[fix] google news dom xpath fix

This commit is contained in:
misnyo 2017-08-31 17:48:07 +02:00
parent e74aaa781e
commit 3182ba7069
2 changed files with 58 additions and 8 deletions

View file

@ -67,8 +67,8 @@ def response(resp):
for result in dom.xpath('//div[@class="g"]|//div[@class="g _cy"]'): for result in dom.xpath('//div[@class="g"]|//div[@class="g _cy"]'):
try: try:
r = { r = {
'url': result.xpath('.//div[@class="_cnc"]//a/@href')[0], 'url': result.xpath('.//a[@class="l _PMs"]')[0].attrib.get("href"),
'title': ''.join(result.xpath('.//div[@class="_cnc"]//h3//text()')), 'title': ''.join(result.xpath('.//a[@class="l _PMs"]//text()')),
'content': ''.join(result.xpath('.//div[@class="st"]//text()')), 'content': ''.join(result.xpath('.//div[@class="st"]//text()')),
} }
except: except:

File diff suppressed because one or more lines are too long