News download: Add support for turning off SSL certificate verification

src/calibre/__init__.py

@@ -377,20 +377,21 @@ def random_user_agent(choose=None):
         choose = random.randint(0, len(choices)-1)
     return choices[choose]
 
-def browser(honor_time=True, max_time=2, mobile_browser=False, user_agent=None, use_robust_parser=False):
+def browser(honor_time=True, max_time=2, mobile_browser=False, user_agent=None, use_robust_parser=False, verify_ssl_certificates=True):
     '''
     Create a mechanize browser for web scraping. The browser handles cookies,
     refresh requests and ignores robots.txt. Also uses proxy if available.
 
     :param honor_time: If True honors pause time in refresh requests
     :param max_time: Maximum time in seconds to wait during a refresh request
+    :param verify_ssl_certificates: If false SSL certificates errors are ignored
     '''
     from calibre.utils.browser import Browser
     if use_robust_parser:
         import mechanize
-        opener = Browser(factory=mechanize.RobustFactory())
+        opener = Browser(factory=mechanize.RobustFactory(), verify_ssl=verify_ssl_certificates)
     else:
-        opener = Browser()
+        opener = Browser(verify_ssl=verify_ssl_certificates)
     opener.set_handle_refresh(True, max_time=max_time, honor_time=honor_time)
     opener.set_handle_robots(False)
     if user_agent is None:

src/calibre/srv/errors.py

@@ -11,6 +11,9 @@ import httplib
 class JobQueueFull(Exception):
     pass
 
+class RouteError(ValueError):
+    pass
+
 class HTTPSimpleResponse(Exception):
 
     def __init__(self, http_code, http_message='', close_connection=False, location=None):

src/calibre/utils/browser.py

@@ -5,23 +5,51 @@ __license__   = 'GPL v3'
 __copyright__ = '2010, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
-import copy
+import copy, httplib, ssl
 from cookielib import CookieJar
 
-from mechanize import Browser as B
+from mechanize import Browser as B, HTTPSHandler
+
+class ModernHTTPSHandler(HTTPSHandler):
+
+    ssl_context = None
+
+    def https_open(self, req):
+        if self.client_cert_manager is not None:
+            key_file, cert_file = self.client_cert_manager.find_key_cert(
+                req.get_full_url())
+            if cert_file:
+                self.ssl_context.load_cert_chain(cert_file, key_file)
+        def conn_factory(hostport):
+            return httplib.HTTPSConnection(hostport, context=self.ssl_context)
+        return self.do_open(conn_factory, req)
+
 
 class Browser(B):
     '''
     A cloneable mechanize browser. Useful for multithreading. The idea is that
     each thread has a browser clone. Every clone uses the same thread safe
     cookie jar. All clones share the same browser configuration.
+
+    Also adds support for fine-tuning SSL verification via an SSL context object.
     '''
 
+    handler_classes = B.handler_classes.copy()
+    handler_classes['https'] = ModernHTTPSHandler
+
     def __init__(self, *args, **kwargs):
         self._clone_actions = {}
+        sc = kwargs.pop('ssl_context', None)
+        if sc is None:
+            sc = ssl.create_default_context() if kwargs.pop('verify_ssl', True) else ssl._create_unverified_context(cert_reqs=ssl.CERT_NONE)
 
         B.__init__(self, *args, **kwargs)
         self.set_cookiejar(CookieJar())
+        self._ua_handlers['https'].ssl_context = sc
+
+    @property
+    def https_handler(self):
+        return self._ua_handlers['https']
 
     def set_handle_refresh(self, *args, **kwargs):
         B.set_handle_refresh(self, *args, **kwargs)
@@ -89,6 +117,7 @@ class Browser(B):
 
     def clone_browser(self):
         clone = Browser()
+        clone.https_handler.ssl_context = self.https_handler.ssl_context
         clone.addheaders = copy.deepcopy(self.addheaders)
         for func, args, kwargs in self._clone_actions.values():
             func = getattr(clone, func)