Fix document deletion: use point IDs instead of filter-based delete, add dropdown selector

app/core/vectorstore.py

@@ -8,6 +8,7 @@ from qdrant_client.http.models import (
     MatchAny,
     MatchValue,
     PayloadSchemaType,
+    PointIdsList,
     PointStruct,
     Range,
     VectorParams,
@@ -127,14 +128,37 @@ class VectorStoreService:
         ]
 
     def delete_document(self, document_id: str) -> int:
+        # First, find all point IDs belonging to this document
+        doc_filter = Filter(
+            must=[FieldCondition(key="document_id", match=MatchValue(value=document_id))]
+        )
+        point_ids = []
+        offset = None
+        while True:
+            results, next_offset = self.client.scroll(
+                collection_name=self.collection_name,
+                scroll_filter=doc_filter,
+                limit=100,
+                offset=offset,
+                with_payload=False,
+                with_vectors=False,
+            )
+            point_ids.extend([r.id for r in results])
+            if next_offset is None:
+                break
+            offset = next_offset
+
+        if not point_ids:
+            logger.warning(f"No points found for document '{document_id}'")
+            return 0
+
+        # Delete by point IDs (requires only write permission, not manage)
         self.client.delete(
             collection_name=self.collection_name,
-            points_selector=Filter(
-                must=[FieldCondition(key="document_id", match=MatchValue(value=document_id))]
-            ),
+            points_selector=PointIdsList(points=point_ids),
         )
-        logger.info(f"Deleted document '{document_id}' from '{self.collection_name}'")
-        return 0
+        logger.info(f"Deleted {len(point_ids)} points for document '{document_id}'")
+        return len(point_ids)
 
     def scroll_all(self, batch_size: int = 100) -> list[dict]:
         all_points = []

app/ui/gradio_app.py

@@ -123,26 +123,42 @@ def upload_document(file):
         return f"**Upload failed:** {e}"
 
 
-def list_documents():
+def _fetch_documents():
+    """Fetch documents from API. Returns list of doc dicts or empty list."""
     try:
         response = httpx.get(f"{API_BASE}/api/documents", timeout=30)
         if response.status_code == 200:
-            data = response.json()
-            docs = data.get("documents", [])
-            if not docs:
-                return [["—", "—", "—", "—"]]
-            return [
-                [
-                    d.get("source", ""),
-                    d.get("doc_type", "").upper(),
-                    str(d.get("num_chunks", 0)),
-                    d.get("document_id", "")[:12] + "...",
-                ]
-                for d in docs
-            ]
-        return [["Error loading", "", "", ""]]
-    except Exception as e:
-        return [[f"Error: {e}", "", "", ""]]
+            return response.json().get("documents", [])
+    except Exception:
+        pass
+    return []
+
+
+def list_documents():
+    docs = _fetch_documents()
+    if not docs:
+        return [["—", "—", "—", "—"]]
+    return [
+        [
+            d.get("source", ""),
+            d.get("doc_type", "").upper(),
+            str(d.get("num_chunks", 0)),
+            d.get("document_id", ""),
+        ]
+        for d in docs
+    ]
+
+
+def get_delete_choices():
+    """Build dropdown choices: 'filename (doc_id)' for each indexed document."""
+    docs = _fetch_documents()
+    if not docs:
+        return gr.update(choices=[], value=None)
+    choices = [
+        f"{d.get('source', 'unknown')}  [{d.get('document_id', '')}]"
+        for d in docs
+    ]
+    return gr.update(choices=choices, value=None)
 
 
 def get_doc_count():
@@ -159,14 +175,21 @@ def get_doc_count():
         return "Connecting..."
 
 
-def delete_document(doc_id):
-    if not doc_id or not doc_id.strip():
-        return "Enter a document ID to delete."
+def delete_document(selection):
+    if not selection:
+        return "Select a document to delete."
+    # Extract document ID from "filename  [doc_id]" format
+    if "[" in selection and selection.endswith("]"):
+        doc_id = selection.rsplit("[", 1)[1][:-1]
+    else:
+        doc_id = selection.strip()
     try:
-        response = httpx.delete(f"{API_BASE}/api/documents/{doc_id.strip()}", timeout=30)
+        response = httpx.delete(f"{API_BASE}/api/documents/{doc_id}", timeout=30)
         if response.status_code == 200:
-            return f"Document `{doc_id.strip()[:12]}...` deleted successfully."
-        return f"**Error:** {response.text}"
+            filename = selection.split("  [")[0] if "  [" in selection else doc_id[:12]
+            return f"**'{filename}'** deleted successfully."
+        detail = response.json().get("detail", response.text)
+        return f"**Error:** {detail}"
     except Exception as e:
         return f"**Delete failed:** {e}"
 
@@ -375,11 +398,6 @@ def create_gradio_app() -> gr.Blocks:
                     )
 
             upload_status = gr.Markdown()
-            upload_btn.click(
-                fn=upload_document,
-                inputs=file_upload,
-                outputs=upload_status,
-            )
 
             gr.Markdown("---")
             gr.Markdown("#### Indexed Documents")
@@ -392,14 +410,13 @@ def create_gradio_app() -> gr.Blocks:
                 elem_classes=["doc-table"],
             )
             refresh_btn = gr.Button("Refresh", size="sm")
-            refresh_btn.click(fn=list_documents, outputs=doc_table)
 
             gr.Markdown("---")
             gr.Markdown("#### Delete a Document")
             with gr.Row():
-                delete_id_input = gr.Textbox(
-                    placeholder="Paste full document ID here...",
-                    show_label=False,
+                delete_dropdown = gr.Dropdown(
+                    choices=[],
+                    label="Select document to delete",
                     scale=3,
                 )
                 delete_btn = gr.Button(
@@ -409,10 +426,34 @@ def create_gradio_app() -> gr.Blocks:
                     elem_classes=["danger-btn"],
                 )
             delete_status = gr.Markdown()
+
+            # Refresh: updates table, dropdown, and stats
+            def refresh_all():
+                return list_documents(), get_delete_choices(), get_doc_count()
+
+            refresh_btn.click(
+                fn=refresh_all,
+                outputs=[doc_table, delete_dropdown, stats_display],
+            )
+
+            # After upload: refresh table, dropdown, and stats
+            upload_btn.click(
+                fn=upload_document,
+                inputs=file_upload,
+                outputs=upload_status,
+            ).then(
+                fn=refresh_all,
+                outputs=[doc_table, delete_dropdown, stats_display],
+            )
+
+            # Delete: delete then refresh everything
             delete_btn.click(
                 fn=delete_document,
-                inputs=delete_id_input,
+                inputs=delete_dropdown,
                 outputs=delete_status,
+            ).then(
+                fn=refresh_all,
+                outputs=[doc_table, delete_dropdown, stats_display],
             )
 
         # Footer