Fixes #516: replace bare except with Exception and fix file handle leaks

dataflow/operators/code/eval/python_executor.py

@@ -469,7 +469,7 @@ def _ensure_worker(self):
             # Worker process died, restart it
             try:
                 self.persistent_worker.terminate()
-            except:
+            except Exception:
                 pass
             self.persistent_worker = PersistentWorker()
 
@@ -478,7 +478,7 @@ def _restart_worker(self):
         if self.persistent_worker is not None:
             try:
                 self.persistent_worker.terminate()
-            except:
+            except Exception:
                 pass
         self.persistent_worker = PersistentWorker()
 
@@ -625,7 +625,7 @@ def batch_apply(self, batch_code, messages):
                         # If reset fails, try restarting worker
                         try:
                             self._restart_worker()
-                        except:
+                        except Exception:
                             pass
             except Exception as error:
                 print(f"Error in batch_apply: {error}")

dataflow/operators/knowledge_cleaning/generate/mathbook_question_extract.py

@@ -144,7 +144,8 @@ def organize_pics(
         global_json_data = []
 
         # read mineru content json
-        json_data = json.load(open(mineru_content_json_path, 'r'))
+        with open(mineru_content_json_path, 'r') as f:
+            json_data = json.load(f)
 
         # if output_pic_folder is not exist, create it
         if not os.path.exists(output_pic_folder):
@@ -207,7 +208,8 @@ def process_input(self,
         max_page_idx = max(idx_list)
 
         # load img_json
-        img_json = json.load(open(img_json_path, "r"))
+        with open(img_json_path, "r") as f:
+            img_json = json.load(f)
         img_dict = {}
         for item in img_json:
             if item["page_idx"] not in img_dict:

dataflow/operators/text_pt/eval/Qurating/modeling/modeling_flash_llama.py

@@ -531,7 +531,7 @@ def forward(
         ):
             try: # for flash-attn latest version
                 hidden_states, unpad_indices, cu_seqlens, max_seqlen, _ = unpad_input(hidden_states, attention_mask)
-            except: # for flash-attn 2.3.3 verstion
+            except Exception:  # for flash-attn 2.3.3 verstion
                 hidden_states, unpad_indices, cu_seqlens, max_seqlen = unpad_input(hidden_states, attention_mask)
             unpadded_lengths = (cu_seqlens, max_seqlen)
         else:

dataflow/utils/pdf2vqa/format_utils.py

@@ -9,7 +9,7 @@ def refine_title(title: str, strict_title_match=False):
         try:
             # 优先提取阿拉伯数字章节编号（如1.1，2等）
             new_title = re.search(r"\d+\.\d+|\d+", title).group()
-        except:    
+        except AttributeError:
             try:
                 # 其次提取中文数字章节编号（如六、二十四等）
                 new_title = re.search(r'[一二三四五六七八九零十百]+', title).group()