From ccdb59cce10f385b0079a32508fa3a2afaeaa96d Mon Sep 17 00:00:00 2001
From: Daniel <kingdread@gmx.de>
Date: Wed, 16 Sep 2015 16:31:17 +0200
Subject: [PATCH] Add adblock host whitelisting

The config option "content host-blocking-whitelist" may contain comma
separated domains that are exempt from host blocking.

The listed domains may contain the wildcards "*" and "?" to match many
and one character, respectively.

You need to run :adblock-update after modifying the list.
---
 doc/help/settings.asciidoc       | 11 +++++++++++
 qutebrowser/browser/adblock.py   | 23 ++++++++++++++++++++++-
 qutebrowser/config/configdata.py |  8 ++++++++
 3 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/doc/help/settings.asciidoc b/doc/help/settings.asciidoc
index 796199a19..499c9c88b 100644
--- a/doc/help/settings.asciidoc
+++ b/doc/help/settings.asciidoc
@@ -159,6 +159,7 @@
 |<<content-cookies-store,cookies-store>>|Whether to store cookies.
 |<<content-host-block-lists,host-block-lists>>|List of URLs of lists which contain hosts to block.
 |<<content-host-blocking-enabled,host-blocking-enabled>>|Whether host blocking is enabled.
+|<<content-host-blocking-whitelist,host-blocking-whitelist>>|List of domains that should always be loaded, despite being ad-blocked.
 |==============
 
 .Quick reference for section ``hints''
@@ -1433,6 +1434,16 @@ Valid values:
 
 Default: +pass:[true]+
 
+[[content-host-blocking-whitelist]]
+=== host-blocking-whitelist
+List of domains that should always be loaded, despite being ad-blocked.
+
+Domains may contain * and ? wildcards and are otherwise required to exactly match the requested domain.
+
+Local domains are always exempt from hostblocking.
+
+Default: empty
+
 == hints
 Hinting settings.
 
diff --git a/qutebrowser/browser/adblock.py b/qutebrowser/browser/adblock.py
index bec74f096..0f4b50b75 100644
--- a/qutebrowser/browser/adblock.py
+++ b/qutebrowser/browser/adblock.py
@@ -24,6 +24,8 @@ import os.path
 import functools
 import posixpath
 import zipfile
+import fnmatch
+import re
 
 from qutebrowser.config import config
 from qutebrowser.utils import objreg, standarddir, log, message
@@ -59,6 +61,24 @@ def get_fileobj(byte_io):
     return io.TextIOWrapper(byte_io, encoding='utf-8')
 
 
+def is_whitelisted_domain(host):
+    """Check if the given host is on the adblock whitelist.
+
+    Args:
+        host: The host as given by the adblocker as string.
+    """
+    whitelist = objreg.get('config').get('content', 'host-blocking-whitelist')
+    if whitelist is None:
+        return False
+
+    for domain in whitelist:
+        fnmatch_translated = fnmatch.translate(domain)
+        domain_regex = re.compile(fnmatch_translated, re.IGNORECASE)
+        if domain_regex.match(host):
+            return True
+    return False
+
+
 class FakeDownload:
 
     """A download stub to use on_download_finished with local files."""
@@ -188,7 +208,8 @@ class HostBlocker:
             else:
                 error_count += 1
                 continue
-            if host not in self.WHITELISTED:
+            if (host not in self.WHITELISTED
+                    and not is_whitelisted_domain(host)):
                 self.blocked_hosts.add(host)
         log.misc.debug("{}: read {} lines".format(byte_io.name, line_count))
         if error_count > 0:
diff --git a/qutebrowser/config/configdata.py b/qutebrowser/config/configdata.py
index c9d82838d..cbede4a00 100644
--- a/qutebrowser/config/configdata.py
+++ b/qutebrowser/config/configdata.py
@@ -735,6 +735,14 @@ def data(readonly=False):
              SettingValue(typ.Bool(), 'true'),
              "Whether host blocking is enabled."),
 
+            ('host-blocking-whitelist',
+             SettingValue(typ.List(none_ok=True), ''),
+             "List of domains that should always be loaded, despite being "
+             "ad-blocked.\n\n"
+             "Domains may contain * and ? wildcards and are otherwise "
+             "required to exactly match the requested domain.\n\n"
+             "Local domains are always exempt from hostblocking."),
+
             readonly=readonly
         )),