+```
+PIPELINE_MAX_DRAFTS_PER_DAY=2
+PIPELINE_PUBLISH_HOURS=9,14
+```
+
+---
+
+## API-Endpunkte (N8N / extern)
+
+Alle externen Endpunkte benötigen den Header `X-API-Key: <N8N_API_KEY>`.
+
+| Methode | Endpunkt | Funktion |
+|---------|----------|----------|
+| `POST` | `/api/n8n/pipeline` | Komplette Pipeline starten |
+| `POST` | `/api/n8n/ingest` | Nur RSS-Import (ohne Rewrite) |
+
+---
+
+## Deployment (Hetzner via GitHub)
+
+Das Deployment läuft automatisch über GitHub Actions beim Push auf `main`:
+
+1. GitHub Action führt Tests aus
+2. Bei Erfolg: SSH-Deploy auf Hetzner
+3. `pip install -r requirements.txt`
+4. Systemd-Dienst `rss-app` neu starten
+
+Workflow-Dateien: `.github/workflows/test.yml` und `.github/workflows/deploy.yml`
+
+---
+
+## Troubleshooting
+
+**Pipeline läuft, aber keine Telegram-Nachrichten:**
+- `TELEGRAM_BOT_TOKEN` und `TELEGRAM_CHAT_ID` prüfen
+- Webhook-Status prüfen: `GET https://api.telegram.org/bot<TOKEN>/getWebhookInfo`
+
+**N8N bekommt 401:**
+- `N8N_API_KEY` in `.env` und N8N-Workflow-Header müssen übereinstimmen
+
+**Alle Artikel werden abgelehnt:**
+- `PIPELINE_RELEVANCE_WARN` temporär auf 40 senken zum Testen
+- Über `/rejected` + Override-Button manuell testen
+
+**Artikel werden doppelt importiert:**
+- Deduplication läuft über `source_url` (eindeutig). Bereits verarbeitete Artikel werden nie erneut als Draft angelegt.

docs/PROJECT_PLAN.md

@@ -0,0 +1,91 @@
+# Projektplan (Neustart)
+
+## Leitentscheidungen
+- Bestehendes Repository wird weiterverwendet.
+- Kein harter Endtermin: lauffaehig werden, dann iterativ verbessern.
+- Hetzner bleibt Laufzeitplattform.
+- WordPress (IONOS) bleibt vorerst Ziel fuer Publikation.
+- Auth initial nur mit einem User/Password.
+
+## Zielbild
+Eine modulare News-Pipeline mit klaren Stufen:
+1. Feed-Ingestion
+2. Inhaltsanalyse und Normalisierung
+3. Rewrite/Anreicherung
+4. Legal- und Qualitaetschecks
+5. WordPress-Publikation (Draft-first, Queue + Retry)
+6. Monitoring/Logging
+
+## Grobe Zeitplanung (ohne Fixtermine)
+- Phase 0: ca. 1 Woche
+- Phase 1: ca. 2-4 Wochen
+- Phase 2: ca. 2-3 Wochen
+- Phase 3: fortlaufend
+
+## Phasen
+
+### Phase 0 - Grundlagen (jetzt)
+- Doku und Wiki strukturieren
+- Source-Policy definieren
+- Redirect fuer `news.vanityontour.de` setzen
+- GitHub Project als zentrale Planung scharfstellen
+
+### Phase 1 - MVP Core
+- Neues FastAPI-Projektgeruest
+- SQLite-Datenmodell (feeds, articles, runs, source_policy)
+- Feed-Import mit Duplikaterkennung
+- Admin-Login (ein User)
+- Manuelle Review vor Publish
+- Admin-UI fuer Rechtscheck, Bildauswahl, Relevanzbewertung
+
+### Phase 2 - Automation
+- Job-Queue (asynchron)
+- Regelbasierte Scheduler
+- Retry/Dead-Letter-Handling
+- Robustes Error-Reporting
+- WordPress-Publisher (Draft) mit Mapping `article_id -> wp_post_id`
+
+### Phase 3 - Compliance und Skalierung
+- Source-Whitelisting mit Pflichtfeldern
+- Pflicht-Attribution pro Artikel
+- Qualitaetsmetriken und Audit-Logs
+- Optional: Passkey/WebAuthn
+
+## Aktueller Stand (Snapshot)
+- Backend/API + Admin-UI lauffaehig
+- Feed-Ingestion inkl. Originalartikel-Extraktion (Autor, Pressekontakt, Bilder)
+- Bildkuration:
+  - automatische Scoring-Reduktion (u. a. Presseportal `story_big` priorisiert)
+  - manuelle Auswahl/Ausblendung im UI
+- Rechts-/Publish-Gates aktiv:
+  - `legal_checked` Pflicht
+  - Hauptbild-Auswahl Pflicht
+  - Status-Workflow bis `published`
+- WordPress-Publishing:
+  - Queue + Retry + Job-Historie
+  - Draft-Erstellung/Update erfolgreich getestet
+- Exporte:
+  - JSON/CSV inkl. Datum/Alter/Relevanz + Attribution/Legal-Felder
+
+## Naechste Iteration (konkret)
+1. WordPress `featured_media` Upload aus ausgewaehltem Hauptbild
+2. Publish-HTML je Artikel verfeinern (strukturierter Body + konsistenter Quellenblock)
+3. Publisher als periodischen Worker (Timer/Cron/Systemd) auf Hetzner betreiben
+4. Monitoring/Alerting fuer Queue-Fehler + WP-API Fehlercodes
+
+## Architekturprinzipien
+- Idempotente Jobs
+- Trennung von UI, API, Worker
+- Strikte Validierung bei Quell-/Lizenzdaten
+- Expliziter Publish-Schritt, kein blindes Autoposting
+
+## Risiken
+- Lizenz-/Nutzungsbedingungen je Quelle variieren stark
+- Feeds aendern Struktur/Verfuegbarkeit
+- WordPress-API und Auth koennen regressionsanfaellig sein
+
+## Erfolgsmetriken
+- Zeit von Feed-Eingang bis Review-Ready
+- Quote sauber attribuierter Artikel
+- Fehlerrate pro Pipeline-Stufe
+- Anzahl manueller Eingriffe pro Woche

docs/SOURCE_POLICY.md

@@ -0,0 +1,81 @@
+# Source Policy und Feed-Vorschlaege
+
+## Grundsatz
+Es werden nur Quellen genutzt, deren Nutzungsbedingungen die geplante Nutzung erlauben oder fuer die eine explizite Genehmigung vorliegt.
+
+## Pflichtdaten pro Quelle
+- Quellname
+- Feed-URL
+- Originalartikel-URL
+- Autor/Herausgeber (wenn vorhanden)
+- Lizenz/Nutzungsgrundlage
+- Einschraenkungen (kommerziell, Bearbeitung, Bildrechte, Archivierung)
+- Datum der letzten Pruefung
+- Link auf Nutzungsbedingungen
+
+## Einstufung (Ampel)
+- Gruen: Nutzung fuer geplantes Modell klar erlaubt
+- Gelb: teilklar/mit Einschraenkungen, manuelle Pruefung erforderlich
+- Rot: fuer das Modell nicht geeignet ohne Zusatzvertrag
+
+## Verbindliche Regeln
+- Keine neue Quelle ohne Eintrag im Source-Register
+- Kein automatischer Publish bei Gelb/Rot
+- Bilder separat pruefen (Textrecht != Bildrecht)
+- Quartalsweiser Re-Check der Terms
+
+## Ersteinschaetzung (Stand: 16.02.2026)
+
+### Rot
+1. Reuters / Thomson Reuters
+- Grund: Inhalte sind urheberrechtlich geschuetzt; Reproduktion/Verteilung laut Terms nur mit vorheriger Zustimmung.
+- Folge: Nur mit explizitem Vertrag/Lizenz.
+- Referenz:
+  - https://www.thomsonreuters.com/en/terms-of-use
+
+2. tagesschau.de RSS
+- Grund: Inhalte nur privat/nicht-kommerziell; Veroeffentlichung grundsaetzlich nicht erlaubt (ausser explizit CC-lizenziert).
+- Folge: Nicht fuer das geplante Modell geeignet.
+- Referenz:
+  - https://www.tagesschau.de/infoservices/rssfeeds
+
+### Gelb
+1. Presseportal / ots
+- Grund: Redaktionelle Nutzung grundsaetzlich moeglich, aber Verantwortung liegt beim Verwender; darueber hinausgehende Geschaeftsnutzung nur mit Genehmigung.
+- Folge: Nur mit strikter Einzelpruefung pro Meldung (insb. Bild-/Drittrechte).
+- Referenz:
+  - https://www.presseportal.de/nutzungsbedingungen
+  - https://www.presseportal.de/feeds/
+
+2. Bundesbehoerden-RSS ohne explizite freie Weiterverwendungs-Lizenz
+- Grund: RSS wird bereitgestellt, aber nicht immer als offene Lizenz zur kommerziellen Nachnutzung formuliert.
+- Folge: Je Behoerde einzeln pruefen und dokumentieren.
+- Beispiele:
+  - https://www.bundesfinanzministerium.de/Content/DE/Standardartikel/Service/rss_base.html
+  - https://bmas.bund.de/EN/Services/RSS/rss.html
+
+### Gruen (mit korrekter Attribution)
+1. GovData / Open-Data-Portale mit `dl-de/by-2-0`, `dl-de/zero-2-0`, `CC BY 4.0` oder `CC0`
+- Grund: Diese Lizenzen erlauben grundsaetzlich auch kommerzielle Weiterverwendung (je nach Lizenzbedingungen).
+- Folge: Sehr gut fuer stabile Automatisierung geeignet.
+- Referenz:
+  - https://www.govdata.de/dl-de/by-2-0
+  - https://data.gov.de/informationen/lizenzen
+  - https://www.dcat-ap.de/def/licenses/dl-zero-de/2.0
+
+2. EU-Quellen mit expliziter `CC BY 4.0` Wiederverwendungsregel
+- Grund: EU-Inhalte sind haeufig unter CC BY 4.0 wiederverwendbar, sofern nicht anders gekennzeichnet.
+- Folge: Geeignet, wenn Drittinhalte ausgenommen werden.
+- Referenz:
+  - https://commission.europa.eu/legal-notice_en
+  - https://eur-lex.europa.eu/content/help/content/legal-notice/legal-notice.html
+
+## Quelle im Register freischalten (Definition of Done)
+- Terms-Link hinterlegt
+- Lizenzklasse (Gruen/Gelb/Rot) gesetzt
+- Pflicht-Attribution dokumentiert
+- Bildrechtsregel dokumentiert
+- Letzte Pruefung und Verantwortlicher gepflegt
+
+## Hinweis
+Keine Rechtsberatung. Bei unklaren oder wirtschaftlich kritischen Quellen ist eine juristische Prüfung sinnvoll.

