async def parse_page(self, session, url):
proxys = await self.get(session, url) >> to_doc >> extra_xpath(
'//table[@id="ip_list"]//tr[position()>1]')
rst = []
for proxy in proxys:
host = proxy >> extra_xpath(
'.//td[position()=2]/text()') >> extra_head
port = proxy >> extra_xpath(
'.//td[position()=3]/text()') >> extra_head
rst.append((host, port))
return rst
async def parse_page(self, session, url):
text = await self.get(session, url)
proxys = text >> to_doc >> extra_xpath("//table//tr[position()>1]")
rst = []
for proxy in proxys:
host = proxy >> extra_xpath(
"./td[position()=1]/text()") >> extra_head
port = proxy >> extra_xpath(
"./td[position()=2]/text()") >> extra_head
rst.append((host, port))
return rst
async def parse_page(self, session, url):
text = await self.get(session, url)
self.headers["Referer"] = 'https://ip.ihuan.me/ti.html'
text = await self.get(session, 'https://ip.ihuan.me/mouse.do')
val = self.pattern.search(text).group(0).replace('val("', '')
data = {
'num': 500,
'port': '',
'kill_port': '',
'address': '',
'kill_address': '',
'anonymity': '',
'type': '',
'post': '',
'sort': '',
'key': val
}
self.headers['Origin'] = 'https://ip.ihuan.me'
self.headers['Content-Type'] = 'application/x-www-form-urlencoded'
text = await self.post(session,
'https://ip.ihuan.me/tqdl.html',
data=data)
proxys = text >> to_doc >> extra_xpath(
'//div[@class="panel-body"]/text()')
rst = []
for proxy in proxys:
try:
host, port = (proxy >> extra_head).split(':')
except Exception:
continue
rst.append((host, port))
return rst
async def prepare(self, session):
urls = await self.get(
session, "http://www.xsdaili.com/") >> to_doc >> extra_xpath(
'//div[@class="panel-body"]//div[@class="title"]/a/@href')
self.funcmap = {
self.parse_page: ['http://www.xsdaili.com' + ele for ele in urls]
}
async def parse_page(self, session, url):
proxies = await self.get(session,
url) >> to_doc >> extra_xpath("//text()")
rst = []
for proxy in proxies[7:-1]:
[host, port] = proxy.strip('\n\t\r ').split(':')
rst.append((host, port))
return rst
async def parse_page(self, session, url):
text = await self.get(session, url)
proxys = text >> to_doc >> extra_xpath("//body//text()")
rst = []
for proxy in proxys[:-10]:
proxy = proxy.strip('" \t\r\n')
[host, port] = proxy.split(':')
rst.append((host, port))
return rst
async def parse_page(self, session, url):
proxys = await self.get(
session,
url) >> to_doc >> extra_xpath('//div[@class="cont"]/text()')
rst = []
for proxy in proxys:
proxy = proxy.strip().split('@')[0]
rst.append(tuple(proxy.split(":")))
return rst