Merge remote-tracking branch 'origin/master'

comData/annualReport_ZJH/证监会-年报.py

-
+
@@ -49,6 +49,7 @@ def RequestUrl(url, payload, item_id, start_time):
     if response.status_code == 200:
         # 请求成功，处理响应数据
         # print(response.text)
+        soup = BeautifulSoup(response.text, 'html.parser')
         pass
     else:
         # 请求失败，输出错误信息
@@ -56,9 +57,7 @@ def RequestUrl(url, payload, item_id, start_time):
         state = 0
         takeTime = baseCore.getTimeCost(start_time, time.time())
         baseCore.recordLog(item_id, taskType, state, takeTime, url, '请求失败')
-
-    soup = BeautifulSoup(response.text, 'html.parser')
-
+        soup = ''
     return soup
 
 
@@ -89,6 +88,8 @@ def SpiderByZJH(url, payload, dic_info, num, start_time):
     # years = dic_info['call_year']
     short_name = dic_info[4]
     soup = RequestUrl(url, payload, item_id, start_time)
+    if soup == '':
+        return
     # 先获取页数
 
     page = soup.find('div', class_='pages').find('ul', class_='g-ul').text
@@ -106,8 +107,9 @@ def SpiderByZJH(url, payload, dic_info, num, start_time):
         else:
             # http://eid.csrc.gov.cn/101811/index_3_f.html
             href = url.split('index')[0] + f'index_{i}_f.html'
-
         soup = RequestUrl(href, payload, item_id, start_time)
+        if soup == '':
+            continue
         tr_list = soup.find('div', id='txt').find_all('tr')
         for tr in tr_list[1:]:
             td_list = tr.find_all('td')
@@ -210,7 +212,7 @@ def SpiderByZJH(url, payload, dic_info, num, start_time):
                         baseCore.recordLog(item_id, taskType, state, takeTime, pdf_url, e)
                         continue
             else:
-                continue
+                    continue
 
 
 def getUrl(code, url_parms, Catagory2_parms):