prevent-abbreviations - non-ASCII characters ignored in filenames

[MichaelBlm]

Fixes #2292

[MichaelBlm]

Here is an explanation of the updated regular expression:

1. (?=\P{Ll}): This is a positive lookahead assertion. It matches a position where the next character is not a Unicode lowercase letter. \P{Ll} is a Unicode character class that negates the Ll (lowercase letter) class.

2. (?<=\P{L}): This is a positive lookbehind assertion. It matches a position where the previous character is not a Unicode letter (uppercase or lowercase). \P{L} negates the L (letter) Unicode character class.

[fisker]

The solution looks tricky to me, I prefer just ignore if non-ascii characters included.

[sindresorhus]

Use the verbose names (Lowercase_Letter), for readability.

[MichaelBlm]

Updated to include the verbose names

[fisker]

We should make a rule for it 😄

[fisker]

ota-meshi/eslint-plugin-regexp#720

[MichaelBlm]

I'll work on it actually

[fisker]

Note: the proposal is on another repo.

[sindresorhus]

@fisker Tricky how? The solution looks fine to me.

[MichaelBlm]

My explanation of the regex wasn't that clear. \P{Letter} is equivalent to [^a-zA-Z] but it includes non-ASCII characters

[fisker]

@fisker Tricky how? The solution looks fine to me.

In different languages, the latin letters can be used with other characters together to be a word, so the non-letter characters not a word boundary.

For example, in Chinese, "B超" "T恤" are words. Similar in Japanese, I think.

[MichaelBlm]

@fisker Tricky how? The solution looks fine to me.

In different languages, the latin letters can be used with other characters together to be a word, so the non-letter characters not a word boundary.

For example, in Chinese, "B超" "T恤" are words. Similar in Japanese, I think.

Can you please provide me a valid/invalid test case to see if this fix doesn't already account for that?

[fisker]

I don't really have a case now.

[fregante]

@fisker you're saying that 龖.js will be counted as "one-letter long", but in Chinese it's a fully-formed "word", not an abbreviation. Hence "ignore if non-ascii characters included"

[fisker]

I don't think Chinese characters are "letter"s, but I can be wrong.

[fregante]

That's right, it's why quoted "one-letter word". The rule counts characters as "letters" but not all writing systems have letters.

https://en.wikipedia.org/wiki/Writing_system#Basic_terminology

The rule is based on the idea that no concept can be defined in 2 characters because in most languages words are longer than that.

[fregante]

Probably not helpful here, but: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Segmenter

[MichaelBlm]

I'll explore a solution using the segmenter. Looks promising

[MichaelBlm]

I added this code to the valid tests and it is passing with the current implementation? Is that the correct behavior?
{ code: 'foo();', filename: '龖.js', },