How to write a DownloadHandler for scrapy that makes requests through socksipy?

Question

I'm trying to use scrapy over Tor. I've been trying to get my head around how to write a DownloadHandler for scrapy that uses socksipy connections.

Scrapy's HTTP11DownloadHandler is here: https://github.com/scrapy/scrapy/blob/master/scrapy/core/downloader/handlers/http11.py

Here is an example for creating a custom download handler: https://github.com/scrapinghub/scrapyjs/blob/master/scrapyjs/dhandler.py

Here's the code for creating a SocksiPyConnection class: http://blog.databigbang.com/distributed-scraping-with-multiple-tor-circuits/

class SocksiPyConnection(httplib.HTTPConnection):
    def __init__(self, proxytype, proxyaddr, proxyport = None, rdns = True, username = None, password = None, *args, **kwargs):
        self.proxyargs = (proxytype, proxyaddr, proxyport, rdns, username, password)
        httplib.HTTPConnection.__init__(self, *args, **kwargs)

    def connect(self):
        self.sock = socks.socksocket()
        self.sock.setproxy(*self.proxyargs)
        if isinstance(self.timeout, float):
            self.sock.settimeout(self.timeout)
        self.sock.connect((self.host, self.port))

With the complexity of twisted reactors in the scrapy code, I can't figure out how plug socksipy into it. Any thoughts?

Please do not answer with privoxy-like alternatives or post answers saying "scrapy doesn't work with socks proxies" - I know that, which is why I'm trying to write a custom Downloader that makes requests using socksipy.

Answer 1

I was able to make this work with https://github.com/habnabit/txsocksx.

After doing a pip install txsocksx, I needed to replace scrapy's ScrapyAgent with txsocksx.http.SOCKS5Agent.

I simply copied the code for HTTP11DownloadHandler and ScrapyAgent from scrapy/core/downloader/handlers/http.py, subclassed them and wrote this code:

class TorProxyDownloadHandler(HTTP11DownloadHandler):

    def download_request(self, request, spider):
        """Return a deferred for the HTTP download"""
        agent = ScrapyTorAgent(contextFactory=self._contextFactory, pool=self._pool)
        return agent.download_request(request)


class ScrapyTorAgent(ScrapyAgent):
    def _get_agent(self, request, timeout):
        bindaddress = request.meta.get('bindaddress') or self._bindAddress
        proxy = request.meta.get('proxy')
        if proxy:
            _, _, proxyHost, proxyPort, proxyParams = _parse(proxy)
            scheme = _parse(request.url)[0]
            omitConnectTunnel = proxyParams.find('noconnect') >= 0
            if  scheme == 'https' and not omitConnectTunnel:
                proxyConf = (proxyHost, proxyPort,
                             request.headers.get('Proxy-Authorization', None))
                return self._TunnelingAgent(reactor, proxyConf,
                    contextFactory=self._contextFactory, connectTimeout=timeout,
                    bindAddress=bindaddress, pool=self._pool)
            else:
                _, _, host, port, proxyParams = _parse(request.url)
                proxyEndpoint = TCP4ClientEndpoint(reactor, proxyHost, proxyPort,
                    timeout=timeout, bindAddress=bindaddress)
                agent = SOCKS5Agent(reactor, proxyEndpoint=proxyEndpoint)
                return agent

        return self._Agent(reactor, contextFactory=self._contextFactory,
            connectTimeout=timeout, bindAddress=bindaddress, pool=self._pool)

In settings.py, something like this is needed:

DOWNLOAD_HANDLERS = {
    'http': 'crawler.http.TorProxyDownloadHandler'
}

Now proxying with Scrapy with work through a socks proxy like Tor.

Answer 2

Try txsocksx that comes from one of the Twisted developers.

Thanks to greatness of Twsisted and Scrapy, you can easily use SOCKS as proxy:

---

In downloader.py:

import scrapy.core.downloader.handlers.http11 as handler
from twisted.internet import reactor
from txsocksx.http import SOCKS5Agent
from twisted.internet.endpoints import TCP4ClientEndpoint
from scrapy.core.downloader.webclient import _parse


class TorScrapyAgent(handler.ScrapyAgent):
    _Agent = SOCKS5Agent

    def _get_agent(self, request, timeout):
        proxy = request.meta.get('proxy')

        if proxy:
            proxy_scheme, _, proxy_host, proxy_port, _ = _parse(proxy)

            if proxy_scheme == 'socks5':
                endpoint = TCP4ClientEndpoint(reactor, proxy_host, proxy_port)

                return self._Agent(reactor, proxyEndpoint=endpoint)

        return super(TorScrapyAgent, self)._get_agent(request, timeout)


class TorHTTPDownloadHandler(handler.HTTP11DownloadHandler):
    def download_request(self, request, spider):
        agent = TorScrapyAgent(contextFactory=self._contextFactory, pool=self._pool,
                               maxsize=getattr(spider, 'download_maxsize', self._default_maxsize),
                               warnsize=getattr(spider, 'download_warnsize', self._default_warnsize))

        return agent.download_request(request)

Register new handler in settings.py:

DOWNLOAD_HANDLERS = {
    'http': 'crawler.downloader.TorHTTPDownloadHandler',
    'https': 'crawler.downloader.TorHTTPDownloadHandler'
}

---

Now, you only have to tell crawlers to use proxy. I recommend to do it via middleware:

class ProxyDownloaderMiddleware(object):
    def process_request(self, request, spider):
        request.meta['proxy'] = 'socks5://127.0.0.1:950'