||*tube.com^ for youtube.com is not working

[gwarser]

Filter issues MUST NOT be reported here. Read first: https://github.com/gorhill/uBlock/blob/master/CONTRIBUTING.md

Describe the issue

I want to optimize a bit this filter list https://github.com/azet12/KAD by enclosing domains inside ||...^
to be sure domain filters match inside domain (in ABP - UBO does this by default I think).
Most of combinations works fine:

||*.youtube.com^
||*youtube.com^
||youtube.*^
||youtube.com^

and ||you*.com^!

but not:

||*tube.com^
or
||*ww.youtube.com^

There is no match for youtube.com (www.youtube.com) in UBO. ABP works fine.

Example filter from filter list which I want to convert:

! 55pakiety-lte-5gb.blogspot.com
! 61pakiety-lte-5gb.blogspot.com
! e54pakiety-lte-5gb.blogspot.com
! e51pakiety-lte-5gb.blogspot.com
*pakiety-lte-5gb.blogspot.com

*pakiety-lte-5gb.blogspot.com is working, but I think it may be bit slower in APB (will match in path?) and I want to make these domain filters consistent.
Also, in path part of URL, UBO requires * to match word partially, but not here.

One or more specific URLs where the issue occurs

http://www.spidersweb.pl/2017/05/nvidia-max-q-laptopy.html

Steps for anyone to reproduce the issue

Create filter ||*tube.com^ in my filters
Open http://www.spidersweb.pl/2017/05/nvidia-max-q-laptopy.html
Youtube iframe is not blocked

Your settings

[If you fail to provide this info, I will mark the issue as invalid. Lists all settings which differs from default settings]

• OS/version: Mint 17.3
• Browser/version: Fx 55, 1.9.818.50 (Stable channel) (64-bit)
• uBlock Origin version: 1.12.4, 1.12.5b17

Your filter lists

default

Your custom filters (if any)

||*tube.com^

[lewisje]

I believe that this is intentional; the relevant regexes are .reHostnameRule2 and .reCleanupHostnameRule2, and the cleanup work is done here.

That is, it strips the leading * out of a hostname-anchored rule.

[gwarser]

These are working:

||you*.com^
||youtube.c*^
*tube.com^

[gorhill]

That is, it strips the leading * out of a hostname-anchored rule

That would change the meaning of the filter. The expected behavior is that the filter will be instanciated internally as a FilterGenericHnAnchored, however it's being classified as a pure hostname, which is wrong. I am investigating why this happens.

[gorhill]

Ok so reHostnameRule2 regex /^\**[0-9a-z][0-9a-z.-]*[0-9a-z]\^?$/i is wrong, I can't remember if there was a reason why I crafted it the way it is or whether it's just an oversight.

The ||*tube.com^ case is handled later in the function, but because of the faulty regex, that code path is never reached.

So it looks like it's just an oversight, given that I can't find any such filter in EasyList, EasyPrivacy, RU AdList, EasyList Germany, EasyList China (I stopped looking after this).

[gorhill]

UBO does this by default I think

Just to be sure there is no misunderstanding, uBO won't automatically anchor with || this if there is at least one asterisk in the filter.

[gwarser]

given that I can't find any such filter in...

Check in JPN: ABP Japanese filters (日本用フィルタ)

Just to be sure there is no misunderstanding,...

I was thinking about hosts-file-like plain domains.

[gorhill]

Yes, that list has 2 instances which were improperly categorized by uBO:

||*adexchange.com^$document
||*adexchange.com^$popup

The others ||*... instances were properly categorized because they still did not match reHostnameRule2 because of other stuff in these filters.

[lewisje]

That would change the meaning of the filter.

I agree, but I still thought it was intentional: I remember an earlier case where syntax was re-interpreted in a way that changed the meaning of filters and it was regarded as intentional (I believe it was later changed).

[gwarser]

@lewisje I was thinking about this, but cannot find this issue. Will be good to create wiki page with quirks like this.

[gorhill]

Unlike with ABP, tube.com/allo won't match https://youtube.com/allo with uBO, but *tube.com/allo will match with uBO, that is the difference. So given this, ||*tube.com^ should have matched in the current issue.