dongzhaorui il y a 3 ans
Parent
commit
d37b13adfa
1 fichiers modifiés avec 5 ajouts et 6 suppressions
  1. 5 6
      codes_hospital/crawl_hospital.py

+ 5 - 6
codes_hospital/crawl_hospital.py

@@ -142,7 +142,6 @@ def get_jgdm(query, proxies):
         timeout=60,
         proxies=proxies
     )
-    print('proxies >>> ', proxies)
     try:
         response = requests.post(url, **request_params)
     except requests.exceptions.ProxyError:
@@ -157,7 +156,7 @@ def get_jgdm(query, proxies):
         if item['jyzt'] != '注销':
             # results.append({'query': query, 'encJgdm': item['encJgdm']})
             results.append(item['encJgdm'])
-    logger.info(f"[列表查询成功]{proxies} - {query}")
+    logger.info(f"[列表查询成功]{query}")
     return results
 
 
@@ -191,7 +190,6 @@ def get_hospital(query, jgdm, proxies):
         verify=False,
         timeout=60
     )
-    print("proxies >>> ", proxies)
     try:
         response = requests.post(url, **request_params)
     except requests.exceptions.ProxyError:
@@ -224,7 +222,7 @@ def get_hospital(query, jgdm, proxies):
     }
     f_hospital_codes.insert_one(hospital)
     # print(json.dumps(hospital, indent=4, ensure_ascii=False))
-    logger.info(f'[详情查询成功]{proxies} - {hospital["hospital_name"]}')
+    logger.info(f'[详情查询成功]{hospital["hospital_name"]}')
     return hospital
 
 
@@ -233,10 +231,11 @@ def callback_requests(func, *args, **kwargs):
     while True:
         try:
             proxies = proxy.proxies if proxy is not None else None
+            logger.debug(f"[当前代理]:{proxies}")
             kwargs.setdefault('proxies', proxies)
             return func(*args, **kwargs)
-        except (requests.RequestException, AssertionError) as e:
-            logger.error(e)
+        except (IOError, AssertionError) as e:
+            logger.error(f"[访问异常]:{e}")
             time.sleep(3)
             if proxy is not None:
                 proxy.switch()