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>

backend/app/main.py

@@ -1,8 +1,10 @@
+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
@@ -637,20 +639,26 @@ def _require_api_key(request: Request) -> None:
         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)
-    import asyncio
-    import logging
+
+    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():
-        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)
+        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"}

backend/app/scheduler.py

@@ -11,6 +11,7 @@ from __future__ import annotations
 
 import base64
 import json
+import threading
 import urllib.request
 from datetime import date, datetime, timedelta, timezone
 from typing import Any
@@ -18,6 +19,9 @@ 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)
@@ -256,34 +260,77 @@ def reserve_publish_slot(article_id: int) -> str:
 
     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.
     """
-    # Check if already has a slot
-    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 slot, re-assign below
-
+    # 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()
-    result = _find_next_free_slot(wp_occupied, lookahead_days=30)
-    if result:
-        candidate, hour = result
-    else:
-        candidate = _today_cet() + timedelta(days=1)
-        hours = _preferred_hours()
-        hour = hours[0] if hours else 9
 
-    iso_ts = f"{candidate.isoformat()}T{hour:02d}:00:00"
-    with get_conn() as conn:
-        conn.execute(
-            "UPDATE articles SET scheduled_publish_at = ? WHERE id = ?",
-            (iso_ts, article_id),
-        )
-    return _format_slot(candidate, hour)
+    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)