feat(#179 Phase 4.3): pm-sample-bundle.sh fuer 5 PMs (PM + Thread) zur Sichtung

Skript laeuft fuer N_SAMPLES (Default 5) hochbewertete Antraege jeweils generate_draft() mit style='pm' und style='thread' aus. Idempotent ueber das presse_drafts.style-Schema. Manueller Aufruf: ./scripts/pm-sample-bundle.sh gwoe-antragspruefer-dev Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-06 16:12:41 +02:00 · 2026-05-06 16:12:41 +02:00 · 6a78dee2d1
commit 6a78dee2d1
parent a2b8f8c6fe
1 changed files with 90 additions and 0 deletions
--- a/scripts/pm-sample-bundle.sh
+++ b/scripts/pm-sample-bundle.sh
@ -0,0 +1,90 @@
+#!/bin/bash
+# PM-Sample-Bundle: erzeugt fuer 5 hochbewertete Anträge je eine PM
+# (Klassisch + Thread) zur manuellen Qualitätssichtung.
+#
+# Manueller Aufruf:
+#   ./scripts/pm-sample-bundle.sh gwoe-antragspruefer-dev
+#
+# Bei mehrfachem Aufruf werden bestehende PMs nicht regeneriert
+# (Idempotenz pro drucksache+news_url+style).
+
+set -euo pipefail
+
+CONTAINER="${1:-gwoe-antragspruefer-dev}"
+N_SAMPLES="${N_SAMPLES:-5}"
+
+if ! docker ps --format '{{.Names}}' | grep -q "^${CONTAINER}$"; then
+    echo "$(date -Iseconds) SKIP — ${CONTAINER} is not running"
+    exit 0
+fi
+
+echo "$(date -Iseconds) START pm-sample-bundle (n=${N_SAMPLES})"
+
+docker exec -i -e N_SAMPLES="$N_SAMPLES" "$CONTAINER" python <<'EOF'
+import os
+import asyncio
+import sqlite3
+from app.presse_generator import generate_draft
+
+
+N_SAMPLES = int(os.environ.get("N_SAMPLES", "5"))
+
+
+def find_candidates(n: int) -> list:
+    """Top-N hochbewertete Anträge ohne existierende PM (style=pm)."""
+    conn = sqlite3.connect("/app/data/gwoe-antraege.db")
+    try:
+        cur = conn.execute("""
+            SELECT a.drucksache, a.bundesland, a.gwoe_score
+            FROM assessments a
+            WHERE a.gwoe_score >= 7.0
+            ORDER BY a.gwoe_score DESC, a.datum DESC
+        """)
+        out = []
+        for ds, bl, score in cur.fetchall():
+            existing = conn.execute(
+                "SELECT id FROM presse_drafts WHERE drucksache=? AND style='pm' LIMIT 1",
+                (ds,),
+            ).fetchone()
+            if existing:
+                continue
+            news_match = conn.execute("""
+                SELECT n.url FROM news_articles n
+                ORDER BY n.datum DESC LIMIT 1
+            """).fetchone()
+            if not news_match:
+                continue
+            out.append({"drucksache": ds, "bundesland": bl,
+                        "score": score, "news_url": news_match[0]})
+            if len(out) >= n:
+                break
+        return out
+    finally:
+        conn.close()
+
+
+async def main() -> None:
+    cands = find_candidates(N_SAMPLES)
+    if not cands:
+        print("  keine Kandidaten gefunden")
+        return
+    print(f"  {len(cands)} Kandidaten:")
+    for i, c in enumerate(cands, 1):
+        print(f"  [{i}/{len(cands)}] {c['bundesland']}/{c['drucksache']} (GWÖ {c['score']})")
+        try:
+            pm = await generate_draft(
+                drucksache=c["drucksache"], news_url=c["news_url"], style="pm",
+            )
+            print(f"     PM:     id={pm.get('id')}  was_existing={pm.get('_was_existing')}")
+            thread = await generate_draft(
+                drucksache=c["drucksache"], news_url=c["news_url"], style="thread",
+            )
+            print(f"     Thread: id={thread.get('id')}  was_existing={thread.get('_was_existing')}")
+        except Exception as e:
+            print(f"     ERROR:  {str(e)[:120]}")
+
+
+asyncio.run(main())
+EOF
+
+echo "$(date -Iseconds) END pm-sample-bundle"