中国证券报·中证网

comData/dingzhi/zzcx.py

@@ -51,7 +51,7 @@ def zzcx():
     driver = create_driver()
     driver.maximize_window()
     url = 'https://zzcx.cs.com.cn/dist/publishManuscript/listES'
-    payload = {"pageNo": 1, "pageSize": 15, "statusList": [0], "keyword": ""}
+    # payload = {"pageNo": 1, "pageSize": 15, "statusList": [0], "keyword": ""}
     headers = {
         'Accept': 'application/json',
         'Accept-Encoding': 'gzip, deflate, br',
@@ -69,10 +69,11 @@ def zzcx():
         'Origin': 'https://zzcx.cs.com.cn',
         'Referer': 'https://zzcx.cs.com.cn/app/zzb/list?spm=0.0.0.0.wjnSUZ'
     }
-    payload = json.dumps(payload)
-    result_json = requests.post(url=url, data=payload, headers=headers).json()
-    print(result_json)
-    pages = result_json['data']['pages']
+    # payload = json.dumps(payload)
+    # result_json = requests.post(url=url, data=payload, headers=headers).json()
+    # print(result_json)
+    # pages = result_json['data']['pages']
+    pages = 5
     for page in range(1, int(pages) + 1):
         payload_page = {"pageNo": page, "pageSize": 15, "statusList": [0], "keyword": ""}
         payload_page = json.dumps(payload_page)
@@ -81,6 +82,7 @@ def zzcx():
         for news in records:
             title = news['title']
             news_url = 'https://zzcx.cs.com.cn/app/zzb/detail?id=' + news['manuscriptId']
+            # news_url = 'https://zzcx.cs.com.cn/app/zzb/detail?id=3ec65751b63e40d7813a0c6bbe9b3135'
 
             try:
                 flag = r.sismember('IN-20240129-0001', news_url)
@@ -96,6 +98,7 @@ def zzcx():
 
             div_photo = driver.find_elements(By.ID, 'line')
             for png_ in div_photo:
+                try:
                     div = png_.find_element(By.XPATH, './/div/div[1]/div')
                     # div = png_.find_element(By.CLASS_NAME, 'ant-col ant-col-17')
                     # todo:滚轮需要滑动
@@ -122,6 +125,8 @@ def zzcx():
                         "var newElement = document.createElement('img'); newElement.src = 'http://zzsn.luyuen.com" + path + "'; arguments[0].insertBefore(newElement, arguments[0].firstChild);",
                         append_tag)
                     os.remove(file_path)
+                except:
+                    continue
 
             # div_undefined_line = driver.find_elements(By.ID, 'k-line-undefined')
             div_undefined_line = driver.find_elements(By.ID, 'KLineSubscription')
@@ -235,6 +240,7 @@ def zzcx():
                 log.info(e)
             finally:
                 producer.close()
+
 if __name__ == "__main__":
     pathType = 'PhotoDingzhi/'
     r = redis.Redis(host='114.115.236.206', port=6379, password='clbzzsn', db=5)