Protokoll #17808

{
    "event": "PreToolUse",
    "tool_name": "Bash",
    "tool_input": {
        "command": "cat << 'PYTHON' | DB_PASSWORD=\"v8mFLmkF2fth_r\" timeout 1800 .\/venv\/bin\/python\nimport sys\nimport time\nsys.path.insert(0, \"\/var\/www\/scripts\/pipeline\")\n\nfrom db import db\ndb.connect()\n\nfrom analyzers.entity_extractor import extract_entities_ollama, store_entities\n\ndoc_id = 1\n\n# Get all chunks\ncursor = db.execute(\"SELECT id, content FROM chunks WHERE document_id = %s ORDER BY chunk_index\", (doc_id,))\nchunks = cursor.fetchall()\ncursor.close()\n\nprint(f\"Processing {len(chunks)} chunks with 2-pass entity extraction...\")\nprint(\"=\" * 60)\n\ntotal_entities = 0\nall_entities = []\nstart_time = time.time()\n\nfor i, chunk in enumerate(chunks, 1):\n    chunk_id = chunk['id']\n    content = chunk['content']\n    \n    # Extract entities using 2-pass\n    entities = extract_entities_ollama(content)\n    \n    if entities:\n        # Store entities\n        stored = store_entities(doc_id, entities)\n        total_entities += stored\n        \n        # Link entities to chunk\n        for entity in entities:\n            # Find entity ID\n            cursor = db.execute(\n                \"SELECT id FROM entities WHERE name = %s AND type = %s LIMIT 1\",\n                (entity['name'], entity['type'])\n            )\n            result = cursor.fetchone()\n            cursor.close()\n            \n            if result:\n                try:\n                    cursor = db.execute(\n                        \"INSERT IGNORE INTO chunk_entities (chunk_id, entity_id, relevance) VALUES (%s, %s, 0.8)\",\n                        (chunk_id, result['id'])\n                    )\n                    db.commit()\n                    cursor.close()\n                except:\n                    pass\n        \n        all_entities.extend(entities)\n    \n    if i % 10 == 0 or i == len(chunks):\n        elapsed = time.time() - start_time\n        print(f\"[{i:3d}\/{len(chunks)}] {total_entities} entities | {elapsed:.0f}s\")\n\nelapsed = time.time() - start_time\nprint(\"=\" * 60)\nprint(f\"DONE: {total_entities} entities in {elapsed:.0f}s ({elapsed\/len(chunks):.1f}s\/chunk)\")\n\n# Stats by type\ncursor = db.execute(\"SELECT type, COUNT(*) as cnt FROM entities GROUP BY type ORDER BY cnt DESC\")\ntypes = cursor.fetchall()\ncursor.close()\n\nprint(\"\\nEntities by type:\")\nfor t in types:\n    print(f\"  {t['type']:12s}: {t['cnt']}\")\nPYTHON",
        "timeout": 1800000,
        "description": "Run 2-pass entity extraction on all chunks"
    }
}

-