def main(): SiteRule.drop_collection() fin = open('scripts/sub_or_article.csv','r') for line in fin: title, site_url, sub_index_url, sub_index_title, sub_index_cate = line.strip().split(',') if sub_index_title == 'Home': continue if 'http' not in site_url: site_url = 'http://%s' % site_url site_url = re.sub('/$', '', site_url) rules = SiteRule() rules.title = title.strip() rules.site_url = site_url.strip() rules.sub_index_title = sub_index_title.strip().decode('gb18030').encode('utf8') rules.sub_index_url = sub_index_url.strip() if not sub_index_cate == 'Null': rules.category_name = sub_index_cate.strip() print title, site_url, sub_index_cate, sub_index_title.decode('gb18030').encode('utf8'), sub_index_url rules.save() fin.close()