This concerns the static network filtering engine.
Related issue:
- https://github.com/uBlockOrigin/uBlock-issues/issues/943
* * *
New static network filter type: `cname`
By default, network requests which are result of
resolving a canonical name are subject to filtering.
This filtering can be bypassed by creating exception
filters using the `cname` option. For example:
@@*$cname
The filter above tells the network filtering engine
to except network requests which fulfill all the
following conditions:
- network request is blocked
- network request is that of an unaliased hostname
Filter list authors are discouraged from using
exception filters of `cname` type, unless there no
other practical solution such that maintenance
burden become the greater issue. Of course, such
exception filters should be as narrow as possible,
i.e. apply to specific domain, etc.
* * *
New static network filter option: `denyallow`
The purpose of `denyallow` is bring
default-deny/allow-exceptionally ability into static
network filtering arsenal. Example of usage:
*$3p,script, \
denyallow=x.com|y.com \
domain=a.com|b.com
The above filter tells the network filtering engine that
when the context is `a.com` or `b.com`, block all
3rd-party scripts except those from `x.com` and `y.com`.
Essentially, the new `denyallow` option makes it easier
to implement default-deny/allow-exceptionally in static
filter lists, whereas before this had to be done with
unwieldy regular expressions[1], or through the mix of
broadly blocking filters along with exception filters[2].
[1] https://hg.adblockplus.org/ruadlist/rev/f362910bc9a0
[2] Typically filters which pattern are of the
form `|http*://`
Related issue:
- https://github.com/uBlockOrigin/uBlock-issues/issues/923
Use current page's hostname instead of that of
image URLs to decide whether to reset pattern
union with previous picker sessions.
The fixed issue arose from the fact that the
page uses URLs from different origins in a
single srcset attribute.
Related commit:
- https://github.com/gorhill/uBlock/commit/703c525b01aa
This adds an indentation requirement for line
continuation to take place. The conditions are now
as follow:
- Current line ends with ` \`: ASCII space + backslash
- Next line starts with ` `: four ASCII spaces
If a line in a filter list ends with a space
(ASCII code 32) followed by a backslash
(ASCII code 92), those two characters will be
removed, the line will be trimmed and the next
line will be trimmed and concatenated to form
a new, longer line.
The purpose is to give filter list authors
a way to visually break apart unduly long
filters and thus make maintenance easier.
When line continuation is used, it is suggested
that the extra lines are prepended with four
space so as to make it more visually obvious that
the extra line(s) are the continuation of a
previous line.
Related issue:
- https://github.com/uBlockOrigin/uBlock-issues/issues/943
The filter referenced in the above issue was the
motivation to implement this feature:
- https://hg.adblockplus.org/ruadlist/rev/f362910bc9a0
I verified and could not find any instance in major
filter lists of lines ending with ` \`, thus the
change should be safe.
The new syntax deprecate the old syntax, though the
old syntax will still be supported until it's no
longer used in mainstream filter lists.
The new syntax is:
example.com##+js(window.open-defuser, pattern, seconds)
`pattern`:
A pattern to match for the defusing to take place.
Patterns which starts and ends with `/` will be
interpreted as regular expressions. To NOT match a
pattern, prefix with `!`.
`seconds`:
If not present, no window will be opened and the
scriptlet will return `null`.
If present and a valid integer value, the defuser
will return a valid window object even though no
popup window is opened. The returned window object
will cease to be valid after the specified number
of seconds.
Reported internally. The following invalid filter was not
discarded by uBO:
123tvnow.com##.123tv-ads
The correct form should be:
123tvnow.com##.\31 23tv-ads
Not discarding invalid CSS selector-based cosmetic
filter may break CSS selector-based cosmetic
filtering.
***
New procedural cosmetic operator: `:remove()`
Related issue:
- https://github.com/gorhill/uBlock/issues/2252
The purpose is to outright remove elements from the
DOM tree. Since `:remove()` is an "action" operator,
it must only be used as a trailing operator (just
like the `:style()` operator).
AdGuard's cosmetic filter syntax `{ remove: true; }`
will be converted to uBO's `:remove()` operator
internally.
***
New procedural cosmetic operator: `:upward(...)`
The purpose is to lookup an ancestor element.
When used with an integer argument, it is synonym of
`:nth-ancestor()`, which will be deprecated and which
will no longer be supported once no longer used in
mainstream filter lists.
Filter lists maintainers must only use `:upward(int)`
instead of `:nth-ancestor(int)` once the new operator
become available in all stable releases of uBO.
`:upward()` can also accept a CSS selector as argument,
in which case the nearest ancestor which matches the
CSS selector will be selected.
The scriptlet will now still try to trap a specific
property if a segment of the chain is not undefined
while yet not an object either.
For example, this now allows to set a value on
`document.body.onselectstart` when `document.body` has
not been instantiated yet by the browser parser,
whereas this would previously fail because
`document.body` would be `null` while the scriptlet
was testing against `undefined`.
The stricter mode `disable_non_proxied_udp` is preferable
to `default_public_interface_only` to prevent local IP
address leakage through WebRTC.
This mode is properly supported since Firefox 70, so the
less strict `default_public_interface_only` will now be
used only for Firefox 69 and older.
Override the Backspace key, not just the Delete key, as Mac keyboards have Backspace as the only delete key and label it Delete.
Source of key value: https://developer.mozilla.org/en-US/docs/Web/API/KeyboardEvent/key/Key_Values#Editing_keys
More background on Mac keyboard shortcuts: Mac keyboards can emulate Delete using fn+Delete, meaning Forward Delete, but Mac software does not use it except in text editing. When deletion is dangerous, Mac software requires holding a modifier key in conjuction with Delete, but I think it’s better to make deletion easy in this case.
This new binding has a potential downside: if the user Backspace key normally goes Back in history (which can differ across OSs and browsers), this will change the behavior to delete the selected element instead. If the user really wants to go back in history, they will have to press Escape to leave the mode and then press Backspace, or they will have to press an alternative keyboard shortcuts such as Alt+Left. I think the user will rarely want to go back in history in the middle of picking an element, though.
That downside could be mitigated by conditioning the key check on `runtime.PlatformOs` (https://developer.mozilla.org/en-US/docs/Mozilla/Add-ons/WebExtensions/API/runtime/PlatformOs). But that would complicate the code a lot compared to the importance of this feature, and such detection would still fail to capture the user’s intent accurately in all cases. I think it’s better to unconditionally accept both Backspace (Delete) and Delete (Forward Delete).
Raymond Hill