docs/TODO.md

@@ -0,0 +1,38 @@
+# ToDo (Ein-Entwickler Setup)
+
+## Jetzt
+- [ ] WordPress Beitragsbild-Upload implementieren (`featured_media` aus ausgewaehltem Hauptbild)
+- [ ] WordPress-HTML-Ausgabe pro Artikel weiter verbessern (sauberes Layout, Quellenblock, Shortcodes falls noetig)
+- [ ] Publisher Fehlertexte fuer WP-Auth/Media/API in UI klarer darstellen
+- [ ] End-to-end Publish Smoke-Test dokumentieren (lokal + Hetzner)
+
+## MVP
+- [x] Neues Backend-Skelett (`backend/`) aufsetzen (FastAPI)
+- [x] Datenmodell in SQLite anlegen
+- [x] Feed-Ingestion Service bauen (ETag/Last-Modified)
+- [x] Duplikaterkennung ueber `source_url`, `guid`, Hash
+- [x] Login mit 1 Admin-Account implementieren
+- [x] Artikel-Review-Maske mit Statusworkflow
+- [x] WordPress-Publisher als separaten Service implementieren (Queue + Retry + Mapping)
+- [x] Bildvorschau + manuelle Bildauswahl im Admin-UI
+- [x] Automatische Bildreduktion/Scoring fuer Presseportal-Quellen
+- [x] Artikel-Datum + Relevanzscore im UI/Export
+
+## Recht/Qualitaet
+- [x] Source-Policy in DB + Admin-UI abbilden
+- [x] Pflichtfelder je Quelle erzwingen (Autor, URL, Lizenz, Hinweise)
+- [x] Auto-Block bei fehlender Lizenzinfo
+- [x] Pro Artikel Attribution-Block generieren
+- [x] Manuelle Rechtsfreigabe als Publish-Gate
+
+## Betrieb
+- [ ] Systemd-Service(s) fuer API/Worker erstellen
+- [ ] Nginx-Routing fuer neue App einrichten
+- [ ] Healthcheck-Endpunkte + Monitoring einrichten
+- [ ] Backup/Restore fuer DB dokumentieren
+
+## Spaeter
+- [ ] Passkey/WebAuthn evaluieren und optional einfuehren
+- [ ] Migration auf PostgreSQL bewerten
+- [ ] Teilautomatische Freigabe-Regeln definieren
+- [ ] KI-Rewrite mit Prompt-Versionierung + Qualitaetsmetriken wieder aktivieren

docs/roadmap-image-dedup.md

@@ -0,0 +1,37 @@
+# Roadmap: Bild-Deduplizierung & Medien-Hygiene
+
+## Ziele
+- Speicherverbrauch reduzieren
+- Medienbestand konsistent halten
+- Pipeline stabilisieren (keine Mehrfach-Uploads und -Speicherungen)
+
+## Vorgehen (sicher und reversibel)
+1. **Index aufbauen (Read-Only):**
+   - Alle Bilder (`.jpg/.jpeg/.png/.webp/.gif`) in definierten Verzeichnissen scannen
+   - Für jede Datei: `sha256` (Byte-Hash) + `pHash` (perzeptuell) berechnen
+   - Ergebnis als SQLite-Index + CSV-Report speichern
+
+2. **Kanonisierung & Referenzen prüfen:**
+   - Pro Duplikatgruppe genau **eine** kanonische Datei wählen (größte/neueste)
+   - Alle internen Referenzen (DB/JSON) testweise auf Kanon aktualisieren (Dry-Run)
+
+3. **Speicher sparen ohne Risiko:**
+   - Nicht-kanonische Dateien durch **Hardlinks** auf den Kanon ersetzen (gleiches FS)
+   - Alternativ: nur löschen, wenn Referenzen **sicher** auf Kanon zeigen
+
+4. **Prävention für die Zukunft:**
+   - Beim Speichern: **Content-Addressed Storage** (`<sha256>.<ext>`)
+   - In DB ein `content_hash`-Feld mit **Unique-Constraint**
+   - Vor jedem Speichern/Upload: Hash lookup → vorhandene Datei wiederverwenden
+
+## Akzeptanzkriterien
+- Report listet alle Duplikatgruppen mit Pfaden und Größenersparnis
+- Dry-Run zeigt geplante Änderungen ohne Schreibzugriff
+- Nach „Anwenden“ verweisen alle Referenzen auf die kanonische Datei
+- Re-Run findet **keine** Duplikate mehr (idempotent)
+- Rollback möglich via Backup der Reports/Indexdatei
+
+## Metriken
+- Anzahl Bilder vorher/nachher
+- Ersparter Speicher (MB/GB)
+- Anzahl gruppierter Duplikate

docs/wiki/Architektur.md

@@ -0,0 +1,29 @@
+# Architektur
+
+## Zielarchitektur
+- API: FastAPI
+- Worker: Queue-basierte Hintergrundjobs
+- DB: SQLite (MVP), spaeter optional PostgreSQL
+- Publisher: WordPress REST API
+- Frontend/Admin: schlanke Web-UI mit Login
+
+## Pipeline
+1. Feed Fetch
+2. Parse + Normalize
+3. Deduplicate
+4. Enrichment (Rewrite/Tags)
+5. Legal/Policy Check
+6. Publish (pending)
+
+## Datenobjekte (MVP)
+- `sources`
+- `feeds`
+- `articles`
+- `article_versions`
+- `runs`
+- `policy_checks`
+
+## Nichtziele (MVP)
+- Multi-User und Rollen
+- Vollautomatische Freigabe ohne Review
+- Komplexe externe SSO-Integration

docs/wiki/Deployment.md

@@ -0,0 +1,20 @@
+# Deployment (Hetzner + CloudPanel)
+
+## Umgebung
+- Host: Hetzner
+- Reverse Proxy: Nginx via CloudPanel
+- Ziel-Domain: `news.vanityontour.de`
+
+## Aktueller Zustand
+- Domain ist bis zum Go-Live auf `https://vanityontour.de` umgeleitet.
+
+## Zielzustand
+- `news.vanityontour.de` zeigt auf neue App (interner Port, z. B. `127.0.0.1:8501`)
+- API/Worker laufen als systemd-Services
+- TLS bleibt ueber CloudPanel/Nginx
+
+## Mindest-Checks nach Deployment
+- `curl -I https://news.vanityontour.de`
+- Login erreichbar
+- Feed-Import laeuft
+- WordPress-Testpublikation (pending) erfolgreich

docs/wiki/Home.md

@@ -0,0 +1,19 @@
+# Wiki Home
+
+## Zweck
+Dieses Wiki dokumentiert Architektur, Betrieb, Sicherheit, Recht und Roadmap des Neuaufbaus von `rss-news`.
+
+## Inhalte
+- `Architektur.md`
+- `Deployment.md`
+- `Security-Auth.md`
+- `Recht-Quellen.md`
+- `Operations-Runbook.md`
+- `Roadmap.md`
+- `Project-Board.md`
+
+## Projektsteuerung
+- GitHub Project #3: https://github.com/users/OliverGiertz/projects/3/views/1
+
+## Prinzip
+Dokumentation wird bei jeder relevanten Aenderung im selben Pull Request aktualisiert.

docs/wiki/Operations-Runbook.md

@@ -0,0 +1,43 @@
+# Operations Runbook
+
+## Daily Checks
+- App erreichbar
+- Queue/Worker aktiv
+- Letzte Feed-Laeufe erfolgreich
+- Keine auffaelligen Fehler im Log
+
+## Incident: Feed-Import faellt aus
+1. RSS-Quelle erreichbar?
+2. Parser-Fehler im Log?
+3. Rate Limits oder Blockaden?
+4. Retry-Queue pruefen
+
+## Incident: WordPress Publish faellt aus
+1. WP API erreichbar?
+2. Credentials gueltig?
+3. Payload-Validation/Tag-Fehler?
+4. Artikel in `pending` statt `failed` markieren, wenn unklar
+
+## Incident: Telegram-Buttons reagieren nicht / Befehle ignoriert
+
+**Ursache:** N8N "App Release - Telegram Bot"-Workflow hat den Webhook überschrieben.
+
+**Prüfen:**
+```bash
+curl -s "https://api.telegram.org/bot8403822424:AAGp8gZoNIGZv3IIan45q7P9HfM868qzXi4/getWebhookInfo" | python3 -m json.tool
+```
+→ `url` muss auf `https://news.vanityontour.de/telegram/webhook` zeigen
+→ `allowed_updates` muss `["message", "callback_query"]` enthalten
+
+**Webhook zurücksetzen:**
+```bash
+curl -s -X POST "https://api.telegram.org/bot8403822424:AAGp8gZoNIGZv3IIan45q7P9HfM868qzXi4/setWebhook" \
+  -H "Content-Type: application/json" \
+  -d '{"url":"https://news.vanityontour.de/telegram/webhook","allowed_updates":["message","callback_query"],"secret_token":"RWWAaBwfCUX9Y573JVkB9zAeloHsZZoruXOBBgUtsvU"}'
+```
+
+Vollständige Dokumentation: `projects/webhook/telegram-webhook-reset.md`
+
+## Backups
+- SQLite-Dump taeglich
+- Konfiguration und `.env` sicher sichern

docs/wiki/Project-Board.md

@@ -0,0 +1,28 @@
+# Project Board Workflow
+
+## Zentrale Steuerung
+- Board: https://github.com/users/OliverGiertz/projects/3/views/1
+- Board ist die einzige Quelle fuer Planungsstatus.
+
+## Arbeitsmodus (1 Entwickler)
+- Neue Arbeit immer als Issue anlegen
+- Issue direkt ins Project aufnehmen
+- Status nur im Project pflegen
+- PR/Commit auf Issue referenzieren
+
+## Empfohlene Status-Disziplin
+- `Todo`: noch nicht begonnen
+- `In Progress`: aktiv in Arbeit
+- `Done`: umgesetzt und dokumentiert
+
+## Konventionen fuer Issues
+- Prefix fuer Klarheit:
+  - `[MVP]`
+  - `[Infra]`
+  - `[Legal]`
+  - `[Bug]`
+- Definition of Done in jedem Issue notieren
+
+## Aktueller Backlog-Hinweis
+- Thema Userverwaltung ist fuer MVP obsolet (ein Admin-User).
+- Entsprechende Issues als `deferred` oder `closed` kennzeichnen.

docs/wiki/Recht-Quellen.md

@@ -0,0 +1,35 @@
+# Recht und Quellen
+
+## Grundregeln
+- Nur freigegebene Quellen aus Source-Register
+- Pflicht-Attribution pro Artikel
+- Rechte fuer Bilder separat pruefen
+- Kein Autopublish bei unklarer Lizenz
+
+## Bewertungsmodell
+- Gruen: Freie Nachnutzung klar erlaubt
+- Gelb: Nutzung mit Einschraenkungen/Einzelfallpruefung
+- Rot: Ohne Zusatzlizenz nicht geeignet
+
+## Aktuelle Referenzen
+- Reuters/Thomson Reuters Terms: https://www.thomsonreuters.com/en/terms-of-use
+- Presseportal Nutzungsbedingungen: https://www.presseportal.de/nutzungsbedingungen
+- tagesschau RSS-Hinweise: https://www.tagesschau.de/infoservices/rssfeeds
+- Datenlizenz Deutschland BY 2.0: https://www.govdata.de/dl-de/by-2-0
+- GovData Lizenzen: https://data.gov.de/informationen/lizenzen
+- EU Legal Notice (CC BY 4.0): https://commission.europa.eu/legal-notice_en
+
+## Review-Checkliste je Quelle
+1. Sind Bearbeitung und Veroeffentlichung erlaubt?
+2. Ist kommerzielle Nutzung erlaubt?
+3. Gibt es gesonderte Bildrechte?
+4. Ist die Quellenangabe vorgeschrieben?
+5. Gibt es Archivierungs- oder Weitergabebeschraenkungen?
+
+## Operativer Schutz
+- Source-Register als Pflicht vor Feed-Aktivierung
+- Auto-Block bei fehlenden Lizenzdaten
+- Quartalsweiser Terms-Recheck
+
+## Hinweis
+Keine Rechtsberatung. Finale Freigabe kritischer Quellen bei Bedarf juristisch validieren.

