#!/usr/bin/python import MySQLdb import re links = dict() content = dict() def get_data(): db = MySQLdb.connect(user='my_user', passwd='my_password', host='localhost', db='my_db') cursor = db.cursor() query = "select id, linkUrl, contentUrl from wp_shashin_photo" cursor.execute(query) res = cursor.fetchall() for (id, linkurl, contenturl) in res: links[str(id)] = linkurl content[str(id)] = contenturl cursor.close() db.close() get_data() blogexport = open('oltcuisine.wordpress.2015-06-04-with-gallery-links.xml', 'r') blogexportfixed = open('oltcuisine.wordpress.2015-06-04-with-gallery-links-fixed.xml', 'w') for line in blogexport: repline = line m = re.search('\[simage=([0-9]+).*?\]', line) if not m is None: tag = m.group(0) id = m.group(1) if(id in links and id in content): repline = repline.replace(tag, '
') m2 = re.search('\[shashin type="photo" id="([0-9]+)".*?\]', line) if not m2 is None: tag = m2.group(0) id = m2.group(1) if(id in links and id in content): repline = repline.replace(tag, '') blogexportfixed.write(repline) blogexportfixed.close()