code update

src/leaderboard/read_evals.py

@@ -153,28 +153,95 @@ def get_request_file_for_model(requests_path, model_name, precision):
     return request_file
 
 
+# def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResult]:
+#     """From the path of the results folder root, extract all needed info for results"""
+#     model_result_filepaths = []
+
+#     print(f"DEBUG: Results path: {results_path}")
+#     files = os.listdir(results_path)
+#     for f in files:
+#         print(f)
+
+#     for root, _, files in os.walk(results_path):
+#         # We should only have json files in model results
+#         if len(files) == 0 or any([not f.endswith(".json") for f in files]):
+#             continue
+
+#         # Sort the files by date
+#         try:
+#             files.sort(key=lambda x: x.removesuffix(".json").removeprefix("results_")[:-7])
+#         except dateutil.parser._parser.ParserError:
+#             files = [files[-1]]
+
+#         for file in files:
+#             model_result_filepaths.append(os.path.join(root, file))
+
+#     eval_results = {}
+#     for model_result_filepath in model_result_filepaths:
+#         # Creation of result
+#         eval_result = EvalResult.init_from_json_file(model_result_filepath)
+#         eval_result.update_with_request_file(requests_path)
+
+#         # Store results of same eval together
+#         eval_name = eval_result.eval_name
+#         if eval_name in eval_results.keys():
+#             eval_results[eval_name].results.update({k: v for k, v in eval_result.results.items() if v is not None})
+#         else:
+#             eval_results[eval_name] = eval_result
+
+#     results = []
+#     for v in eval_results.values():
+#         try:
+#             v.to_dict() # we test if the dict version is complete
+#             results.append(v)
+#         except KeyError:  # not all eval values present
+#             continue
+
+#     return results
+
 def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResult]:
     """From the path of the results folder root, extract all needed info for results"""
     model_result_filepaths = []
 
     print(f"DEBUG: Results path: {results_path}")
-    files = os.listdir(results_path)
-    for f in files:
-        print(f)
-
-    for root, _, files in os.walk(results_path):
-        # We should only have json files in model results
-        if len(files) == 0 or any([not f.endswith(".json") for f in files]):
+    try:
+        files = os.listdir(results_path)
+        for f in files:
+            print(f)
+    except FileNotFoundError:
+        print(f"Error: Directory not found at {results_path}")
+        return [] # Exit early if path doesn't exist
+
+
+    
+    # Change signature from (root, _, files) to (root, dirs, files)
+    # This allows us to modify 'dirs' to stop unwanted recursion.
+    for root, dirs, files in os.walk(results_path):
+        
+        # 1. Prevent recursion into hidden directories (like .cache, .git)
+        dirs[:] = [d for d in dirs if not d.startswith('.')]
+
+        # 2. Filter the 'files' list to only include .json files
+        json_files = [f for f in files if f.endswith(".json")]
+
+        # 3. If no .json files were found in this directory, skip it
+        if not json_files:
             continue
 
-        # Sort the files by date
+        # 4. Now, apply the sorting logic ONLY to the filtered json_files
         try:
-            files.sort(key=lambda x: x.removesuffix(".json").removeprefix("results_")[:-7])
-        except dateutil.parser._parser.ParserError:
-            files = [files[-1]]
-
-        for file in files:
+            json_files.sort(key=lambda x: x.removesuffix(".json").removeprefix("results_")[:-7])
+        except (dateutil.parser._parser.ParserError, IndexError): # Added IndexError for safety
+            # Original logic: if parsing fails, just use the last file.
+            if json_files:
+                json_files = [json_files[-1]]
+            else:
+                continue # Should not happen, but good to be safe
+
+        # 5. Add the full paths of the sorted json files to our list
+        for file in json_files:
             model_result_filepaths.append(os.path.join(root, file))
+            
 
     eval_results = {}
     for model_result_filepath in model_result_filepaths:
@@ -195,6 +262,7 @@ def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResu
             v.to_dict() # we test if the dict version is complete
             results.append(v)
         except KeyError:  # not all eval values present
+            print(f"Warning: Skipping {v.eval_name} due to missing eval values.")
             continue
 
     return results