docs/wiki/Roadmap.md

@@ -0,0 +1,19 @@
+# Roadmap
+
+## Jetzt
+- Doku und Projektstruktur bereinigen
+- Redirect aktiv
+- Backlog auf Neustart ausrichten
+
+## Naechster Schritt
+- FastAPI-MVP implementieren
+- Login + Feed-Ingestion + Review + WordPress pending
+
+## Danach
+- Worker/Queue
+- Source-Policy Enforcement
+- Monitoring/Reporting
+- Optional Passkey
+
+## Steuerung
+Alle Arbeitsitems liegen im GitHub Project #3.

docs/wiki/Security-Auth.md

@@ -0,0 +1,16 @@
+# Security und Auth
+
+## Mindestanforderungen
+- Zugriff auf die WebApp nur mit Login
+- Ein aktiver Admin-User (kein Rollenmodell im MVP)
+- Passwort nicht im Repo, nur als Secret auf Server
+
+## Empfohlene Umsetzung
+- Session-basierte Auth (HTTP-only Cookies)
+- Passwort gehasht (Argon2 oder bcrypt)
+- Rate Limiting auf Login-Endpunkt
+- CSRF-Schutz fuer Form-Aktionen
+
+## Spaeter (optional)
+- Passkey/WebAuthn als zusaetzlicher Login-Faktor
+- IP-Allowlist fuer Admin-Zugang

pages/01_feed_manager.py

@@ -2,8 +2,13 @@
 
 import streamlit as st
 from main import load_feeds, save_feeds, load_articles
+from utils.css_loader import load_css, apply_dark_theme
 import logging
 
+# === CSS & Theme laden ===
+load_css()
+apply_dark_theme()
+
 # === Logging vorbereiten ===
 log_dir = "logs"
 log_file = f"{log_dir}/rss_tool.log"
@@ -15,17 +20,29 @@ logging.basicConfig(
 
 st.set_page_config(page_title="📡 Feed-Verwaltung")
 
-st.title("📡 RSS Feed-Verwaltung")
+# Header
+st.markdown("""
+<div class="main-header">
+    <h1>📡 RSS Feed-Verwaltung</h1>
+    <p>Verwalte deine RSS-Feeds zentral und effizient</p>
+</div>
+""", unsafe_allow_html=True)
+
 feeds = load_feeds()
 articles = load_articles()
 
 # === Neuen Feed hinzufügen ===
+st.markdown('<div class="filter-section">', unsafe_allow_html=True)
 st.subheader("➕ Neuen Feed hinzufügen")
 
 with st.form("add_feed_form"):
-    new_url = st.text_input("Feed URL", "")
-    new_name = st.text_input("Feed Name", "")
-    submitted = st.form_submit_button("Feed hinzufügen")
+    col1, col2 = st.columns(2)
+    with col1:
+        new_url = st.text_input("Feed URL", "", placeholder="https://example.com/feed.xml")
+    with col2:
+        new_name = st.text_input("Feed Name", "", placeholder="Beispiel News")
+    
+    submitted = st.form_submit_button("Feed hinzufügen", use_container_width=True)
     if submitted:
         if new_url and new_name:
             if not any(f.get("url") == new_url for f in feeds):
@@ -39,33 +56,185 @@ with st.form("add_feed_form"):
         else:
             st.error("❌ Bitte gib sowohl URL als auch Name ein.")
 
+st.markdown('</div>', unsafe_allow_html=True)
+
 # === Bestehende Feeds bearbeiten ===
-st.subheader("🛠️ Vorhandene Feeds bearbeiten oder löschen")
+st.subheader("🛠️ Vorhandene Feeds verwalten")
 
-for idx, feed in enumerate(feeds):
-    with st.expander(f"🔗 {feed.get('name')}"):
-        url = st.text_input(f"Feed-URL {idx}", value=feed.get("url"), key=f"url_{idx}")
-        name = st.text_input(f"Feed-Name {idx}", value=feed.get("name"), key=f"name_{idx}")
-        count = sum(1 for a in articles if a.get("source") == feed.get("url"))
-
-        col1, col2 = st.columns(2)
+if not feeds:
+    st.info("Noch keine Feeds konfiguriert. Füge oben deinen ersten Feed hinzu!")
+else:
+    for idx, feed in enumerate(feeds):
+        feed_url = feed.get("url", "")
+        feed_name = feed.get("name", "Unbekannt")
+        article_count = sum(1 for a in articles if a.get("source") == feed_url)
+        
+        # Feed Card
+        st.markdown(f"""
+        <div class="article-card">
+            <div style="display: flex; justify-content: space-between; align-items: center; margin-bottom: 1rem;">
+                <div>
+                    <h3 class="article-title">{feed_name}</h3>
+                    <div class="article-meta">{feed_url}</div>
+                </div>
+                <div>
+                    <span class="status-badge status-online">{article_count} Artikel</span>
+                </div>
+            </div>
+            <div class="article-footer">
+                📰 Verknüpfte Artikel: {article_count}
+            </div>
+        </div>
+        """, unsafe_allow_html=True)
+        
+        # Actions
+        col1, col2, col3 = st.columns(3)
 
         with col1:
-            if st.button("💾 Änderungen speichern", key=f"save_{idx}"):
-                old_url, old_name = feed.get("url"), feed.get("name")
-                feeds[idx]["url"] = url
-                feeds[idx]["name"] = name
-                save_feeds(feeds)
-                logging.info(f"✏️ Feed geändert: '{old_name}' ({old_url}) → '{name}' ({url})")
-                st.success("Änderungen gespeichert.")
-                st.rerun()
+            if st.button("💾 Bearbeiten", key=f"edit_{idx}", use_container_width=True):
+                st.session_state[f"edit_mode_{idx}"] = not st.session_state.get(f"edit_mode_{idx}", False)
 
         with col2:
-            if st.button("🗑️ Feed löschen", key=f"delete_{idx}"):
-                deleted_feed = feeds.pop(idx)
-                save_feeds(feeds)
-                logging.info(f"❌ Feed gelöscht: {deleted_feed.get('name')} ({deleted_feed.get('url')})")
-                st.warning("Feed gelöscht.")
-                st.rerun()
+            if st.button("🔄 Aktualisieren", key=f"refresh_{idx}", use_container_width=True):
+                with st.spinner(f"Aktualisiere Feed '{feed_name}'..."):
+                    # Hier könntest du eine einzelne Feed-Update-Funktion implementieren
+                    from main import process_articles
+                    existing_ids = [a["id"] for a in articles]
+                    process_articles(existing_ids)
+                    st.success(f"Feed '{feed_name}' aktualisiert!")
+                    st.rerun()
 
-        st.caption(f"📰 Verknüpfte Artikel: {count}")
+        with col3:
+            if st.button("🗑️ Löschen", key=f"delete_{idx}", use_container_width=True):
+                # Bestätigung über Session State
+                if not st.session_state.get(f"confirm_delete_{idx}", False):
+                    st.session_state[f"confirm_delete_{idx}"] = True
+                    st.warning(f"Klicke erneut um '{feed_name}' wirklich zu löschen!")
+                else:
+                    deleted_feed = feeds.pop(idx)
+                    save_feeds(feeds)
+                    logging.info(f"❌ Feed gelöscht: {deleted_feed.get('name')} ({deleted_feed.get('url')})")
+                    st.success(f"Feed '{feed_name}' wurde gelöscht.")
+                    # Cleanup Session State
+                    if f"confirm_delete_{idx}" in st.session_state:
+                        del st.session_state[f"confirm_delete_{idx}"]
+                    st.rerun()
+
+        # Edit Form (wenn aktiviert)
+        if st.session_state.get(f"edit_mode_{idx}", False):
+            st.markdown('<div class="filter-section" style="margin-top: 1rem;">', unsafe_allow_html=True)
+            st.write("**Feed bearbeiten:**")
+            
+            with st.form(f"edit_form_{idx}"):
+                col1, col2 = st.columns(2)
+                with col1:
+                    edited_url = st.text_input("Feed-URL", value=feed_url, key=f"edit_url_{idx}")
+                with col2:
+                    edited_name = st.text_input("Feed-Name", value=feed_name, key=f"edit_name_{idx}")
+                
+                form_col1, form_col2 = st.columns(2)
+                with form_col1:
+                    if st.form_submit_button("💾 Änderungen speichern", use_container_width=True):
+                        old_url, old_name = feed.get("url"), feed.get("name")
+                        feeds[idx]["url"] = edited_url
+                        feeds[idx]["name"] = edited_name
+                        save_feeds(feeds)
+                        logging.info(f"✏️ Feed geändert: '{old_name}' ({old_url}) → '{edited_name}' ({edited_url})")
+                        st.success("Änderungen gespeichert!")
+                        st.session_state[f"edit_mode_{idx}"] = False
+                        st.rerun()
+                
+                with form_col2:
+                    if st.form_submit_button("❌ Abbrechen", use_container_width=True):
+                        st.session_state[f"edit_mode_{idx}"] = False
+                        st.rerun()
+            
+            st.markdown('</div>', unsafe_allow_html=True)
+
+# === Feed-Statistiken ===
+if feeds:
+    st.markdown("---")
+    st.subheader("📊 Feed-Statistiken")
+    
+    col1, col2, col3 = st.columns(3)
+    
+    with col1:
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number">{}</div>
+            <div>Feeds Gesamt</div>
+        </div>
+        """.format(len(feeds)), unsafe_allow_html=True)
+    
+    with col2:
+        total_articles = len(articles)
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number">{}</div>
+            <div>Artikel Gesamt</div>
+        </div>
+        """.format(total_articles), unsafe_allow_html=True)
+    
+    with col3:
+        avg_articles = total_articles // len(feeds) if feeds else 0
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number">{}</div>
+            <div>Ø Artikel pro Feed</div>
+        </div>
+        """.format(avg_articles), unsafe_allow_html=True)
+
+# === Bulk Actions ===
+if feeds:
+    st.markdown("---")
+    st.subheader("⚡ Bulk-Aktionen")
+    
+    col1, col2 = st.columns(2)
+    
+    with col1:
+        if st.button("🔄 Alle Feeds aktualisieren", use_container_width=True):
+            with st.spinner("Aktualisiere alle Feeds..."):
+                from main import process_articles
+                existing_ids = [a["id"] for a in articles]
+                process_articles(existing_ids)
+                st.success(f"Alle {len(feeds)} Feeds wurden aktualisiert!")
+                st.rerun()
+    
+    with col2:
+        if st.button("📊 Feed-Performance anzeigen", use_container_width=True):
+            st.subheader("📈 Feed-Performance")
+            
+            # Performance-Daten sammeln
+            feed_performance = []
+            for feed in feeds:
+                feed_url = feed.get("url", "")
+                feed_name = feed.get("name", "Unbekannt")
+                feed_articles = [a for a in articles if a.get("source") == feed_url]
+                
+                performance = {
+                    "name": feed_name,
+                    "url": feed_url,
+                    "total_articles": len(feed_articles),
+                    "new_articles": len([a for a in feed_articles if a.get("status") == "New"]),
+                    "processed_articles": len([a for a in feed_articles if a.get("status") in ["Process", "Online", "WordPress Pending"]])
+                }
+                feed_performance.append(performance)
+            
+            # Sortiere nach Artikel-Anzahl
+            feed_performance.sort(key=lambda x: x["total_articles"], reverse=True)
+            
+            # Anzeige als Cards
+            for perf in feed_performance:
+                success_rate = (perf["processed_articles"] / perf["total_articles"] * 100) if perf["total_articles"] > 0 else 0
+                
+                st.markdown(f"""
+                <div class="article-card">
+                    <h3 class="article-title">{perf["name"]}</h3>
+                    <div class="article-footer">
+                        📰 {perf["total_articles"]} Artikel | 
+                        🆕 {perf["new_articles"]} Neu | 
+                        ✅ {perf["processed_articles"]} Verarbeitet | 
+                        📊 {success_rate:.1f}% Success Rate
+                    </div>
+                </div>
+                """, unsafe_allow_html=True)

pages/log_viewer.py

@@ -1,23 +1,297 @@
-# log_viewer.py
+# pages/log_viewer.py
 
 import streamlit as st
 import os
+from utils.css_loader import load_css, apply_dark_theme
+from datetime import datetime
+
+# === CSS & Theme laden ===
+load_css()
+apply_dark_theme()
 
 st.set_page_config(page_title="🧾 Log Viewer", layout="wide")
-st.title("🧾 Letzte Logeinträge anzeigen")
+
+# Header
+st.markdown("""
+<div class="main-header">
+    <h1>🧾 Log Viewer</h1>
+    <p>Überwache Systemaktivitäten und Debug-Informationen</p>
+</div>
+""", unsafe_allow_html=True)
 
