SOCKS proxy (#281)

warcserver: SOCKS proxy:
- add support for running warcserver through a socks proxy specified via SOCKS_HOST and SOCKS_PORT
- move socks patch setup, http max_header adjustment to http module
- logging: print stack trace only if debugging
- add pysocks to extra_requirements, enable in ci
- add simple test (not actual proxy) to check that connection through proxy is attempted
- docs: add SOCKS proxy section to docs

appveyor.yml

@@ -17,7 +17,7 @@ install:
   - "pip install coverage pytest-cov coveralls"
   - "pip install cffi"
   - "pip install pyopenssl"
-  - "pip install certauth boto3 youtube-dl"
+  - "pip install certauth boto3 youtube-dl pysocks"
   - "pip install codecov"
 
 build_script:

docs/manual/configuring.rst

@@ -223,6 +223,16 @@ This configures the ``/live/`` route to point to the live web.
 This collection can be useful for testing, or even more powerful, when combined with recording.
 
 
+SOCKS Proxy for Live Web
+""""""""""""""""""""""""
+
+pywb can be configured to use a SOCKS5 proxy when connecting to the live web. This allows pywb to be used with `Tor <https://torproject.org/>`_ and other
+services that require a SOCKS proxy.
+
+If the ``SOCKS_HOST`` and optionally ``SOCKS_PORT`` environment variables are set, pywb will attempt to route all live web traffic through the SOCKS5 proxy.
+Note that, at this time, it is not possible to configure a SOCKS proxy per pywb collection -- all live web traffic will use the SOCKS proxy if enabled.
+
+
 .. _auto-all:
 
 Auto "All" Aggregate Collection

extra_requirements.txt

@@ -4,3 +4,4 @@ boto3
 uwsgi
 git+https://github.com/t0m/pyamf.git@python3
 git+https://github.com/esnme/ultrajson.git
+pysocks

pywb/warcserver/http.py

@@ -1,9 +1,78 @@
 from requests.adapters import HTTPAdapter
 import requests
+import os
 
+import six.moves.http_client
+six.moves.http_client._MAXHEADERS = 10000
+
+SOCKS_PROXIES = None
+orig_getaddrinfo = None
+
+
+#=============================================================================
 class DefaultAdapters(object):
     live_adapter = HTTPAdapter(max_retries=3)
     remote_adapter = HTTPAdapter(max_retries=3)
 
 requests.packages.urllib3.disable_warnings()
 
+
+#=============================================================================
+def patch_socks():
+    try:
+        import socks
+    except ImportError:  #pragma: no cover
+        print('Ignoring SOCKS_HOST: PySocks must be installed to use SOCKS proxy')
+        return
+
+    import socket
+
+    socks_host = os.environ.get('SOCKS_HOST')
+    socks_port = os.environ.get('SOCKS_PORT', 9050)
+
+    # Set socks proxy and wrap the urllib module
+    socks.set_default_proxy(socks.PROXY_TYPE_SOCKS5, socks_host, socks_port, True)
+    #socket.socket = socks.socksocket # sets default socket to be the sockipy socket
+
+    # store original getaddrinfo
+    global orig_getaddrinfo
+    orig_getaddrinfo = socks.socket.getaddrinfo
+
+    # Perform DNS resolution through socket
+    def getaddrinfo(*args):
+        if args[0] in ('127.0.0.1', 'localhost'):
+            res = orig_getaddrinfo(*args)
+
+        else:
+            res = [(socket.AF_INET, socket.SOCK_STREAM, 6, '', (args[0], args[1]))]
+
+        return res
+
+    socks.socket.getaddrinfo = getaddrinfo
+
+    socks_url = 'socks5h://{0}:{1}'.format(socks_host, socks_port)
+
+    global SOCKS_PROXIES
+    SOCKS_PROXIES = {'http': socks_url,
+                     'https': socks_url}
+
+# =============================================================================
+def unpatch_socks():
+    global orig_getaddrinfo
+    if not orig_getaddrinfo:
+        return
+
+    import socks
+    socks.socket.getaddrinfo = orig_getaddrinfo
+    orig_getaddrinfo = None
+
+    global SOCKS_PROXIES
+    SOCKS_PROXIES = None
+
+
+# =============================================================================
+if os.environ.get('SOCKS_HOST'):
+    patch_socks()
+
+
+

pywb/warcserver/resource/responseloader.py

@@ -14,7 +14,7 @@ from pywb.utils.format import ParamFormatter
 from pywb.warcserver.resource.resolvingloader import ResolvingLoader
 from pywb.warcserver.resource.pathresolvers import DefaultResolverMixin
 
-from pywb.warcserver.http import DefaultAdapters
+from pywb.warcserver.http import DefaultAdapters, SOCKS_PROXIES
 
 from six.moves.urllib.parse import urlsplit, quote, unquote
 
@@ -30,9 +30,6 @@ import logging
 
 from requests.models import PreparedRequest
 
-import six.moves.http_client
-six.moves.http_client._MAXHEADERS = 10000
-
 logger = logging.getLogger('warcserver')
 
 
@@ -447,11 +444,15 @@ class LiveWebLoader(BaseLoader):
 
     def _do_request(self, method, load_url, data, req_headers, params, is_live):
         adapter = DefaultAdapters.live_adapter if is_live else DefaultAdapters.remote_adapter
-        pool = adapter.poolmanager
         max_retries = adapter.max_retries
 
+        if SOCKS_PROXIES:
+            conn = adapter.get_connection(load_url, SOCKS_PROXIES)
+        else:
+            conn = adapter.poolmanager
+
         try:
-            upstream_res = pool.urlopen(method=method,
+            upstream_res = conn.urlopen(method=method,
                                         url=load_url,
                                         body=data,
                                         headers=req_headers,
@@ -465,7 +466,11 @@ class LiveWebLoader(BaseLoader):
             return upstream_res
 
         except Exception as e:
+            if logger.isEnabledFor(logging.DEBUG):
+                import traceback
+                traceback.print_exc()
                 logger.debug('FAILED: ' + method + ' ' + load_url + ': ' + str(e))
+
             raise LiveResourceException(load_url)
 
     def get_custom_metadata(self, content_type, dt):

tests/test_socks.py

@@ -0,0 +1,38 @@
+from .base_config_test import BaseConfigTest, fmod_sl
+
+import pywb.warcserver.http as pywb_http
+import os
+import socket
+import gevent
+import pytest
+
+
+# ============================================================================
+class TestSOCKSProxy(BaseConfigTest):
+    @classmethod
+    def setup_class(cls):
+        os.environ['SOCKS_HOST'] = 'localhost'
+        os.environ['SOCKS_PORT'] = '8080'
+
+        pywb_http.patch_socks()
+        import pywb.warcserver.resource.responseloader
+        pywb.warcserver.resource.responseloader.SOCKS_PROXIES = pywb_http.SOCKS_PROXIES
+        super(TestSOCKSProxy, cls).setup_class('config_test.yaml')
+
+    @classmethod
+    def teardown_class(cls):
+        pywb_http.unpatch_socks()
+        super(TestSOCKSProxy, cls).teardown_class()
+
+    def test_socks_proxy_set(self):
+        assert pywb_http.SOCKS_PROXIES == {'http': 'socks5h://localhost:8080',
+                                           'https': 'socks5h://localhost:8080'
+                                          }
+
+    def test_socks_attempt_connect(self, fmod_sl):
+        pytest.importorskip('socks')
+        # no proxy is set, expect to fail if socks is being used
+        resp = self.get('/live/{0}http://httpbin.org/get', fmod_sl, status=400)
+        assert resp.status_int == 400
+
+