Improve/refine syntax highlighter

[gorhill]

Instead of having countless issues opened for every single improvements/refining that could be brought to the syntax highlighter, here is one meta issue to handle them all. I will list all the actionable items in the top comment. Feel free to report un-handled cases or suggest more refinements.

To do:

• Add a filter list which purpose is to validate the static filtering parser
  • Done:
    • https://raw.githubusercontent.com/gorhill/uBlock/master/docs/tests/static-filtering-parser-checklist.txt
    • Subscribe to it and disable it after the list is imported, then use asset viewer to see the result.
• *$xhr,empty should be reported as valid
• The following redirect rules should be reported as invalid:
  • *$beacon,redirect-rule=empty
  • *$ping,redirect-rule=empty
  • *$websocket,redirect-rule=empty
• Detect invalid !# directives
  • Support auto-completion for valid directives
• Detect unmatched !#if-!#endif directives
• Handle line continuation -- I need to read and wrap my mind around CodeMirror's state object
• Detect and fix regex-based filters starting with /.*: they are inefficient and the leading .* accomplishes nothing.
• Ambiguous option separator, see Improve/refine syntax highlighter #1134 (comment).

Nice to have:

• Buttons to navigate to next/prev problematic filter in asset viewer and "My filters
• Warn in gutter about untokenizable static network filters
• Create clickable links for URLs present in comments
  • Implemented as follow: double clicking the URL will select whole URL, at which point you can use your browser's "Open in new tab" feature on the selection.
• Code folding ability:
  • !#if ... / !#endif blocks
  • !#include blocks

[Yuki2718]

Warn in gutter about untokenizable static network filters
This will be huge plus!

This is invalid now:

$all,~document

easylist/easylist#5621 (comment)

gorhill/uBlock#2385

[gorhill]

I will fix this asap. However I would like to understand the purpose of that /adguard. filter in EasyList. Someone knows?

[krystian3w]

Maybe some ad engine have name "adguard...".

[Yuki2718]

I will fix this asap. However I would like to understand the purpose of that /adguard. filter in EasyList. Someone knows?

Apparently came from the Fanboy's list so @ryanbr would know. Clearly it has nothing to do with AdGuard.
easylist/easylist@648b822

This look like first occurrence: ryanbr/fanboy-adblock@d51ac92

[gorhill]

If I am getting this right, that /adguard. filter has always been rejected by uBO, since it does not use the all property, ~document can't be used on its own.

I am trying to assess whether I should publish an emergency fix for the all,~document regression.

~document was added recently easylist/easylist@24784f0

This is puzzling: easylist/easylist#5286

---

[rdk@on assets.json_resources]$ grep '~document' *
DEU-0_DEU EasyList Germany.txt:@@||adonline.anonza.de^$~document,~third-party
easylist_EasyList.txt:/adguard.$~document,domain=~adguard-com.cdn.ampproject.org|~adguard.com|~adguard.mobi|~adguard.oneskyapp.com|~greinr.com|~marketplace.visualstudio.com
easylist_EasyList.txt:@@||lemon-ads.com^$~document,~third-party
[rdk@on assets.json_resources]$ 

[gorhill]

This is puzzling: easylist/easylist#5286

Yeah, /adguard. on its own would never cause strict-blocking to kick in, so it's unclear what the original issue was.

The other cases seem to be about not blocking 1st-party secondary resources without wholly disabling filtering -- so probably they are targeting ABP specifically.

Yeah, /adguard. on its own would never cause strict-blocking to kick in, so it's unclear what the original issue was.

This was fixed after gorhill/uBlock@41636c5 v1.21 and I think it was fixed very recently in legacy builds. In 1.16.4.22