 LOG_FILE = "logs/rss_tool.log"
 MAX_LINES = 500
 
-if not os.path.exists(LOG_FILE):
-    st.warning("Keine Logdatei gefunden.")
-else:
-    with open(LOG_FILE, "r") as f:
-        lines = f.readlines()
+# === Log-Datei Kontrollen ===
+st.markdown('<div class="filter-section">', unsafe_allow_html=True)
+st.subheader("📁 Log-Datei Optionen")
 
-    st.write(f"Letzte {min(len(lines), MAX_LINES)} Zeilen aus `{LOG_FILE}`:")
+col1, col2, col3, col4 = st.columns(4)
 
-    st.code("".join(lines[-MAX_LINES:]), language="text")
+with col1:
+    lines_to_show = st.selectbox(
+        "Anzahl Zeilen", 
+        [50, 100, 200, 500, 1000], 
+        index=3,  # Default: 500
+        key="lines_select"
+    )
 
-    if st.button("🔄 Neu laden"):
+with col2:
+    if st.button("🔄 Neu laden", use_container_width=True):
         st.rerun()
+
+with col3:
+    log_level_filter = st.selectbox(
+        "Log Level Filter",
+        ["Alle", "INFO", "WARNING", "ERROR", "DEBUG"],
+        key="level_filter"
+    )
+
+with col4:
+    search_term = st.text_input(
+        "Suche in Logs",
+        placeholder="Suchbegriff...",
+        key="log_search"
+    )
+
+st.markdown('</div>', unsafe_allow_html=True)
+
+# === Log-Datei Status ===
+if not os.path.exists(LOG_FILE):
+    st.markdown("""
+    <div class="wp-status">
+        <strong>⚠️ Keine Log-Datei gefunden</strong><br>
+        <div class="text-secondary">
+            Die Log-Datei wurde noch nicht erstellt oder befindet sich an einem anderen Ort.<br>
+            Erwarteter Pfad: <code>{}</code>
+        </div>
+    </div>
+    """.format(LOG_FILE), unsafe_allow_html=True)
+else:
+    # Datei-Informationen
+    file_size = os.path.getsize(LOG_FILE)
+    file_mtime = datetime.fromtimestamp(os.path.getmtime(LOG_FILE))
+    
+    col1, col2, col3 = st.columns(3)
+    
+    with col1:
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number">{:.1f} KB</div>
+            <div>Dateigröße</div>
+        </div>
+        """.format(file_size / 1024), unsafe_allow_html=True)
+    
+    with col2:
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number" style="font-size: 1.5rem;">{}</div>
+            <div>Letzte Änderung</div>
+        </div>
+        """.format(file_mtime.strftime("%H:%M:%S")), unsafe_allow_html=True)
+    
+    with col3:
+        # Zeilen zählen
+        try:
+            with open(LOG_FILE, "r", encoding="utf-8") as f:
+                total_lines = sum(1 for _ in f)
+        except:
+            total_lines = 0
+            
+        st.markdown("""
+        <div class="stats-card">
+            <div class="stats-number">{}</div>
+            <div>Zeilen Gesamt</div>
+        </div>
+        """.format(total_lines), unsafe_allow_html=True)
+
+    # === Log-Inhalt anzeigen ===
+    try:
+        with open(LOG_FILE, "r", encoding="utf-8") as f:
+            lines = f.readlines()
+
+        # Filter anwenden
+        filtered_lines = []
+        
+        for line in lines:
+            # Log Level Filter
+            if log_level_filter != "Alle":
+                if f" - {log_level_filter} - " not in line:
+                    continue
+            
+            # Suchfilter
+            if search_term and search_term.lower() not in line.lower():
+                continue
+                
+            filtered_lines.append(line)
+
+        # Anzahl begrenzen
+        display_lines = filtered_lines[-lines_to_show:] if len(filtered_lines) > lines_to_show else filtered_lines
+        
+        # Header für Log-Anzeige
+        st.subheader(f"📋 Log-Einträge ({len(display_lines)} von {len(filtered_lines)} gefilterten Zeilen)")
+        
+        if display_lines:
+            # Log-Inhalt mit Syntax-Highlighting
+            log_content = "".join(display_lines)
+            
+            # Farbkodierung für verschiedene Log-Level
+            colored_content = log_content
+            colored_content = colored_content.replace(" - ERROR - ", " - 🔴 ERROR - ")
+            colored_content = colored_content.replace(" - WARNING - ", " - 🟡 WARNING - ")
+            colored_content = colored_content.replace(" - INFO - ", " - 🔵 INFO - ")
+            colored_content = colored_content.replace(" - DEBUG - ", " - ⚪ DEBUG - ")
+            
+            # Log in Card anzeigen
+            st.markdown("""
+            <div class="article-card">
+                <h3 class="article-title">📄 Log-Ausgabe</h3>
+                <div class="article-meta">
+                    Letzte {count} Einträge | Filter: {level} | Suche: "{search}"
+                </div>
+            </div>
+            """.format(
+                count=len(display_lines),
+                level=log_level_filter,
+                search=search_term or "Keine"
+            ), unsafe_allow_html=True)
+            
+            # Code-Block mit Logs
+            st.code(colored_content, language="text")
+            
+            # Download-Button
+            st.download_button(
+                label="💾 Log-Datei herunterladen",
+                data=log_content,
+                file_name=f"rss_tool_log_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt",
+                mime="text/plain",
+                use_container_width=True
+            )
+            
+        else:
+            st.markdown("""
+            <div class="wp-status">
+                <strong>🔍 Keine Log-Einträge gefunden</strong><br>
+                <div class="text-secondary">
+                    Mit den aktuellen Filtern wurden keine Log-Einträge gefunden.<br>
+                    Versuche andere Filter-Einstellungen.
+                </div>
+            </div>
+            """, unsafe_allow_html=True)
+
+    except Exception as e:
+        st.markdown(f"""
+        <div class="wp-status">
+            <strong>❌ Fehler beim Lesen der Log-Datei</strong><br>
+            <div class="text-secondary">
+                {str(e)}
+            </div>
+        </div>
+        """, unsafe_allow_html=True)
+
+# === Log-Level Erklärung ===
+with st.expander("ℹ️ Log-Level Erklärung", expanded=False):
+    st.markdown("""
+    <div class="article-card">
+        <h3 class="article-title">📖 Log-Level Bedeutung</h3>
+        <div class="article-summary">
+            <strong>🔵 INFO:</strong> Normale Programmaktivitäten (Feed-Updates, Artikel verarbeitet)<br>
+            <strong>🟡 WARNING:</strong> Potentielle Probleme (Duplikate, fehlende Daten)<br>
+            <strong>🔴 ERROR:</strong> Fehler die das Programm beeinträchtigen<br>
+            <strong>⚪ DEBUG:</strong> Detaillierte Entwickler-Informationen
+        </div>
+    </div>
+    """, unsafe_allow_html=True)
+
+# === Log-Datei verwalten ===
+st.markdown("---")
+st.subheader("🛠️ Log-Datei Verwaltung")
+
+col1, col2, col3 = st.columns(3)
+
+with col1:
+    if st.button("🗑️ Log-Datei leeren", use_container_width=True):
+        if st.button("⚠️ Wirklich leeren?", key="confirm_clear"):
+            try:
+                with open(LOG_FILE, "w", encoding="utf-8") as f:
+                    f.write("")
+                st.success("Log-Datei wurde geleert!")
+                st.rerun()
+            except Exception as e:
+                st.error(f"Fehler beim Leeren der Log-Datei: {e}")
+
+with col2:
+    if st.button("📦 Log archivieren", use_container_width=True):
+        try:
+            archive_name = f"rss_tool_log_archive_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt"
+            with open(LOG_FILE, "r", encoding="utf-8") as f:
+                log_data = f.read()
+            
+            st.download_button(
+                label=f"💾 {archive_name}",
+                data=log_data,
+                file_name=archive_name,
+                mime="text/plain",
+                key="archive_download"
+            )
+        except Exception as e:
+            st.error(f"Fehler beim Archivieren: {e}")
+
+with col3:
+    if st.button("📊 Log-Statistiken", use_container_width=True):
+        if os.path.exists(LOG_FILE):
+            try:
+                with open(LOG_FILE, "r", encoding="utf-8") as f:
+                    all_lines = f.readlines()
+                
+                # Statistiken berechnen
+                total_lines = len(all_lines)
+                info_count = sum(1 for line in all_lines if " - INFO - " in line)
+                warning_count = sum(1 for line in all_lines if " - WARNING - " in line)
+                error_count = sum(1 for line in all_lines if " - ERROR - " in line)
+                debug_count = sum(1 for line in all_lines if " - DEBUG - " in line)
+                
+                st.subheader("📈 Log-Statistiken")
+                
+                col1, col2, col3, col4 = st.columns(4)
+                
+                with col1:
+                    st.markdown("""
+                    <div class="stats-card">
+                        <div class="stats-number">{}</div>
+                        <div>🔵 INFO</div>
+                    </div>
+                    """.format(info_count), unsafe_allow_html=True)
+                
+                with col2:
+                    st.markdown("""
+                    <div class="stats-card">
+                        <div class="stats-number">{}</div>
+                        <div>🟡 WARNING</div>
+                    </div>
+                    """.format(warning_count), unsafe_allow_html=True)
+                
+                with col3:
+                    st.markdown("""
+                    <div class="stats-card">
+                        <div class="stats-number">{}</div>
+                        <div>🔴 ERROR</div>
+                    </div>
+                    """.format(error_count), unsafe_allow_html=True)
+                
+                with col4:
+                    st.markdown("""
+                    <div class="stats-card">
+                        <div class="stats-number">{}</div>
+                        <div>⚪ DEBUG</div>
+                    </div>
+                    """.format(debug_count), unsafe_allow_html=True)
+                
+            except Exception as e:
+                st.error(f"Fehler beim Berechnen der Statistiken: {e}")
+
+# === Auto-Refresh Option ===
+if st.checkbox("🔄 Auto-Refresh (30s)", key="auto_refresh"):
+    import time
+    time.sleep(30)
+    st.rerun()

pytest.ini

@@ -0,0 +1,4 @@
+[pytest]
+testpaths = backend/tests
+python_files = test_*.py
+addopts = -q --maxfail=1

