Upload app.py

app.py

@@ -30,6 +30,7 @@ folder_tree_data: dict[str, list[dict]] = {}
 folder_browser_data: dict[str, dict[str, dict]] = {}
 source_records_map: dict[str, list[int]] = {}
 extension_counts: dict[str, int] = {}
+source_extension_counts: dict[str, dict[str, int]] = {}
 word_index: dict[str, set[int]] = {}
 
 def query_tokens(text: str) -> list[str]:
@@ -156,13 +157,18 @@ def load_json_gz(path: Path):
     return json.loads(gzip.decompress(path.read_bytes()).decode("utf-8"))
 
 def build_indexes() -> None:
-    global word_index, source_records_map, extension_counts
+    global word_index, source_records_map, extension_counts, source_extension_counts
     word_index = {}
     source_records_map = {}
     extension_counts = {}
+    source_extension_counts = {}
     for idx, rec in enumerate(records):
-        source_records_map.setdefault(rec["source"], []).append(idx)
-        extension_counts[rec["extension"]] = extension_counts.get(rec["extension"], 0) + 1
+        source = rec["source"]
+        extension = rec["extension"]
+        source_records_map.setdefault(source, []).append(idx)
+        extension_counts[extension] = extension_counts.get(extension, 0) + 1
+        source_counts_for_ext = source_extension_counts.setdefault(source, {})
+        source_counts_for_ext[extension] = source_counts_for_ext.get(extension, 0) + 1
         file_text = " ".join([rec["display_name"], rec["source_name"]])
         path_text = rec["display_rel_path"]
         tokens = tokenize(f"{file_text} {path_text}")
@@ -270,7 +276,7 @@ def search(q="", sources_filter=None, folders=None, min_size=None, max_size=None
     if not q and sort == "relevance" and not has_filters(sources_filter, folders, min_size, max_size):
         result_items = [trim_record(rec) for rec in records[start:start + page_size]]
         result_items = add_summaries(result_items)
-        return {"results": result_items, "total": len(records), "page": page, "page_size": page_size, "did_you_mean": None}
+        return {"results": result_items, "total": len(records), "page": page, "page_size": page_size}
     if not q:
         indices = list(range(len(records)))
         filtered = apply_filters(indices, sources_filter, folders, min_size, max_size)
@@ -300,7 +306,7 @@ def search(q="", sources_filter=None, folders=None, min_size=None, max_size=None
     total = len(filtered)
     result_items = [trim_record(records[idx]) for idx in filtered[start:start + page_size]]
     result_items = add_summaries(result_items)
-    return {"results": result_items, "total": total, "page": page, "page_size": page_size, "did_you_mean": None}
+    return {"results": result_items, "total": total, "page": page, "page_size": page_size}
 
 def fulltext_search(q="", sources_filter=None, folders=None, min_size=None, max_size=None, page=1, page_size=100, sort="relevance", exact=False, search_paths=True):
     q = q.strip()
@@ -327,7 +333,7 @@ def fulltext_search(q="", sources_filter=None, folders=None, min_size=None, max_
         q,
         matched_snippets=True,
     )
-    return {"results": result_items, "total": total, "page": page, "page_size": page_size, "did_you_mean": None}
+    return {"results": result_items, "total": total, "page": page, "page_size": page_size}
 
 def get_doc_storage_path(doc: dict) -> Path:
     return BASE_DIR / doc["storage_root"] / doc["storage_rel_path"]
@@ -391,10 +397,7 @@ def api_sources():
 @app.get("/api/extensions")
 
 def api_extensions(source: Optional[str] = Query(default=None)):
-    pool = [records[idx] for idx in source_records_map.get(source, [])] if source else records
-    counts = {}
-    for rec in pool:
-        counts[rec["extension"]] = counts.get(rec["extension"], 0) + 1
+    counts = source_extension_counts.get(source, {}) if source else extension_counts
     return JSONResponse([{"name": ext, "count": count} for ext, count in sorted(counts.items())])
 @app.get("/api/folders/{source_slug}")
 
@@ -407,10 +410,14 @@ def api_folder_contents(source_slug: str, path: str = Query(default="")):
 @app.get("/api/random")
 
 def api_random(source: Optional[str] = Query(default=None)):
-    pool = [records[idx] for idx in source_records_map.get(source, [])] if source else records
-    if not pool:
+    if source:
+        indices = source_records_map.get(source, [])
+        if not indices:
+            return JSONResponse({"error": "no record"}, status_code=404)
+        return JSONResponse(trim_record(records[random.choice(indices)]))
+    if not records:
         return JSONResponse({"error": "no record"}, status_code=404)
-    return JSONResponse(trim_record(random.choice(pool)))
+    return JSONResponse(trim_record(random.choice(records)))
 @app.get("/api/preview/{doc_id}")
 
 def api_preview(doc_id: str):