modify content refresh

tasks/zhihu/content_refresh.py

@@ -39,27 +39,41 @@ class RefreshContent(object):
             替换url，更新回答内容
         '''
         import re
-        sql = """select {}, url, new_url from {}""".format(key_id, pic_table)
+        sql = """select {}, url, new_url from {} """.format(key_id, pic_table)
         self.cur.execute(sql)
         tuple = self.cur.fetchall()
         self.conn.commit()
         for i in range(len(tuple)):
-            if tuple[i][2] == None:
+            find_id, url, new_url = tuple[i][0], tuple[i][1], tuple[i][2]
+            if new_url is None:
                 continue
-            find_id = tuple[i][0]
-            temp = str(tuple[i][1])
+
+            print('deal : ', find_id, new_url, datetime.now())
+            temp = str(url)
             temp1 = temp.replace("?", "#")
+            
             sql = """select new_content from {} where {} = '{}' """.format(table, key_id, find_id)
             self.cur.execute(sql)
             tuples = self.cur.fetchall()
             # tuples = str(tuples)
             content = tuples[0][0]
-            pattern = r'%s(.+?)%s' % ("<noscript><img src=\"" + temp1, "</figure>")
+            # pattern = r'%s(.+?)%s' % ("<noscript><img src=\"" + temp1, "</figure>")
             temp_tuples = content.replace("?", "#")
-            new_content = re.sub(pattern, "<noscript><img src=\"" + str(tuple[i][2]) + "\"></noscript></figure>",
-                                 temp_tuples)
+            
+            # new_content = re.sub(pattern, "<noscript><img src=\"" + new_url + "\"></noscript></figure>",
+            #                      temp_tuples)
+            url_replace = str(new_url) + "-w"
+            new_content = content.replace('"'+temp1+'"', '"'+url_replace+'"')
+            
+            new_content = new_content.replace("<noscript>", "")
+            new_content = new_content.replace("</noscript>", "")
+            
+            # url_replace = str(new_url) + "-w"
+            # new_content = new_content.replace('"'+new_url+'"', url_replace)
+            
             new_content = r'%s' % (new_content)
             new_content = escape_string(new_content)
+            
             sql = """update {} set new_content = '{}' WHERE {} = '{}' """.format(table, new_content, key_id, find_id)
             self.cur.execute(sql)
             self.conn.commit()