requirements.txt

@@ -53,3 +53,6 @@ typing_extensions==4.14.0
 tzdata==2025.2
 urllib3==2.5.0
 typer[all]==0.12.3
+# Bild- und Hashing-Bibliotheken für Dedupe
+Pillow>=10.0.0        # Bildverarbeitung (öffnet Bilder für pHash)
+ImageHash>=4.3        # Perzeptueller Hash (pHash) für Near-Duplicate Erkennung

scripts/smoke_backend.sh

@@ -0,0 +1,33 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+if [[ -z "${BASE_URL:-}" ]]; then
+  echo "BASE_URL fehlt (z. B. https://news.vanityontour.de)"
+  exit 1
+fi
+
+if [[ -z "${APP_ADMIN_USERNAME:-}" || -z "${APP_ADMIN_PASSWORD:-}" ]]; then
+  echo "APP_ADMIN_USERNAME/APP_ADMIN_PASSWORD fehlen"
+  exit 1
+fi
+
+cookie_file="$(mktemp)"
+trap 'rm -f "$cookie_file"' EXIT
+
+echo "[1/4] Healthcheck"
+curl -fsS "${BASE_URL}/health" | grep -q '"status":"ok"'
+
+echo "[2/4] Login"
+curl -fsS -c "$cookie_file" \
+  -H "Content-Type: application/json" \
+  -X POST "${BASE_URL}/auth/login" \
+  -d "{\"username\":\"${APP_ADMIN_USERNAME}\",\"password\":\"${APP_ADMIN_PASSWORD}\"}" \
+  | grep -q '"ok":true'
+
+echo "[3/4] Protected Endpoint"
+curl -fsS -b "$cookie_file" "${BASE_URL}/api/protected" | grep -q '"ok":true'
+
+echo "[4/4] Pipeline Status"
+curl -fsS -b "$cookie_file" "${BASE_URL}/api/pipeline/status" | grep -q '"stage":"skeleton+db"'
+
+echo "Smoke test erfolgreich."

static/styles.css

@@ -0,0 +1,491 @@
+/* ===============================================
+   RSS Feed Manager - Zentrale CSS-Datei
+   Dark-Mode optimiert mit Fallbacks
+   =============================================== */
+
+/* === ROOT VARIABLEN === */
+:root {
+    /* Dark Mode Farbpalette */
+    --bg-primary: #1e1e1e;
+    --bg-secondary: #2d2d30;
+    --bg-card: #2d2d30;
+    --bg-header: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    --bg-filter: #363636;
+    
+    /* Text Farben */
+    --text-primary: #ffffff;
+    --text-secondary: #b0b0b0;
+    --text-muted: #888888;
+    --text-accent: #667eea;
+    
+    /* Status Farben */
+    --status-new: #2196f3;
+    --status-new-bg: #1565c0;
+    --status-rewrite: #ff9800;
+    --status-rewrite-bg: #ef6c00;
+    --status-process: #9c27b0;
+    --status-process-bg: #6a1b9a;
+    --status-online: #4caf50;
+    --status-online-bg: #2e7d32;
+    --status-hold: #e91e63;
+    --status-hold-bg: #ad1457;
+    --status-trash: #f44336;
+    --status-trash-bg: #c62828;
+    --status-wp-pending: #00bcd4;
+    --status-wp-pending-bg: #0097a7;
+    
+    /* Borders & Shadows */
+    --border-color: #404040;
+    --shadow-light: 0 2px 8px rgba(0, 0, 0, 0.3);
+    --shadow-hover: 0 8px 20px rgba(0, 0, 0, 0.4);
+    
+    /* Accent Colors */
+    --gradient-primary: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    --gradient-secondary: linear-gradient(135deg, #f093fb 0%, #f5576c 100%);
+}
+
+/* === LIGHT MODE FALLBACKS === */
+[data-theme="light"], .stApp[data-theme="light"] {
+    --bg-primary: #ffffff;
+    --bg-secondary: #f8f9fa;
+    --bg-card: #ffffff;
+    --bg-filter: #f0f2f6;
+    
+    --text-primary: #212529;
+    --text-secondary: #495057;
+    --text-muted: #6c757d;
+    --text-accent: #667eea;
+    
+    --border-color: #dee2e6;
+    --shadow-light: 0 2px 8px rgba(0, 0, 0, 0.1);
+    --shadow-hover: 0 8px 20px rgba(0, 0, 0, 0.15);
+}
+
+/* === GLOBALE RESETS === */
+* {
+    box-sizing: border-box;
+}
+
+/* === HAUPTCONTAINER === */
+.main-header {
+    background: var(--bg-header);
+    padding: 2rem;
+    border-radius: 12px;
+    margin-bottom: 2rem;
+    color: var(--text-primary);
+    text-align: center;
+    box-shadow: var(--shadow-light);
+}
+
+.main-header h1 {
+    color: var(--text-primary) !important;
+    margin: 0 0 0.5rem 0;
+    font-size: 2.5rem;
+    font-weight: 700;
+}
+
+.main-header p {
+    color: rgba(255, 255, 255, 0.9) !important;
+    margin: 0;
+    font-size: 1.1rem;
+}
+
+/* === ARTIKEL CARDS === */
+.article-card {
+    background: var(--bg-card) !important;
+    border-radius: 12px;
+    padding: 1.5rem;
+    margin-bottom: 1rem;
+    box-shadow: var(--shadow-light);
+    border-left: 4px solid var(--text-accent);
+    border: 1px solid var(--border-color);
+    transition: all 0.3s ease;
+    color: var(--text-primary) !important;
+}
+
+.article-card:hover {
+    transform: translateY(-2px);
+    box-shadow: var(--shadow-hover);
+    border-color: var(--text-accent);
+}
+
+.article-card h3,
+.article-card .article-title {
+    color: var(--text-primary) !important;
+    margin: 0 0 0.5rem 0;
+    font-size: 1.2rem;
+    font-weight: 600;
+    line-height: 1.4;
+}
+
+.article-card .article-meta {
+    color: var(--text-secondary) !important;
+    font-size: 0.9rem;
+    margin-bottom: 1rem;
+}
+
+.article-card .article-summary {
+    color: var(--text-secondary) !important;
+    line-height: 1.5;
+    margin-bottom: 1rem;
+}
+
+.article-card .article-footer {
+    color: var(--text-muted) !important;
+    font-size: 0.85rem;
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+}
+
+/* === STATUS BADGES === */
+.status-badge {
+    padding: 0.3rem 0.8rem;
+    border-radius: 20px;
+    font-size: 0.8rem;
+    font-weight: 600;
+    margin-right: 0.5rem;
+    display: inline-block;
+    color: white !important;
+}
+
+.status-new { 
+    background-color: var(--status-new-bg) !important; 
+    color: white !important; 
+}
+
+.status-rewrite { 
+    background-color: var(--status-rewrite-bg) !important; 
+    color: white !important; 
+}
+
+.status-process { 
+    background-color: var(--status-process-bg) !important; 
+    color: white !important; 
+}
+
+.status-online { 
+    background-color: var(--status-online-bg) !important; 
+    color: white !important; 
+}
+
+.status-hold { 
+    background-color: var(--status-hold-bg) !important; 
+    color: white !important; 
+}
+
+.status-trash { 
+    background-color: var(--status-trash-bg) !important; 
+    color: white !important; 
+}
+
+.status-wp-pending { 
+    background-color: var(--status-wp-pending-bg) !important; 
+    color: white !important; 
+}
+
+/* === FILTER SECTION === */
+.filter-section {
+    background: var(--bg-filter) !important;
+    padding: 1.5rem;
+    border-radius: 12px;
+    margin-bottom: 2rem;
+    border: 1px solid var(--border-color);
+    box-shadow: var(--shadow-light);
+}
+
+.filter-section h3 {
+    color: var(--text-primary) !important;
+    margin: 0 0 1rem 0;
+    font-size: 1.3rem;
+    font-weight: 600;
+}
+
+/* === STATS CARDS === */
+.stats-card {
+    background: var(--bg-card) !important;
+    padding: 1.5rem;
+    border-radius: 12px;
+    text-align: center;
+    box-shadow: var(--shadow-light);
+    border: 1px solid var(--border-color);
+    transition: transform 0.2s ease;
+}
+
+.stats-card:hover {
+    transform: translateY(-2px);
+}
+
+.stats-number {
+    font-size: 2.5rem;
+    font-weight: 700;
+    color: var(--text-accent) !important;
+    margin-bottom: 0.5rem;
+    display: block;
+}
+
+.stats-card div:last-child {
+    color: var(--text-secondary) !important;
+    font-weight: 500;
+    font-size: 1rem;
+}
+
+/* === WORDPRESS STATUS === */
+.wp-status {
+    background: var(--bg-card) !important;
+    padding: 1rem;
+    border-radius: 8px;
+    margin: 1rem 0;
+    border-left: 4px solid var(--status-wp-pending);
+    border: 1px solid var(--border-color);
+    box-shadow: var(--shadow-light);
+}
+
+.wp-status strong {
+    color: var(--text-primary) !important;
+}
+
+.wp-status small {
+    color: var(--text-muted) !important;
+}
+
+/* === IMAGE GALLERY === */
+.image-gallery {
+    display: flex;
+    gap: 1rem;
+    overflow-x: auto;
+    padding: 1rem 0;
+}
+
+.image-item {
+    min-width: 200px;
+    text-align: center;
+    background: var(--bg-card);
+    padding: 1rem;
+    border-radius: 8px;
+    border: 1px solid var(--border-color);
+}
+
+.image-item img {
+    border-radius: 6px;
+    max-width: 100%;
+}
+
+.image-item strong,
+.image-item p {
+    color: var(--text-primary) !important;
+}
+
+.image-item small {
+    color: var(--text-muted) !important;
+}
+
+/* === BUTTONS & ACTIONS === */
+.action-button {
+    margin: 0.25rem;
+    border-radius: 6px;
+}
+
+/* Streamlit Button Overrides */
+.stButton > button {
+    background: var(--gradient-primary) !important;
+    color: white !important;
+    border: none !important;
+    border-radius: 8px !important;
+    font-weight: 600 !important;
+    transition: all 0.2s ease !important;
+}
+
+.stButton > button:hover {
+    transform: translateY(-1px) !important;
+    box-shadow: 0 4px 12px rgba(102, 126, 234, 0.4) !important;
+}
+
+/* === SELECTBOX & INPUT OVERRIDES === */
+.stSelectbox > div > div {
+    background-color: var(--bg-card) !important;
+    color: var(--text-primary) !important;
+    border: 1px solid var(--border-color) !important;
+}
+
+.stTextInput > div > div > input {
+    background-color: var(--bg-card) !important;
+    color: var(--text-primary) !important;
+    border: 1px solid var(--border-color) !important;
+}
+
+/* === TABS === */
+.stTabs [data-baseweb="tab-list"] {
+    background-color: var(--bg-secondary) !important;
+    border-radius: 8px;
+    padding: 0.25rem;
+}
+
+.stTabs [data-baseweb="tab"] {
+    color: var(--text-secondary) !important;
+    background-color: transparent !important;
+    border-radius: 6px !important;
+    font-weight: 600 !important;
+}
+
+.stTabs [aria-selected="true"] {
+    background-color: var(--text-accent) !important;
+    color: white !important;
+}
+
+/* === EXPANDER === */
+.streamlit-expanderHeader {
+    background-color: var(--bg-card) !important;
+    color: var(--text-primary) !important;
+    border: 1px solid var(--border-color) !important;
+    border-radius: 8px !important;
+}
+
+.streamlit-expanderContent {
+    background-color: var(--bg-card) !important;
+    border: 1px solid var(--border-color) !important;
+    border-top: none !important;
+}
+
+/* === METRICS === */
+.metric-container {
+    background: var(--bg-card) !important;
+    padding: 1rem;
+    border-radius: 8px;
+    border: 1px solid var(--border-color);
+    text-align: center;
+}
+
+.metric-container [data-testid="metric-container"] {
+    background: transparent !important;
+}
+
+.metric-container [data-testid="metric-container"] > div {
+    color: var(--text-primary) !important;
+}
+
+/* === CODE BLOCKS === */
+.stCodeBlock {
+    background-color: var(--bg-secondary) !important;
+    border: 1px solid var(--border-color) !important;
+    border-radius: 8px;
+}
+
+/* === SUCCESS/ERROR/WARNING/INFO === */
+.stAlert {
+    border-radius: 8px !important;
+    border: 1px solid var(--border-color) !important;
+}
+
+.stSuccess {
+    background-color: rgba(76, 175, 80, 0.1) !important;
+    color: var(--status-online) !important;
+    border-color: var(--status-online) !important;
+}
+
+.stError {
+    background-color: rgba(244, 67, 54, 0.1) !important;
+    color: var(--status-trash) !important;
+    border-color: var(--status-trash) !important;
+}
+
+.stWarning {
+    background-color: rgba(255, 152, 0, 0.1) !important;
+    color: var(--status-rewrite) !important;
+    border-color: var(--status-rewrite) !important;
+}
+
+.stInfo {
+    background-color: rgba(33, 150, 243, 0.1) !important;
+    color: var(--status-new) !important;
+    border-color: var(--status-new) !important;
+}
+
+/* === SIDEBAR === */
+.css-1d391kg {
+    background-color: var(--bg-secondary) !important;
+}
+
+.css-1d391kg .stMarkdown {
+    color: var(--text-primary) !important;
+}
+
+/* === RESPONSIVE DESIGN === */
+@media (max-width: 768px) {
+    .main-header {
+        padding: 1.5rem;
+    }
+    
+    .main-header h1 {
+        font-size: 2rem;
+    }
+    
+    .article-card {
+        padding: 1rem;
+    }
+    
+    .stats-card {
+        padding: 1rem;
+    }
+    
+    .stats-number {
+        font-size: 2rem;
+    }
+}
+
+/* === UTILITY CLASSES === */
+.text-primary { color: var(--text-primary) !important; }
+.text-secondary { color: var(--text-secondary) !important; }
+.text-muted { color: var(--text-muted) !important; }
+.text-accent { color: var(--text-accent) !important; }
+
+.bg-card { background-color: var(--bg-card) !important; }
+.bg-secondary { background-color: var(--bg-secondary) !important; }
+
+.border-radius { border-radius: 8px; }
+.shadow-light { box-shadow: var(--shadow-light); }
+.shadow-hover { box-shadow: var(--shadow-hover); }
+
+/* === SCROLLBAR STYLING === */
+::-webkit-scrollbar {
+    width: 8px;
+    height: 8px;
+}
+
+::-webkit-scrollbar-track {
+    background: var(--bg-secondary);
+    border-radius: 4px;
+}
+
+::-webkit-scrollbar-thumb {
+    background: var(--text-muted);
+    border-radius: 4px;
+}
+
+::-webkit-scrollbar-thumb:hover {
+    background: var(--text-secondary);
+}
+
+/* === LOADING SPINNER === */
+@keyframes spin {
+    0% { transform: rotate(0deg); }
+    100% { transform: rotate(360deg); }
+}
+
+.loading-spinner {
+    border: 4px solid var(--bg-secondary);
+    border-top: 4px solid var(--text-accent);
+    border-radius: 50%;
+    width: 40px;
+    height: 40px;
+    animation: spin 1s linear infinite;
+    margin: 0 auto 1rem auto;
+}
+
+/* === FOCUS STATES === */
+.stButton > button:focus,
+.stSelectbox > div > div:focus,
+.stTextInput > div > div > input:focus {
+    outline: 2px solid var(--text-accent) !important;
+    outline-offset: 2px !important;
+}

