Protokoll #5414

{
    "event": "PreToolUse",
    "tool_name": "Read",
    "tool_input": {
        "file_path": "\/opt\/scripts\/pipeline\/quality_test.py",
        "offset": 120,
        "limit": 100
    }
}

{
    "tool_response": {
        "type": "text",
        "file": {
            "filePath": "\/opt\/scripts\/pipeline\/quality_test.py",
            "content": "        try:\n            import re\n            json_match = re.search(r'\\{[\\s\\S]*\\}', result[\"response\"])\n            if json_match:\n                data = json.loads(json_match.group())\n                entities = data.get(\"entities\", [])\n        except:\n            pass\n\n    result[\"entities\"] = entities\n    result[\"entity_count\"] = len(entities)\n    return result\n\n\ndef classify_taxonomy(text, model_name, model_id, client=None):\n    \"\"\"Classify text into taxonomy categories.\"\"\"\n    prompt = f\"\"\"Klassifiziere den folgenden Text in passende Kategorien.\n\nWähle aus diesen Hauptkategorien:\n- Methoden (Therapiemethoden, Coaching-Techniken)\n- Theorie (Konzepte, Modelle, Grundlagen)\n- Praxis (Anwendung, Fallbeispiele, Übungen)\n- Organisation (Strukturen, Prozesse, Rollen)\n- Kommunikation (Gesprächsführung, Interaktion)\n- Entwicklung (Persönliche Entwicklung, Veränderung)\n- Teamarbeit (Teamdynamik, Zusammenarbeit)\n\nAntworte NUR im JSON-Format:\n{{\"categories\": [\"...\", \"...\"], \"confidence\": 0.0-1.0, \"reasoning\": \"kurze Begründung\"}}\n\nText:\n{text[:2500]}\n\"\"\"\n\n    if model_name == \"anthropic\":\n        result = run_anthropic(client, prompt, model_id)\n    else:\n        result = run_ollama(model_id, prompt)\n\n    # Parse JSON\n    categories = []\n    confidence = 0\n    reasoning = \"\"\n    if result[\"success\"]:\n        try:\n            import re\n            json_match = re.search(r'\\{[\\s\\S]*\\}', result[\"response\"])\n            if json_match:\n                data = json.loads(json_match.group())\n                categories = data.get(\"categories\", [])\n                confidence = data.get(\"confidence\", 0)\n                reasoning = data.get(\"reasoning\", \"\")\n        except:\n            pass\n\n    result[\"categories\"] = categories\n    result[\"confidence\"] = confidence\n    result[\"reasoning\"] = reasoning\n    return result\n\n\ndef generate_questions(text, model_name, model_id, client=None):\n    \"\"\"Generate quiz questions from text.\"\"\"\n    prompt = f\"\"\"Erstelle 3 Verständnisfragen zu folgendem Lerntext.\nDie Fragen sollen das Verständnis der Kernkonzepte prüfen.\n\nAntworte NUR im JSON-Format:\n{{\"questions\": [\n  {{\"question\": \"...\", \"answer\": \"...\", \"difficulty\": \"leicht|mittel|schwer\"}}\n]}}\n\nText:\n{text[:2500]}\n\"\"\"\n\n    if model_name == \"anthropic\":\n        result = run_anthropic(client, prompt, model_id)\n    else:\n        result = run_ollama(model_id, prompt)\n\n    # Parse JSON\n    questions = []\n    if result[\"success\"]:\n        try:\n            import re\n            json_match = re.search(r'\\{[\\s\\S]*\\}', result[\"response\"])\n            if json_match:\n                data = json.loads(json_match.group())\n                questions = data.get(\"questions\", [])\n        except:\n            pass\n\n    result[\"questions\"] = questions\n    result[\"question_count\"] = len(questions)\n    return result\n\n\ndef run_quality_test(document_id):\n    \"\"\"Run full quality comparison test.\"\"\"\n    db.connect()",
            "numLines": 100,
            "startLine": 120,
            "totalLines": 349
        }
    }
}