def infomation_crew(self):
# 一条一条取
total_count = 20000
i = 0
while i < total_count:
i += 1
# 去除诗词的详情页url
info = self.db.select_unanalyzed_infomation()
if info is not None:
# 下载分析
url = info['url']
html = Downloader.get_html(url, 'infomation')
if html:
content = Analyzer.get_info_detail(html)
if content:
self.db.update_infomationurl(url, content)
print '%d/%d %s %s' % (i, total_count, info['title'], url)
else:
self.db.insert_error('analyze_info_detail_error', 7, 'reason', url)
else:
self.db.insert_error('download_info_detail_error', 6, 'reason', url)
else:
# 没有了
return