tools/image_deduper.py

@@ -0,0 +1,303 @@
+#!/usr/bin/env python3
+"""
+image_deduper.py — Finde und bereinige Bild-Dubletten sicher & reversibel.
+
+Funktionen:
+- Scan: Verzeichnisse rekursiv scannen, sha256 + pHash berechnen
+- Report: CSV + menschenlesbare Zusammenfassung mit Gruppen
+- Apply: Duplikate auf kanonische Datei umbiegen (Hardlink/Löschen)
+- Optional: DB-Referenzen aktualisieren (SQLite/SQLModel kompatibel)
+
+Nutzung (Beispiele):
+  # 1) Nur scannen + reporten (keine Änderungen):
+  python tools/image_deduper.py scan --roots media,assets/images --out-dir .dedupe --phash
+
+  # 2) Report anzeigen:
+  python tools/image_deduper.py report --index .dedupe/index.sqlite --csv
+
+  # 3) Anwenden (Hardlinks setzen, Dry-Run):
+  python tools/image_deduper.py apply --index .dedupe/index.sqlite --mode hardlink --dry-run
+
+  # 4) Anwenden (wirklich ändern):
+  python tools/image_deduper.py apply --index .dedupe/index.sqlite --mode hardlink
+
+  # 5) Referenzen in DB aktualisieren (optional):
+  python tools/image_deduper.py apply --index .dedupe/index.sqlite --update-db sqlite:///./rssbot.db --dry-run
+"""
+
+import argparse
+import csv
+import hashlib
+import os
+import sqlite3
+import sys
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Iterable, List, Optional, Tuple
+
+try:
+    from PIL import Image
+except ImportError:
+    Image = None
+
+try:
+    import imagehash  # type: ignore
+except ImportError:
+    imagehash = None
+
+
+IMAGE_EXTS = {".jpg", ".jpeg", ".png", ".webp", ".gif"}
+DEFAULT_INDEX = ".dedupe/index.sqlite"
+DEFAULT_REPORT = ".dedupe/report.csv"
+
+
+def sha256_file(path: Path, bufsize: int = 1024 * 1024) -> str:
+    h = hashlib.sha256()
+    with path.open("rb") as f:
+        while True:
+            b = f.read(bufsize)
+            if not b:
+                break
+            h.update(b)
+    return h.hexdigest()
+
+
+def calc_phash(path: Path) -> Optional[str]:
+    if Image is None or imagehash is None:
+        return None
+    try:
+        with Image.open(path) as im:
+            im = im.convert("RGB")
+            ph = imagehash.phash(im, hash_size=16)  # 16x16 → 256-bit
+            return str(ph)
+    except Exception:
+        return None
+
+
+def ensure_dir(p: Path):
+    p.mkdir(parents=True, exist_ok=True)
+
+
+def init_index(db_path: Path):
+    conn = sqlite3.connect(str(db_path))
+    cur = conn.cursor()
+    cur.execute("""
+        CREATE TABLE IF NOT EXISTS files (
+            id INTEGER PRIMARY KEY,
+            path TEXT NOT NULL UNIQUE,
+            size INTEGER NOT NULL,
+            mtime REAL NOT NULL,
+            sha256 TEXT NOT NULL,
+            phash TEXT,
+            ext TEXT NOT NULL
+        );
+    """)
+    cur.execute("CREATE INDEX IF NOT EXISTS idx_sha256 ON files (sha256);")
+    cur.execute("CREATE INDEX IF NOT EXISTS idx_phash ON files (phash);")
+    conn.commit()
+    conn.close()
+
+
+def is_image(path: Path) -> bool:
+    return path.suffix.lower() in IMAGE_EXTS
+
+
+def walk_images(roots: List[Path]) -> Iterable[Path]:
+    for root in roots:
+        for p in root.rglob("*"):
+            if p.is_file() and is_image(p):
+                yield p
+
+
+def upsert_file(db_path: Path, path: Path, sha256: str, phash: Optional[str]):
+    st = path.stat()
+    row = (str(path), st.st_size, st.st_mtime, sha256, phash, path.suffix.lower())
+    conn = sqlite3.connect(str(db_path))
+    cur = conn.cursor()
+    cur.execute("""
+        INSERT INTO files (path, size, mtime, sha256, phash, ext)
+        VALUES (?, ?, ?, ?, ?, ?)
+        ON CONFLICT(path) DO UPDATE SET
+            size=excluded.size,
+            mtime=excluded.mtime,
+            sha256=excluded.sha256,
+            phash=excluded.phash,
+            ext=excluded.ext;
+    """, row)
+    conn.commit()
+    conn.close()
+
+
+def group_by_sha256(db_path: Path) -> List[List[Tuple[int, str, int]]]:
+    """Return list of groups: [(id, path, size), ...] where sha256 identical and len(group) > 1."""
+    conn = sqlite3.connect(str(db_path))
+    cur = conn.cursor()
+    cur.execute("""
+        SELECT sha256 FROM files GROUP BY sha256 HAVING COUNT(*) > 1;
+    """)
+    hashes = [r[0] for r in cur.fetchall()]
+    groups = []
+    for h in hashes:
+        cur.execute("SELECT id, path, size FROM files WHERE sha256=?", (h,))
+        rows = cur.fetchall()
+        groups.append([(rid, rpath, rsize) for rid, rpath, rsize in rows])
+    conn.close()
+    return groups
+
+
+def write_csv_report(db_path: Path, csv_path: Path) -> Tuple[int, int]:
+    groups = group_by_sha256(db_path)
+    ensure_dir(csv_path.parent)
+    total_dups = 0
+    total_savings = 0
+    with csv_path.open("w", newline="", encoding="utf-8") as f:
+        w = csv.writer(f)
+        w.writerow(["group_id", "canonical_path", "dup_path", "dup_size_bytes"])
+        gid = 0
+        for g in groups:
+            if not g:
+                continue
+            # Kanon: größte Datei (oder erste)
+            canonical = max(g, key=lambda x: x[2])
+            for rid, path, size in g:
+                if path == canonical[1]:
+                    continue
+                total_dups += 1
+                total_savings += size
+                w.writerow([gid, canonical[1], path, size])
+            gid += 1
+    return total_dups, total_savings
+
+
+def apply_hardlink(canonical: Path, dup: Path, dry_run: bool) -> None:
+    # Ersetzt dup durch Hardlink auf canonical (gleiche Partition nötig)
+    if dry_run:
+        return
+    tmp = dup.with_suffix(dup.suffix + ".dedupe.tmp")
+    dup.unlink()                # entferne dup
+    os.link(canonical, tmp)     # hardlink temp
+    tmp.replace(dup)            # atomarer move
+
+
+def apply_delete(dup: Path, dry_run: bool) -> None:
+    if dry_run:
+        return
+    dup.unlink()
+
+
+@dataclass
+class ApplyStats:
+    processed: int = 0
+    errors: int = 0
+    saved_bytes: int = 0
+
+
+def apply_changes(csv_report: Path, mode: str, dry_run: bool) -> ApplyStats:
+    stats = ApplyStats()
+    with csv_report.open("r", encoding="utf-8") as f:
+        r = csv.DictReader(f)
+        for row in r:
+            canonical = Path(row["canonical_path"])
+            dup = Path(row["dup_path"])
+            size = int(row["dup_size_bytes"])
+            try:
+                if mode == "hardlink":
+                    apply_hardlink(canonical, dup, dry_run)
+                elif mode == "delete":
+                    apply_delete(dup, dry_run)
+                else:
+                    raise ValueError("mode must be 'hardlink' or 'delete'")
+                stats.processed += 1
+                stats.saved_bytes += size
+            except Exception as e:
+                stats.errors += 1
+                print(f"[ERROR] {dup}: {e}", file=sys.stderr)
+    return stats
+
+
+def parse_roots(roots_arg: str) -> List[Path]:
+    parts = [Path(p.strip()) for p in roots_arg.split(",") if p.strip()]
+    for p in parts:
+        if not p.exists():
+            raise FileNotFoundError(f"Root not found: {p}")
+    return parts
+
+
+def cmd_scan(args):
+    out_dir = Path(args.out_dir)
+    index = Path(args.index or DEFAULT_INDEX)
+    ensure_dir(out_dir)
+    ensure_dir(index.parent)
+    init_index(index)
+    roots = parse_roots(args.roots)
+
+    count = 0
+    for path in walk_images(roots):
+        try:
+            h = sha256_file(path)
+            ph = calc_phash(path) if args.phash else None
+            upsert_file(index, path, h, ph)
+            count += 1
+            if count % 500 == 0:
+                print(f"... indexed {count} files")
+        except Exception as e:
+            print(f"[WARN] {path}: {e}", file=sys.stderr)
+
+    dups, savings = write_csv_report(index, Path(args.report or DEFAULT_REPORT))
+    print(f"Indexed {count} images. Found duplicate files: {dups}, potential savings: {savings/1_000_000:.2f} MB")
+    print(f"Index: {index}")
+    print(f"Report: {args.report or DEFAULT_REPORT}")
+
+
+def cmd_report(args):
+    index = Path(args.index or DEFAULT_INDEX)
+    csv_path = Path(args.report or DEFAULT_REPORT)
+    dups, savings = write_csv_report(index, csv_path)
+    print(f"Duplicates: {dups}, potential savings: {savings/1_000_000:.2f} MB")
+    if args.csv:
+        print(f"CSV written: {csv_path}")
+
+
+def cmd_apply(args):
+    csv_report = Path(args.report or DEFAULT_REPORT)
+    if not csv_report.exists():
+        raise FileNotFoundError(f"Report not found: {csv_report}")
+    stats = apply_changes(csv_report, args.mode, args.dry_run)
+    print(f"Processed: {stats.processed}, Errors: {stats.errors}, Saved: {stats.saved_bytes/1_000_000:.2f} MB (mode={args.mode}, dry_run={args.dry_run})")
+    if args.update_db:
+        # Platzhalter: hier könntest du eure DB-Referenzen aktualisieren (falls Bilder-Paths in DB gespeichert sind).
+        # Beispiel: SQLModel mit Tabelle ImageMeta(content_hash UNIQUE, local_path) → auf kanonischen Pfad umbiegen.
+        print(f"[INFO] DB update requested for: {args.update_db} (implementierung projektspezifisch)")
+
+
+def main():
+    ap = argparse.ArgumentParser(description="Bild-Deduplizierung (scan/report/apply)")
+    sub = ap.add_subparsers(dest="cmd", required=True)
+
+    sc = sub.add_parser("scan", help="Verzeichnisse scannen und Index/Report erstellen")
+    sc.add_argument("--roots", required=True, help="Kommagetrennte Wurzelpfade, z.B. 'media,assets/images'")
+    sc.add_argument("--out-dir", default=".dedupe", help="Ausgabeverzeichnis für Index/Reports")
+    sc.add_argument("--index", help="Pfad zur SQLite-Indexdatei (default .dedupe/index.sqlite)")
+    sc.add_argument("--report", help="Pfad zum CSV-Report (default .dedupe/report.csv)")
+    sc.add_argument("--phash", action="store_true", help="Perzeptuellen Hash berechnen (für zukünftige Near-Dups)")
+    sc.set_defaults(func=cmd_scan)
+
+    rp = sub.add_parser("report", help="Report neu generieren/anzeigen")
+    rp.add_argument("--index", help="Pfad zur SQLite-Indexdatei")
+    rp.add_argument("--report", help="Pfad zum CSV-Report")
+    rp.add_argument("--csv", action="store_true", help="CSV-Pfad ausgeben")
+    rp.set_defaults(func=cmd_report)
+
+    aply = sub.add_parser("apply", help="Änderungen anwenden (Hardlink/Delete)")
+    aply.add_argument("--report", help="Pfad zum CSV-Report")
+    aply.add_argument("--mode", choices=["hardlink", "delete"], default="hardlink", help="Strategie für Duplikate")
+    aply.add_argument("--dry-run", action="store_true", help="Nur anzeigen, nichts ändern")
+    aply.add_argument("--update-db", help="Optional: DB-URL für Referenz-Updates (projektspezifisch)")
+    aply.set_defaults(func=cmd_apply)
+
+    args = ap.parse_args()
+    args.func(args)
+
+
+if __name__ == "__main__":
+    main()

