логирование ошибок транскрибации для битых файлов

apiApp/routers/audio_management_router.py

@@ -13,7 +13,7 @@ from datetime import datetime
 from apiApp.database import get_db
 from apiApp.database.Audio import Audio
 from apiApp.database.AiConclusion import AiConclusion
-from apiApp.config import AUDIOFILES_PATH
+from apiApp.config import AUDIOFILES_PATH, MAX_RECOGNITION_ATTEMPTS
 
 logger = logging.getLogger(__name__)
 audio_management_router = APIRouter()
@@ -32,10 +32,14 @@ def query_audio_without_conclusion(db, limit=None):
         AiConclusion.audio_id == Audio.id
     )
 
+    # Берём только те, которые еще можно/нужно распознавать
     query = db.query(Audio).filter(
         ~subquery
     ).filter(
         Audio.sourse == "internal"
+    ).filter(
+        (Audio.recognition_status.in_(["pending", "processing"])) |
+        ((Audio.recognition_status == "failed") & (Audio.recognition_attempts < MAX_RECOGNITION_ATTEMPTS))
     ).order_by(Audio.index_date.asc())
 
     if limit:
@@ -112,6 +116,10 @@ async def register_audio_file(
         audio.filename = filename
         audio.file_size = file_size
         audio.index_date = datetime.utcnow()
+        audio.recognition_status = "pending"
+        audio.recognition_attempts = 0
+        audio.recognition_last_error = None
+        audio.recognition_last_attempt_at = None
 
         db.add(audio)
         db.commit()
@@ -150,6 +158,11 @@ def process_audio_file(audio_id: str, db: Session):
             return
 
         logger.info(f"🎵 Запуск распознавания для {audio.filename}")
+        audio.recognition_status = "processing"
+        audio.recognition_attempts = (audio.recognition_attempts or 0) + 1
+        audio.recognition_last_attempt_at = datetime.utcnow()
+        audio.recognition_last_error = None
+        db.commit()
 
         # Проверяем что файл существует на диске
         from apiApp.config import AUDIOFILES_PATH
@@ -158,7 +171,9 @@ def process_audio_file(audio_id: str, db: Session):
 
         if not os.path.exists(file_path):
             logger.error(f"❌ Файл не найден на диске в FileAudioAPI: {file_path}")
-            # Помечаем audio как проблемный
+            audio.recognition_status = "failed"
+            audio.recognition_last_error = f"File not found on disk: {file_path}"
+            db.commit()
             return
 
         file_size = os.path.getsize(file_path)
@@ -187,11 +202,20 @@ def process_audio_file(audio_id: str, db: Session):
                 error_detail = response.text
                 logger.error(f"❌ Ошибка запуска распознавания для {audio.filename}: {response.status_code}")
                 logger.error(f"   Detail: {error_detail}")
+                audio.recognition_status = "failed"
+                audio.recognition_last_error = f"GigaAM start failed: {response.status_code} {error_detail}"
+                db.commit()
 
         except requests.exceptions.Timeout:
             logger.error(f"❌ Таймаут при отправке задачи для {audio.filename}")
+            audio.recognition_status = "failed"
+            audio.recognition_last_error = "Timeout when starting recognition in GigaAM"
+            db.commit()
         except requests.exceptions.ConnectionError as e:
             logger.error(f"❌ Ошибка подключения к GigaAM API для {audio.filename}: {e}")
+            audio.recognition_status = "failed"
+            audio.recognition_last_error = f"Connection error when starting recognition in GigaAM: {e}"
+            db.commit()
 
     except Exception as e:
         logger.error(f"❌ Ошибка при обработке {audio_id}: {e}")