feat: трёхэтапный фильтр мусора — этапы 1-2 (0 токенов) в classify.py

2026-06-27 13:16:30 +04:00
parent a474b4324b
commit 909468b5b7
2 changed files with 45 additions and 1 deletions
@@ -53,6 +53,14 @@ def set_classify_failed(doc_id, error):
    )


+def set_classify_garbage(doc_id, reason=""):
+    """Mark document as garbage (Stage 1-2 filter, no LLM call)."""
+    return execute(
+        "UPDATE documents SET doc_type='garbage', classify_status='garbage', error_message=%s WHERE id=%s",
+        (f"garbage: {reason}", doc_id),
+    )
+
+
 def list_pending(batch_id):
    """Documents waiting for classification."""
    return query(