Merge remote-tracking branch 'origin/pr/3765' into adblock

qutebrowser/browser/adblock.py

@@ -94,14 +94,8 @@ class HostBlocker:
         _done_count: How many files have been read successfully.
         _local_hosts_file: The path to the blocked-hosts file.
         _config_hosts_file: The path to a blocked-hosts in ~/.config
-
-    Class attributes:
-        WHITELISTED: Hosts which never should be blocked.
     """
 
-    WHITELISTED = ('localhost', 'localhost.localdomain', 'broadcasthost',
-                   'local')
-
     def __init__(self):
         self._blocked_hosts = set()
         self._config_blocked_hosts = set()
@@ -234,16 +228,14 @@ class HostBlocker:
         parts = line.split()
         if len(parts) == 1:
             # "one host per line" format
-            host = parts[0]
-        elif len(parts) == 2:
-            # /etc/hosts format
-            host = parts[1]
+            hosts = [parts[0]]
         else:
-            log.misc.error("Failed to parse: {!r}".format(line))
-            return False
+            # /etc/hosts format
+            hosts = parts[1:]
 
-        if host not in self.WHITELISTED:
-            self._blocked_hosts.add(host)
+        for host in hosts:
+            if '.' in host and not host.endswith('.localdomain'):
+                self._blocked_hosts.add(host)
 
         return True
 

tests/unit/browser/test_adblock.py

@@ -120,8 +120,10 @@ def assert_urls(host_blocker, blocked=BLOCKLIST_HOSTS,
 
     Ensure URLs in 'blocked' and not in 'whitelisted' are blocked.
     All other URLs must not be blocked.
+
+    localhost is an example of a special case that shouldn't be blocked.
     """
-    whitelisted = list(whitelisted) + list(host_blocker.WHITELISTED)
+    whitelisted = list(whitelisted) + ['localhost']
     for str_url in urls_to_check:
         url = QUrl(str_url)
         host = url.host()
@@ -247,6 +249,16 @@ def test_successful_update(config_stub, basedir, download_stub,
     assert_urls(host_blocker, whitelisted=[])
 
 
+def test_parsing_multiple_hosts_on_line(config_stub, basedir, download_stub,
+                                        data_tmpdir, tmpdir, win_registry,
+                                        caplog):
+    """Ensure multiple hosts on a line get parsed correctly."""
+    host_blocker = adblock.HostBlocker()
+    bytes_host_line = ' '.join(BLOCKLIST_HOSTS).encode('utf-8')
+    host_blocker._parse_line(bytes_host_line)
+    assert_urls(host_blocker, whitelisted=[])
+
+
 def test_failed_dl_update(config_stub, basedir, download_stub,
                           data_tmpdir, tmpdir, win_registry, caplog):
     """One blocklist fails to download.
@@ -341,7 +353,7 @@ def test_blocking_with_whitelist(config_stub, basedir, download_stub,
     """Ensure hosts in content.host_blocking.whitelist are never blocked."""
     # Simulate adblock_update has already been run
     # by creating a file named blocked-hosts,
-    # Exclude localhost from it, since localhost is in HostBlocker.WHITELISTED
+    # Exclude localhost from it as localhost is never blocked via list
     filtered_blocked_hosts = BLOCKLIST_HOSTS[1:]
     blocklist = create_blocklist(data_tmpdir,
                                  blocked_hosts=filtered_blocked_hosts,