@xtermjs/addon-search regex case-sensitive should behave like monaco

[esimkowitz]

Currently, the @xtermjs/addon-search plugin applies the case-sensitive flag to the full regex, which means that if it's false, meta-characters like \S, which is non-whitespace chars, will be interpreted as lowercase (in this case, \s is whitespace chars). Monaco uses a different method, where only non-meta chars are interpreted as case-sensitive when the regex flag is set. While it's kind of silly to treat a regex as case-insensitive, it makes sense for xtermjs and monaco to align on their implementation.

Details

• Browser and browser version: Electron 33.2.1
• OS version: macOS Sequoia 15.2
• xterm.js version: 5.5.0

Steps to reproduce

1. run echo "path wrath hath" in the VSCode terminal
2. activate the terminal search, set case-sensitive and whole-word to false, and set regex to true
3. type \S+th in the search bar
4. observe that nothing is highlighted
5. Open a new tab in VSCode and type path wrath hath into it as its only contents
6. activate the Monaco search, set case-sensitive and whole-word to false, and set regex to true
7. type \S+th in the search bar
8. observe that each word is being highlighted separately

[esimkowitz]

I'm willing to look into a fix for this

[jerch]

Currently, the @xtermjs/addon-search plugin applies the case-sensitive flag to the full regex, which means that if it's false, meta-characters like \S, which is non-whitespace chars, will be interpreted as lowercase (in this case, \s is whitespace chars).

Sounds like a bug to me, special chars may never flip case, as it typically flips their in/out filter meaning. Shouldn't the i flag alone deal with that? 🤔

While it's kind of silly to treat a regex as case-insensitive, ...

Yes, I second that. Imho regexps should be treated the exact way they were given by default. If someone wants to ignore case, just append the i-flag to the regexp ctor on top.

For more context on the i-flag behavior, also see https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/RegExp/ignoreCase.

[Tyriar]

Yep looks like a bug. Relevant code is here:

https://github.com/anouartouati/xterm.js/blob/c71ef1b91a736db5fad024600f0303809215052d/addons/addon-search/src/SearchAddon.ts#L551-L552

So we just need a conditional flow there that takes this._searchOptions.regex into account. Could lift the handling of that from monaco too if we can find it.

[esimkowitz]

i can work on this if you wanna assign it to me

[AnouarTouati]

@esimkowitz I already worked on this I will create a pull request.