utils/config.py

@@ -0,0 +1,51 @@
+import os
+from typing import Dict, List
+from dotenv import load_dotenv
+
+load_dotenv()
+
+
+def validate_env() -> Dict:
+    """Validiert sicherheitsrelevante .env-Variablen.
+
+    Returns dict with: ok: bool, errors: List[str], warnings: List[str], summary: Dict[str, bool]
+    """
+    errors: List[str] = []
+    warnings: List[str] = []
+
+    wp_base_url = os.getenv("WP_BASE_URL", "").strip()
+    wp_user = os.getenv("WP_USERNAME", "").strip()
+    wp_pass = os.getenv("WP_PASSWORD", "").strip()
+    wp_b64 = os.getenv("WP_AUTH_BASE64", "").strip()
+    openai_key = os.getenv("OPENAI_API_KEY", "").strip()
+
+    # WP_BASE_URL Pflicht
+    if not wp_base_url:
+        errors.append("WP_BASE_URL fehlt in .env")
+    elif not (wp_base_url.startswith("http://") or wp_base_url.startswith("https://")):
+        errors.append("WP_BASE_URL muss mit http:// oder https:// beginnen")
+
+    # Auth-Creds: entweder Base64 ODER Username+Password
+    if not wp_b64 and not (wp_user and wp_pass):
+        errors.append("Entweder WP_AUTH_BASE64 oder WP_USERNAME + WP_PASSWORD in .env setzen")
+
+    # Empfehlungen
+    if not wp_b64 and (wp_user and wp_pass):
+        warnings.append("WP_AUTH_BASE64 nicht gesetzt – Empfehlung: Base64 nutzen (Application Password)")
+
+    if not openai_key:
+        warnings.append("OPENAI_API_KEY ist nicht gesetzt – Umschreibungsfunktion ist deaktiviert")
+
+    summary = {
+        "WP_BASE_URL": bool(wp_base_url),
+        "WP_USERNAME": bool(wp_user),
+        "WP_PASSWORD": bool(wp_pass),
+        "WP_AUTH_BASE64": bool(wp_b64),
+        "OPENAI_API_KEY": bool(openai_key),
+    }
+
+    return {"ok": len(errors) == 0, "errors": errors, "warnings": warnings, "summary": summary}
+
+
+__all__ = ["validate_env"]
+

