feat: content_hash + дедупликация при загрузке

deploy/db/documents.py

@@ -2,12 +2,12 @@
 from .connection import query, execute, execute_returning
 
 
-def insert(filename, mime_type, original_bytes, status="uploaded", batch_id=None, zip_source=None):
+def insert(filename, mime_type, original_bytes, status="uploaded", batch_id=None, zip_source=None, content_hash=None):
     """Insert document, return row dict."""
     return execute_returning(
-        """INSERT INTO documents (filename, mime_type, original_bytes, status, batch_id, zip_source)
-           VALUES (%s, %s, %s, %s, %s, %s) RETURNING *""",
-        (filename, mime_type, original_bytes, status, batch_id, zip_source),
+        """INSERT INTO documents (filename, mime_type, original_bytes, status, batch_id, zip_source, content_hash)
+           VALUES (%s, %s, %s, %s, %s, %s, %s) RETURNING *""",
+        (filename, mime_type, original_bytes, status, batch_id, zip_source, content_hash),
     )
 
 
@@ -16,6 +16,15 @@ def get(doc_id):
     return rows[0] if rows else None
 
 
+def get_by_hash(batch_id, content_hash):
+    """Find document by content hash within batch."""
+    rows = query(
+        "SELECT id FROM documents WHERE batch_id = %s AND content_hash = %s LIMIT 1",
+        (batch_id, content_hash),
+    )
+    return rows[0] if rows else None
+
+
 def set_parsed(doc_id, elements_json):
     """Update elements_json + status='parsed'."""
     import json