Preserve ingestion data across rebuilds

backend/app/db.py

@@ -245,6 +245,57 @@ def clear_library_documents(library_id: str) -> Dict[str, Any]:
         conn.close()
 
 
+def replace_library_documents(
+    library_id: str,
+    chunks: List[Dict[str, Any]],
+) -> Dict[str, Any]:
+    """Atomically replace all document chunks for a library."""
+    conn = get_connection()
+
+    try:
+        now = datetime.utcnow().isoformat()
+        conn.execute("BEGIN")
+        cursor = conn.execute(
+            "DELETE FROM documents WHERE library_id = ?", (library_id,)
+        )
+        deleted = cursor.rowcount
+
+        conn.executemany(
+            """
+            INSERT INTO documents
+                (id, library_id, path, title, content, chunk_index,
+                 token_estimate, created_at)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """,
+            [
+                (
+                    chunk["id"],
+                    library_id,
+                    chunk["path"],
+                    chunk.get("title"),
+                    chunk.get("content"),
+                    chunk.get("chunk_index"),
+                    chunk.get("token_estimate", 0),
+                    now,
+                )
+                for chunk in chunks
+            ],
+        )
+
+        conn.commit()
+        return {
+            "success": True,
+            "deleted": deleted,
+            "inserted": len(chunks),
+            "library_id": library_id,
+        }
+    except Exception as e:
+        conn.rollback()
+        return {"success": False, "error": str(e)}
+    finally:
+        conn.close()
+
+
 def delete_library(library_id: str) -> Dict[str, Any]:
     """Delete a library row and its document chunks."""
     conn = get_connection()