def get_detail_url(self): detail_url = [] for url in self.get_list_url(): resp = get_response(url=url, sess=self.sess) if resp is not None: resp.encoding = 'gb2312' doc = PyQuery(resp.text) for each in doc("td[width='250']").items(): detail_url.append(each('a').attr.href) return detail_url
def get_download_link(self): download_link = [] for url in self.get_detail_url(): resp = get_response(url=url, sess=self.sess) if resp is not None: resp.encoding = 'gb2312' doc = PyQuery(resp.text) for each in doc("table[bgcolor='#0099cc'] > tbody > tr").items(): if u'网盘' not in each.text() and each('a').attr.href is not None and not each( 'a').attr.href.endswith('html'): download_link.append(each('a').attr.href) break return download_link
def get_download_link(self): download_link = [] for url in self.get_detail_url(): resp = get_response(url=url, sess=self.sess) if resp is not None: resp.encoding = 'gb2312' doc = PyQuery(resp.text) for each in doc( "table[bgcolor='#0099cc'] > tbody > tr").items(): if u'网盘' not in each.text() and each( 'a').attr.href is not None and not each( 'a').attr.href.endswith('html'): download_link.append(each('a').attr.href) break return download_link