update app

app.py

@@ -45,7 +45,7 @@ def unsafe_execute(prompt, completion, suffix, test_case, timeout=3):
     except Exception as e:
         # 捕获异常并记录堆栈跟踪
         output = ''
-        error_output = e
+        error_output = str(e)
     finally:
         # 恢复标准输出和标准错误
         sys.stdout = old_stdout
@@ -57,6 +57,11 @@ def unsafe_execute(prompt, completion, suffix, test_case, timeout=3):
     elif not output and not error_output:
         return 'Pass all test cases!'
     else:
+        error_lines = error_output.splitlines()
+        if error_lines:
+            return f'Error: {error_lines[-1]}'
+        else:
+            return 'Error: Unknown error'
         return f'Error: {error_output}'
 def read_problems() -> Dict[str, Dict]:
     benchmark_file = "HumanEval-SingleLineInfilling.jsonl.gz"
@@ -428,6 +433,8 @@ def get_example_input():
     problem = problems[task_id]
     prefix, suffix = problem['prompt'], problem['suffix']
     test_case = problem['test']
+    pattern = r'METADATA\s*=\s*\{.*?\}\n\n'
+    cleaned_code = re.sub(pattern, '', code, flags=re.DOTALL)
     test_case = test_case.replace('def check(candidate):', 'def run_test():')
     test_case = test_case.replace('candidate', problem['entry_point'])
     test_case = test_case + '\n\n\nrun_test()'