fix code

tasks/zhihu/content_refresh.py

@@ -23,6 +23,8 @@ class RefreshContent(object):
         '''
             初始化数据库，调整js规则
         '''
+        self.update_error_content_id = []
+        
         self.conn = pymysql.connect(host=HOST, port=PORT, user=USER,
                                     passwd=PASSWD,
                                     db=DB, charset='utf8')
@@ -48,26 +50,20 @@ class RefreshContent(object):
             item.replace_with(new_rich_obj)
         return rich_obj.decode()
     
-    def create_new_content(self, content, pic_dict):
+    def create_new_content(self, content_id, content, pic_dict):
         content = self.replace_html_image_to_url(content)
         rich_obj = BeautifulSoup(content, features="html.parser")
         for item in rich_obj.find_all("img"):
             url = item.get("src")
             new_url = pic_dict.get(url)
+            if not new_url:
+                self.update_error_content_id.append({content_id: url})
+                print({content_id: url})
+                continue
             item['src'] = new_url + '-w'
 
         return rich_obj.decode()
     
-    @staticmethod
-    def replace_url_to_new_url(content, url_dict):
-        rich_obj = BeautifulSoup(content, features="html.parser")
-        for item in rich_obj.find_all("img"):
-            url = item.get("src")
-            new_url = url_dict.get(url)
-            item['src'] = new_url
-            
-        return rich_obj.decode()
-    
     def get_all_content_ids(self, table, key_id):
         sql = """select distinct {} from {}""".format(key_id, table)
         self.cur.execute(sql)
@@ -97,9 +93,7 @@ class RefreshContent(object):
             self.conn.commit()
             content = res[0][0]
             
-            new_content = self.create_new_content(content, pic_dict)
-            print(new_content)
-            import pdb; pdb.set_trace()
+            new_content = self.create_new_content(content_id, content, pic_dict)
 
             sql = """update {} set new_content = '{}' WHERE {} = '{}' """.format(table, new_content, key_id, content_id)
             self.cur.execute(sql)
@@ -120,4 +114,6 @@ if __name__ == '__main__':
         refresh.refresh_content('zhihu_article', 'zhihu_article_picture_url', 'article_id')
     elif mark == 2:
         refresh.refresh_content('zhihu_thought', 'zhihu_thought_picture_url', 'thought_id')
+        
+    print(refresh.update_error_content_id)
     print(datetime.now())