def _is_blocked(self,
request_url: QUrl,
first_party_url: QUrl = None) -> bool:
"""Check whether the given request is blocked."""
if not self.enabled:
return False
if first_party_url is not None and not first_party_url.isValid():
first_party_url = None
qtutils.ensure_valid(request_url)
if not config.get("content.blocking.enabled", url=first_party_url):
return False
if blockutils.is_whitelisted_url(request_url):
return False
host = request_url.host()
if config.get("content.blocking.hosts.block_subdomains"):
return any(hostname in self._blocked_hosts
or hostname in self._config_blocked_hosts
for hostname in urlutils.widened_hostnames(host))
else:
return (host in self._blocked_hosts
or host in self._config_blocked_hosts)
def get_for_url(self, url: QUrl = None, *, fallback: bool = True) -> Any:
"""Get a config value, falling back when needed.
This first tries to find a value matching the URL (if given).
If there's no match:
With fallback=True, the global/default setting is returned.
With fallback=False, usertypes.UNSET is returned.
"""
self._check_pattern_support(url)
if url is None:
return self._get_fallback(fallback)
qtutils.ensure_valid(url)
candidates: List[ScopedValue] = []
# Urls trailing with '.' are equivalent to non-trailing types.
# urlutils strips them, so in order to match we will need to as well.
widened_hosts = urlutils.widened_hostnames(url.host().rstrip('.'))
# We must check the 'None' key as well, in case any patterns that
# did not have a domain match.
for host in itertools.chain(widened_hosts, [None]):
host_set = self._domain_map.get(host, ())
for scoped in host_set:
if scoped.pattern is not None and scoped.pattern.matches(url):
candidates.append(scoped)
if candidates:
scoped = max(candidates, key=operator.attrgetter('pattern_id'))
return scoped.value
if not fallback:
return usertypes.UNSET
return self._get_fallback(fallback)
def get_for_url(self,
url: QUrl = None,
*,
fallback: bool = True) -> typing.Any:
"""Get a config value, falling back when needed.
This first tries to find a value matching the URL (if given).
If there's no match:
With fallback=True, the global/default setting is returned.
With fallback=False, UNSET is returned.
"""
self._check_pattern_support(url)
if url is None:
return self._get_fallback(fallback)
candidates = [] # type: typing.List[ScopedValue]
widened_hosts = urlutils.widened_hostnames(url.host())
# We must check the 'None' key as well, in case any patterns that
# did not have a domain match.
for host in itertools.chain(widened_hosts, [None]):
host_set = self._domain_map.get(host, ())
for scoped in host_set:
if scoped.pattern is not None and scoped.pattern.matches(url):
candidates.append(scoped)
if candidates:
scoped = max(candidates, key=operator.attrgetter('pattern_id'))
return scoped.value
if not fallback:
return UNSET
return self._get_fallback(fallback)
def test_bench_widen_hostnames(self, hostname, benchmark):
benchmark(lambda: list(urlutils.widened_hostnames(hostname)))
def test_widen_hostnames(self, hostname, expected):
assert list(urlutils.widened_hostnames(hostname)) == expected