utils/css_loader.py

@@ -0,0 +1,367 @@
+# utils/css_loader.py
+
+import streamlit as st
+import os
+from pathlib import Path
+
+def load_css():
+    """
+    Lädt die zentrale CSS-Datei und injiziert sie in die Streamlit-App
+    """
+    try:
+        # Pfad zur CSS-Datei bestimmen
+        css_file = Path(__file__).parent.parent / "static" / "styles.css"
+        
+        if css_file.exists():
+            with open(css_file, "r", encoding="utf-8") as f:
+                css_content = f.read()
+            
+            # CSS in Streamlit injizieren
+            st.markdown(f"""
+            <style>
+            {css_content}
+            </style>
+            """, unsafe_allow_html=True)
+            
+            return True
+        else:
+            # Fallback: CSS-Datei erstellen
+            create_css_file()
+            return load_css()  # Rekursiver Aufruf nach Erstellung
+            
+    except Exception as e:
+        st.error(f"Fehler beim Laden der CSS-Datei: {e}")
+        return False
+
+def create_css_file():
+    """
+    Erstellt die CSS-Datei falls sie nicht existiert
+    """
+    css_content = """/* ===============================================
+   RSS Feed Manager - Zentrale CSS-Datei
+   Dark-Mode optimiert mit Fallbacks
+   =============================================== */
+
+/* === ROOT VARIABLEN === */
+:root {
+    /* Dark Mode Farbpalette */
+    --bg-primary: #1e1e1e;
+    --bg-secondary: #2d2d30;
+    --bg-card: #2d2d30;
+    --bg-header: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    --bg-filter: #363636;
+    
+    /* Text Farben */
+    --text-primary: #ffffff;
+    --text-secondary: #b0b0b0;
+    --text-muted: #888888;
+    --text-accent: #667eea;
+    
+    /* Status Farben */
+    --status-new: #2196f3;
+    --status-new-bg: #1565c0;
+    --status-rewrite: #ff9800;
+    --status-rewrite-bg: #ef6c00;
+    --status-process: #9c27b0;
+    --status-process-bg: #6a1b9a;
+    --status-online: #4caf50;
+    --status-online-bg: #2e7d32;
+    --status-hold: #e91e63;
+    --status-hold-bg: #ad1457;
+    --status-trash: #f44336;
+    --status-trash-bg: #c62828;
+    --status-wp-pending: #00bcd4;
+    --status-wp-pending-bg: #0097a7;
+    
+    /* Borders & Shadows */
+    --border-color: #404040;
+    --shadow-light: 0 2px 8px rgba(0, 0, 0, 0.3);
+    --shadow-hover: 0 8px 20px rgba(0, 0, 0, 0.4);
+    
+    /* Accent Colors */
+    --gradient-primary: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    --gradient-secondary: linear-gradient(135deg, #f093fb 0%, #f5576c 100%);
+}
+
+/* === LIGHT MODE FALLBACKS === */
+[data-theme="light"], .stApp[data-theme="light"] {
+    --bg-primary: #ffffff;
+    --bg-secondary: #f8f9fa;
+    --bg-card: #ffffff;
+    --bg-filter: #f0f2f6;
+    
+    --text-primary: #212529;
+    --text-secondary: #495057;
+    --text-muted: #6c757d;
+    --text-accent: #667eea;
+    
+    --border-color: #dee2e6;
+    --shadow-light: 0 2px 8px rgba(0, 0, 0, 0.1);
+    --shadow-hover: 0 8px 20px rgba(0, 0, 0, 0.15);
+}
+
+/* === HAUPTCONTAINER === */
+.main-header {
+    background: var(--bg-header);
+    padding: 2rem;
+    border-radius: 12px;
+    margin-bottom: 2rem;
+    color: var(--text-primary);
+    text-align: center;
+    box-shadow: var(--shadow-light);
+}
+
+.main-header h1 {
+    color: var(--text-primary) !important;
+    margin: 0 0 0.5rem 0;
+    font-size: 2.5rem;
+    font-weight: 700;
+}
+
+.main-header p {
+    color: rgba(255, 255, 255, 0.9) !important;
+    margin: 0;
+    font-size: 1.1rem;
+}
+
+/* === ARTIKEL CARDS === */
+.article-card {
+    background: var(--bg-card) !important;
+    border-radius: 12px;
+    padding: 1.5rem;
+    margin-bottom: 1rem;
+    box-shadow: var(--shadow-light);
+    border-left: 4px solid var(--text-accent);
+    border: 1px solid var(--border-color);
+    transition: all 0.3s ease;
+    color: var(--text-primary) !important;
+}
+
+.article-card:hover {
+    transform: translateY(-2px);
+    box-shadow: var(--shadow-hover);
+    border-color: var(--text-accent);
+}
+
+.article-card h3,
+.article-card .article-title {
+    color: var(--text-primary) !important;
+    margin: 0 0 0.5rem 0;
+    font-size: 1.2rem;
+    font-weight: 600;
+    line-height: 1.4;
+}
+
+.article-card .article-meta {
+    color: var(--text-secondary) !important;
+    font-size: 0.9rem;
+    margin-bottom: 1rem;
+}
+
+.article-card .article-summary {
+    color: var(--text-secondary) !important;
+    line-height: 1.5;
+    margin-bottom: 1rem;
+}
+
+.article-card .article-footer {
+    color: var(--text-muted) !important;
+    font-size: 0.85rem;
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+}
+
+/* === STATUS BADGES === */
+.status-badge {
+    padding: 0.3rem 0.8rem;
+    border-radius: 20px;
+    font-size: 0.8rem;
+    font-weight: 600;
+    margin-right: 0.5rem;
+    display: inline-block;
+    color: white !important;
+}
+
+.status-new { 
+    background-color: var(--status-new-bg) !important; 
+    color: white !important; 
+}
+
+.status-rewrite { 
+    background-color: var(--status-rewrite-bg) !important; 
+    color: white !important; 
+}
+
+.status-process { 
+    background-color: var(--status-process-bg) !important; 
+    color: white !important; 
+}
+
+.status-online { 
+    background-color: var(--status-online-bg) !important; 
+    color: white !important; 
+}
+
+.status-hold { 
+    background-color: var(--status-hold-bg) !important; 
+    color: white !important; 
+}
+
+.status-trash { 
+    background-color: var(--status-trash-bg) !important; 
+    color: white !important; 
+}
+
+.status-wp-pending { 
+    background-color: var(--status-wp-pending-bg) !important; 
+    color: white !important; 
+}
+
+/* === FILTER SECTION === */
+.filter-section {
+    background: var(--bg-filter) !important;
+    padding: 1.5rem;
+    border-radius: 12px;
+    margin-bottom: 2rem;
+    border: 1px solid var(--border-color);
+    box-shadow: var(--shadow-light);
+}
+
+.filter-section h3 {
+    color: var(--text-primary) !important;
+    margin: 0 0 1rem 0;
+    font-size: 1.3rem;
+    font-weight: 600;
+}
+
+/* === STATS CARDS === */
+.stats-card {
+    background: var(--bg-card) !important;
+    padding: 1.5rem;
+    border-radius: 12px;
+    text-align: center;
+    box-shadow: var(--shadow-light);
+    border: 1px solid var(--border-color);
+    transition: transform 0.2s ease;
+}
+
+.stats-card:hover {
+    transform: translateY(-2px);
+}
+
+.stats-number {
+    font-size: 2.5rem;
+    font-weight: 700;
+    color: var(--text-accent) !important;
+    margin-bottom: 0.5rem;
+    display: block;
+}
+
+.stats-card div:last-child {
+    color: var(--text-secondary) !important;
+    font-weight: 500;
+    font-size: 1rem;
+}
+
+/* === WORDPRESS STATUS === */
+.wp-status {
+    background: var(--bg-card) !important;
+    padding: 1rem;
+    border-radius: 8px;
+    margin: 1rem 0;
+    border-left: 4px solid var(--status-wp-pending);
+    border: 1px solid var(--border-color);
+    box-shadow: var(--shadow-light);
+}
+
+.wp-status strong {
+    color: var(--text-primary) !important;
+}
+
+.wp-status small {
+    color: var(--text-muted) !important;
+}
+
+/* === BUTTONS & ACTIONS === */
+.stButton > button {
+    background: var(--gradient-primary) !important;
+    color: white !important;
+    border: none !important;
+    border-radius: 8px !important;
+    font-weight: 600 !important;
+    transition: all 0.2s ease !important;
+}
+
+.stButton > button:hover {
+    transform: translateY(-1px) !important;
+    box-shadow: 0 4px 12px rgba(102, 126, 234, 0.4) !important;
+}
+
+/* === SELECTBOX & INPUT OVERRIDES === */
+.stSelectbox > div > div {
+    background-color: var(--bg-card) !important;
+    color: var(--text-primary) !important;
+    border: 1px solid var(--border-color) !important;
+}
+
+.stTextInput > div > div > input {
+    background-color: var(--bg-card) !important;
+    color: var(--text-primary) !important;
+    border: 1px solid var(--border-color) !important;
+}
+
+/* === RESPONSIVE DESIGN === */
+@media (max-width: 768px) {
+    .main-header {
+        padding: 1.5rem;
+    }
+    
+    .main-header h1 {
+        font-size: 2rem;
+    }
+    
+    .article-card {
+        padding: 1rem;
+    }
+    
+    .stats-card {
+        padding: 1rem;
+    }
+    
+    .stats-number {
+        font-size: 2rem;
+    }
+}
+"""
+    
+    try:
+        # Static-Ordner erstellen falls nicht vorhanden
+        static_dir = Path(__file__).parent.parent / "static"
+        static_dir.mkdir(exist_ok=True)
+        
+        # CSS-Datei schreiben
+        css_file = static_dir / "styles.css"
+        with open(css_file, "w", encoding="utf-8") as f:
+            f.write(css_content)
+            
+        return True
+    except Exception as e:
+        st.error(f"Fehler beim Erstellen der CSS-Datei: {e}")
+        return False
+
+def apply_dark_theme():
+    """
+    Wendet das Dark Theme an (zusätzlich zur CSS-Datei)
+    """
+    st.markdown("""
+    <script>
+    // Dark Theme Detection und Anwendung
+    const prefersDark = window.matchMedia('(prefers-color-scheme: dark)').matches;
+    if (prefersDark) {
+        document.documentElement.setAttribute('data-theme', 'dark');
+    } else {
+        document.documentElement.setAttribute('data-theme', 'light');
+    }
+    </script>
+    """, unsafe_allow_html=True)

utils/wordpress_uploader.py

@@ -11,17 +11,16 @@ from dotenv import load_dotenv
 
 load_dotenv()
 
-# WordPress API Konfiguration
-WP_BASE_URL = os.getenv("WP_BASE_URL", "https://vanityontour.de")
-WP_USERNAME = os.getenv("WP_USERNAME", "ogiertz")
-WP_PASSWORD = os.getenv("WP_PASSWORD", "whNEx9aZCIUXViV89Z3e7Z03")
-WP_AUTH_BASE64 = os.getenv("WP_AUTH_BASE64", "b2dpZXJ0ejp3aE5FeDlhWkNJVVhWaVY4OVozZTdaMDM=")
-WP_API_ENDPOINT = f"{WP_BASE_URL}/wp-json/wp/v2"
+# WordPress API Konfiguration – ausschließlich aus .env
+WP_BASE_URL = os.getenv("WP_BASE_URL")
+WP_USERNAME = os.getenv("WP_USERNAME")
+WP_PASSWORD = os.getenv("WP_PASSWORD")
+WP_AUTH_BASE64 = os.getenv("WP_AUTH_BASE64")
 
 # Request-Konfiguration
 REQUEST_TIMEOUT = 30
 MAX_RETRIES = 3
-USER_AGENT = 'RSS-Feed-Manager/1.6.1'
+USER_AGENT = 'RSS-Feed-Manager/1.7.x'
 
 class WordPressUploader:
     """
@@ -30,41 +29,43 @@ class WordPressUploader:
     """
     
     def __init__(self):
-        self.base_url = WP_BASE_URL
-        self.api_endpoint = WP_API_ENDPOINT
+        # Basis-URL validieren und Endpunkt bauen
+        if not WP_BASE_URL:
+            raise ValueError("WP_BASE_URL nicht gesetzt. Bitte .env konfigurieren.")
+        self.base_url = WP_BASE_URL.rstrip('/')
+        self.api_endpoint = f"{self.base_url}/wp-json/wp/v2"
+        
+        # Zugangsdaten (aus .env)
         self.username = WP_USERNAME
         self.password = WP_PASSWORD
         self.auth_base64 = WP_AUTH_BASE64
-        
+
+        if not self.auth_base64 and not (self.username and self.password):
+            raise ValueError("WordPress-Authentifizierung nicht konfiguriert. WP_AUTH_BASE64 oder WP_USERNAME + WP_PASSWORD setzen.")
+
         # Session für bessere Performance
         self.session = requests.Session()
-        
+
         # Authentifizierung über Authorization Header mit Base64
         if self.auth_base64:
-            # Verwende bereitgestellten Base64-String
             self.session.headers.update({
                 'Authorization': f'Basic {self.auth_base64}',
                 'User-Agent': USER_AGENT,
                 'Content-Type': 'application/json',
                 'Accept': 'application/json'
             })
-            logging.info("✅ WordPress-Authentifizierung: Verwende bereitgestellten Base64-String")
+            logging.info("✅ WordPress-Authentifizierung: Base64-String verwendet")
         else:
-            # Fallback: Generiere Base64 aus Username/Password
-            if self.username and self.password:
-                credentials = f"{self.username}:{self.password}"
-                encoded_credentials = base64.b64encode(credentials.encode('utf-8')).decode('utf-8')
-                self.session.headers.update({
-                    'Authorization': f'Basic {encoded_credentials}',
-                    'User-Agent': USER_AGENT,
-                    'Content-Type': 'application/json',
-                    'Accept': 'application/json'
-                })
-                logging.info("✅ WordPress-Authentifizierung: Base64 aus Username/Password generiert")
-            else:
-                logging.error("❌ WordPress-Authentifizierung: Weder Base64-String noch Username/Password verfügbar")
-                raise ValueError("WordPress-Authentifizierung nicht konfiguriert")
-        
+            credentials = f"{self.username}:{self.password}"
+            encoded_credentials = base64.b64encode(credentials.encode('utf-8')).decode('utf-8')
+            self.session.headers.update({
+                'Authorization': f'Basic {encoded_credentials}',
+                'User-Agent': USER_AGENT,
+                'Content-Type': 'application/json',
+                'Accept': 'application/json'
+            })
+            logging.info("✅ WordPress-Authentifizierung: Base64 aus Username/Password generiert")
+
         # Standard-Kategorie ID ermitteln
         self.default_category_id = self._get_default_category_id()
         
@@ -335,11 +336,7 @@ class WordPressUploader:
         Testet die Verbindung zur WordPress API mit Base64-Authentifizierung
         """
         try:
-            logging.info("🔧 Teste WordPress-API-Verbindung mit Base64-Auth...")
-            
-            # Debug: Auth-Header prüfen
-            auth_header = self.session.headers.get('Authorization', 'Nicht gesetzt')
-            logging.info(f"🔑 Authorization Header: {auth_header[:20]}..." if len(auth_header) > 20 else f"🔑 Authorization Header: {auth_header}")
+            logging.info("🔧 Teste WordPress-API-Verbindung...")
             
             # Einfache Abfrage der Kategorien als Test
             response = self.session.get(
@@ -349,7 +346,6 @@ class WordPressUploader:
             )
             
             logging.info(f"📡 API-Response Status: {response.status_code}")
-            logging.info(f"📡 API-Response Headers: {dict(response.headers)}")
             
             if response.status_code == 200:
                 logging.info("✅ WordPress-API-Verbindung erfolgreich")
@@ -465,4 +461,4 @@ def upload_single_article_to_wordpress(article: Dict) -> Tuple[bool, str, Option
         return False, connection_msg, None
     
     # Artikel hochladen
-    return uploader.upload_article(article)
+    return uploader.upload_article(article)