Update app.py

app.py

@@ -23,7 +23,9 @@ def _list_collection_files(pr_number: str) -> Tuple[str, ...]:
     Return the `failure_summary.json` paths stored for a specific PR.
     """
     prefix = f"pr-{pr_number}"
-    print(f"Looking for files with prefix: {prefix}")
+    print(f"DEBUG: Looking for files with prefix: {prefix}")
+    print(f"DEBUG: Dataset ID: {DATASET_ID}")
+    
     try:
         # List all files in the repo and filter by prefix
         entries = API.list_repo_tree(
@@ -32,25 +34,56 @@ def _list_collection_files(pr_number: str) -> Tuple[str, ...]:
             revision="main",
             recursive=True,
         )
+        print("DEBUG: Successfully called list_repo_tree")
     except HfHubHTTPError as error:
-        print(f"Failed to list repo tree: {error}")
+        print(f"ERROR: Failed to list repo tree: {error}")
+        return tuple()
+    except Exception as error:
+        print(f"ERROR: Unexpected error in list_repo_tree: {error}")
+        import traceback
+        traceback.print_exc()
         return tuple()
 
     files = []
     matching_paths = []
-    for entry in entries:
-        entry_type = getattr(entry, "type", None)
-        # Debug: show all files that match the prefix
-        if entry_type == "file" and entry.path.startswith(prefix):
-            matching_paths.append(entry.path)
-        # Filter by prefix and look for failure_summary.json files
-        if entry_type == "file" and entry.path.startswith(prefix) and entry.path.endswith("failure_summary.json"):
-            files.append(entry.path)
+    all_entries = []
+    
+    try:
+        for entry in entries:
+            all_entries.append(entry)
+            entry_type = getattr(entry, "type", type(entry).__name__)
+            entry_path = getattr(entry, "path", str(entry))
+            
+            # Debug: show all entries
+            if len(all_entries) <= 10:
+                print(f"DEBUG: Entry {len(all_entries)}: {entry_path} (type: {entry_type})")
+            
+            # Filter by prefix
+            if entry_path.startswith(prefix):
+                matching_paths.append(entry_path)
+                
+            # Look for failure_summary.json files
+            if entry_path.startswith(prefix) and entry_path.endswith("failure_summary.json"):
+                if "file" in entry_type.lower() or entry_type == "RepoFile":
+                    files.append(entry_path)
+                    print(f"DEBUG: Found matching file: {entry_path}")
+        
+        print(f"DEBUG: Total entries processed: {len(all_entries)}")
+        print(f"DEBUG: Entries with prefix '{prefix}': {len(matching_paths)}")
+        print(f"DEBUG: failure_summary.json files found: {len(files)}")
+        
+        if matching_paths and len(files) == 0:
+            print(f"DEBUG: Sample matching paths (first 5): {matching_paths[:5]}")
+            # Check if we're only getting folders
+            folder_count = sum(1 for p in matching_paths if "RepoFolder" in str(type(p)))
+            print(f"DEBUG: Folders in matching paths: {folder_count}")
+            
+    except Exception as error:
+        print(f"ERROR: Error processing entries: {error}")
+        import traceback
+        traceback.print_exc()
+        return tuple()
     
-    print(f"Found {len(matching_paths)} files with prefix {prefix}")
-    print(f"Found {len(files)} failure_summary.json files")
-    if matching_paths:
-        print(f"Sample matching paths: {matching_paths[:5]}")
     return tuple(files)
 
 
@@ -117,11 +150,15 @@ def query(repo: str, pr: str, sha: str) -> Tuple[List[List[str]], str, str]:
     repo = repo.strip()
     pr = pr.strip()
     sha = sha.strip()
+    
+    print(f"DEBUG: Query called with repo='{repo}', pr='{pr}', sha='{sha}'")
 
     if not pr:
         return [], json.dumps({"error": "PR number is required."}, indent=2), "Provide a PR number to search."
 
     records = _filter_records(repo, pr, sha)
+    print(f"DEBUG: _filter_records returned {len(records)} records")
+    
     if not records:
         return [], json.dumps({"error": "No records found."}, indent=2), f"No records found for PR {pr}."
 
@@ -142,6 +179,7 @@ def query(repo: str, pr: str, sha: str) -> Tuple[List[List[str]], str, str]:
 
     latest_payload = json.dumps(records[0], indent=2)
     status = f"Showing {len(records)} record(s) for PR {pr}."
+    print(f"DEBUG: Returning {len(table_rows)} table rows")
     return table_rows, latest_payload, status