:remove-attr() and :remove-class() pseudo selector operators

[mapx-]

@gorhill

gorhill/uBlock@992255e#commitcomment-92647587

from slack (@MasterKia)

Is this supported?
`##+js(ra, oncontextmenu|onselectstart|ondragstart)`
=>
`##*:remove-attr(oncontextmenu|onselectstart|ondragstart)`

We should know / have examples for the new operators

Also uBlock-user question:

Is there support for "stay"/"complete" option like we have in the scriptlet version ?

Edit: Will need to use :watch-attr(), seems to be a working replacement for the aformentioned options.

[uBlock-user]

##*:remove-attr(oncontextmenu|onselectstart|ondragstart)

Regex is.

[gorhill]

Argument can be text or regex:

// remove exact oncontextmenu attribute
##*:remove-attr(oncontextmenu)

// remove attributes which match the regexes
##*:remove-attr(/oncontextmenu|onselectstart|ondragstart/)
##*:remove-attr(/^on[a-z]+/)

And as with other pseudoclass operators, use quotes when parser is tripped.

[gothic-bum]

Do you recommend the use of ##*:remove-attr(oncontextmenu) vs the regex equivalent of ##*:remove-attr(/^oncontextmenu$/)?

[gorhill]

Well for real cases I recommend simplicity and smallest resultset to act on (as with any other operators):

...##[oncontextmenu]:remove-attr(oncontextmenu)

[mapx-]

@gorhill what about "stay/complete" (added above), should we use watch-attr ?

[gorhill]

Pseudoclass operators do not need stay or complete since they are continually enforced, like other operators. But you will need :watch-attr() if you want the operators to act when attributes change without layout changes.

[uBlock-user]

Pseudoclass operators do not need stay or complete since they are continually enforced, like other operators.

Without :watch-attr(), a filter such as www.google.com##body[jsaction]:remove-attr(jsaction) doesn't work, how come ?

[krystian3w]

How implemented is asap in once or both operators?

gorhill/uBlock@35d7406 uBlockOrigin/uAssets#9528

[krystian3w]

Also sometimes will be nice use only once without loop/MutationObserver:

FiltersHeroes/PolishAnnoyanceFilters#2667 (comment) (so far I do not have any active example where the loop harms the party).

[MasterKia]

https://github.com/uBlockOrigin/uBlock-issues/wiki/Static-filter-syntax#subjectremove-attr-subjectremove-class

[krystian3w]

link down?

https://github.com/uBlockOrigin/uBlock-issues/wiki/Static-filter-syntax#subjectremove-attr-subjectremove-class - old

https://github.com/uBlockOrigin/uBlock-issues/wiki/Static-filter-syntax#subjectremove-attrarg-subjectremove-classarg - latest?

[ghost]

A case with :watch-attr()?

https://www.reddit.com/r/uBlockOrigin/comments/105dkbp/removeclass_works_different_in_comparison_to_rcjs/

[uBlock-user]

class attribute is very generic, would it cause any performance issues ?

[gorhill]

Not an issue when :watch-attr() applies to the actual element(s) being targeted. Attribute changes outside these elements won't trigger the mutation observer.

[Yuki2718]

On welovemanga.one with Chrome, welovemanga.*##body:remove-attr(onclick) does not work - onclick is removed but popup still occurs. welovemanga.*##+js(ra, onclick, body) works fine.

[gorhill]

Given there is the nowoif solution, I will still consider ra to be deprecated in the long term, and nowoif is also preferable anyway because it's properly imported into uBOL, while ra is not (remove-attr would be also properly imported).

[ghost]

the attribute removal is made from the page world, while with remove-attr, the attribute removal is made from the content script world.

Wondering, can this be the cause of why YT or Twitch video still plays after video element is removed from DOM by :remove()? For example:

! https://www.youtube.com/user/LinusTechTips
youtube.com##ytd-channel-video-player-renderer:remove()

I was never able to figure out what is the source of the audio which still plays.

---

Manually removing this element in Dev Tools works like expected breaking the playback, uBO filter does not.

---

https://www.reddit.com/r/uBlockOrigin/comments/106p2ho/how_do_you_block_this_crap_on_youtube/

---

The whole YT interface is build out of custom elements like <ytd-channel-video-player-renderer>, so maybe it match what https://developer.mozilla.org/en-US/docs/Mozilla/Add-ons/WebExtensions/Content_scripts#content_script_environment says.

[joridiculous]

This seems to work: https://www.reddit.com/r/uBlockOrigin/comments/goopft/youtube_channel_trailers/frh43am/
(In one of the google search links you posted in my reddit thread)

Mind im on an older Chrome, and some developer version 1.45.3b12

[uBlock-user]

with ra.js, the attribute removal is made from the page world, while with remove-attr, the attribute removal is made from the content script world.

So is this why :remove() doesn't work on inline javascripts ?

[krystian3w]

Maybe due race condition if are executed with same speed like meta refresh?

[mapx-]

anti right click:

working => warning.or.kr##+js(ra, oncontextmenu)

not working => warning.or.kr##body:remove-attr(oncontextmenu)
(even if it seems oncontextmenu is removed)

tested in chrome
@gorhill

[uBlock-user]

Due to #2403 (reply in thread)

So I think I understand the behavior is because with ra.js, the attribute removal is made from the page world, while with remove-attr, the attribute removal is made from the content script world.