def parse_page(self, response):
if 'tieba' not in response.url:
return
el = ExampleLoader(response=response)
el.add_xpath('name', '//title[1]/text()')
el.add_value('url', response.url)
return el.load_item()
def parse_directory(self, response):
hxs = HtmlXPathSelector(response)
for li in hxs.select('//ul[@class="directory-url"]/li'):
el = ExampleLoader(selector=li)
el.add_xpath('name', 'a/text()')
el.add_xpath('description', 'text()')
el.add_xpath('link', 'a/@href')
el.add_value('url', response.url)
yield el.load_item()
def parse_directory(self, response):
hxs = Selector(response)
for li in hxs.xpath('//ul[@class="directory-url"]/li'):
el = ExampleLoader(selector=li)
el.add_xpath('name', 'a/text()')
el.add_xpath('description', 'text()')
el.add_xpath('link', 'a/@href')
el.add_value('url', response.url)
yield el.load_item()
def parse_directory(self, response):
hxs = Selector(response)
for li in hxs.xpath('//ul[@class="directory-url"]/li'):
el = ExampleLoader(selector=li)
el.add_xpath("name", "a/text()")
el.add_xpath("description", "text()")
el.add_xpath("link", "a/@href")
el.add_value("url", response.url)
yield el.load_item()
def parse_page(self, response):
el = ExampleLoader(response=response)
el.add_xpath('name', '//title[1]/text()')
el.add_value('url', response.url)
return el.load_item()
