Upload app.py

app.py

@@ -289,7 +289,7 @@ def trim_record(rec: dict) -> dict:
         "HasTxt": True,
     }
 
-def add_summaries(items: list[dict], query: str = "", matched_snippets: bool = False) -> list[dict]:
+def add_summaries(items: list[dict], query: str = "", matched_snippets: bool = False, content_indices: set[int] | None = None) -> list[dict]:
     if fulltext_databases is None or not items:
         return items
     summaries = fulltext_databases.summaries([item["doc_id"] for item in items])
@@ -298,8 +298,13 @@ def add_summaries(items: list[dict], query: str = "", matched_snippets: bool = F
         if matched_snippets and query:
             snippet = build_first_match_snippet(summary, query)
             normalized_summary = normalize_text(summary)
-            if not any(token in normalized_summary for token in query_tokens(query)):
-                rec = record_map.get(item["doc_id"])
+            rec = record_map.get(item["doc_id"])
+            rec_index = record_map_index.get(item["doc_id"])
+            if (
+                content_indices is not None
+                and rec_index in content_indices
+                and not any(token in normalized_summary for token in query_tokens(query))
+            ):
                 file_path = get_doc_storage_path(rec) if rec else None
                 if file_path and file_path.exists():
                     text = file_path.read_text(encoding="utf-8", errors="ignore")
@@ -385,9 +390,11 @@ def fulltext_search(q="", sources_filter=None, folders=None, min_size=None, max_
     matched_doc_ids = set()
     for source_slug in source_slugs:
         matched_doc_ids.update(f"{source_slug}:{doc_number}" for doc_number in fulltext_databases.search_source(source_slug, q, exact))
-    indices = {idx for doc_id in matched_doc_ids if (idx := record_map_index.get(doc_id)) is not None}
+    content_indices = {idx for doc_id in matched_doc_ids if (idx := record_map_index.get(doc_id)) is not None}
+    indices = set(content_indices)
     if exact and ("*" in q or "?" in q or len(normalize_text(q)) > 3):
-        indices = verify_fulltext_matches(indices, q)
+        content_indices = verify_fulltext_matches(content_indices, q)
+        indices = set(content_indices)
     indices.update(metadata_matches(q, exact, search_paths))
     filtered = apply_filters(indices, sources_filter, folders, min_size, max_size)
     tokens = query_tokens(q)
@@ -403,6 +410,7 @@ def fulltext_search(q="", sources_filter=None, folders=None, min_size=None, max_
         [trim_record(records[idx]) for idx in filtered[start:start + page_size]],
         q,
         matched_snippets=True,
+        content_indices=content_indices,
     )
     return {"results": result_items, "total": total, "page": page, "page_size": page_size}