雪球网年报

base/BaseCore.py

@@ -767,11 +767,15 @@ class BaseCore:
             log = self.getLogger()
             log.error(f'OBS发送失败')
             return retData
-
+        try:
             with fitz.open(stream=response.content, filetype='pdf') as doc:
                 page_size = doc.page_count
                 for page in doc.pages():
                     retData['content'] += page.get_text()
+        except:
+            log = self.getLogger()
+            log.error(f'文件损坏')
+            return retData
 
         if page_size < 1:
             # pdf解析失败

comData/annualReport/雪球网-年报.py

-# -*- coding: utf-8 -*-
+# -*- coding: utf-8 -*-
@@ -177,7 +177,7 @@ def spider_annual_report(dict_info,num):
                     pass
                 else:
                     log.info(f'====pdf解析失败====')
-                    return False
+                    continue
                 num = num + 1
                 try:
                     origin = '雪球网'
@@ -259,8 +259,8 @@ if __name__ == '__main__':
     while True:
         start_time = time.time()
         # 获取企业信息
-        social_code = baseCore.redicPullData('AnnualEnterprise:gnshqy_socialCode')
-        # social_code = '91100000100003962T'
+        # social_code = baseCore.redicPullData('AnnualEnterprise:gnshqy_socialCode')
+        social_code = '913412007050444417'
         if not social_code:
             time.sleep(20)
             continue