feat(pipeline): add two-stage article quality gate (min word count)

Stage 1 (before OpenAI rewrite): reject if raw content < pipeline_min_words_raw (default 120)
Stage 2 (after rewrite): reject if rewritten text < pipeline_min_words_rewritten (default 150)

Both stages set status='error' with a descriptive note and skip WP draft creation.
The reserved publish slot is released so it stays available for the next article.
Quality rejections don't abort the pipeline — processing continues with the next article.

New config settings (overridable via .env):
  PIPELINE_MIN_WORDS_RAW=120
  PIPELINE_MIN_WORDS_REWRITTEN=150

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

backend/app/config.py

@@ -46,6 +46,8 @@ class Settings(BaseSettings):
     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)
 
 
 @lru_cache(maxsize=1)

backend/app/pipeline.py

@@ -109,10 +109,35 @@ def _store_relevance(article_id: int, relevance: dict[str, Any]) -> None:
 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", article_id)
+    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:
@@ -342,6 +367,14 @@ def _process_article(article: dict[str, Any], stats: PipelineStats, settings: An
                 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)
+            stats.rejected_articles.append(get_article_by_id(article_id) or {})
+            logger.info("Artikel #%d wegen Qualitätsprüfung abgelehnt: %s", article_id, 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}")

backend/app/scheduler.py

@@ -165,6 +165,15 @@ def _find_next_free_slot(
     return tomorrow, _preferred_hours()[0] if _preferred_hours() else 9
 
 
+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()