[css-text] Incorrect text-transform capitalize behavior for Catalan geminate Ls

[zipus]

I've noticed an issue with the text-transform:capitalize property in CSS when used on words containing "l·l" in Catalan (see Interpunct - Catalan). When this rule is applied to such words, the CSS interpreter incorrectly treats the interpunct as a word boundary and capitalizes the letter after it.

To illustrate, consider the word "cancel·lar". When text-transform:capitalize is applied, it currently results in "Cancel·Lar". However, according to Catalan language rules, it should be "Cancel·lar" instead. The interpunct is not considered a word boundary, and the second 'l' should not be capitalized.

I propose updating the algorithm used for text-transform:capitalize to respect language-specific case mapping rules such as this one.

[Loirooriol]

Relevant spec: https://drafts.csswg.org/css-text-4/#text-transform-mapping

For capitalize, what constitutes a “word“ is UA-dependent; [UAX29] is suggested (but not required) for determining such word boundaries.

I'm not sure what UAX29 would imply, but Blink and WebKit seem to behave as expected for Catalan, Firefox is wrong indeed (this reminds me of bug 257073). So filing a bug at https://bugzilla.mozilla.org/ may be a more practical solution.

Interestingly, Firefox considers canceŀlar to be a single word, using U+0140 instead of U+006C U+00B7, but the latter is recommended by unicode (and is more convenient to write).

[r12a]

@zipus thanks for pointing this out.

I created a Gecko bug report at https://bugzilla.mozilla.org/show_bug.cgi?id=1839354

And will be tracking it as a Catalan gap-analysis report at https://w3c.github.io/eurlreq/gap-analysis/latn-ca-gap#transforms

[zipus]

I created already a bugzilla: https://bugzilla.mozilla.org/show_bug.cgi?id=1839308

[r12a]

I updated our gap tracker to point to your bug. Thanks.

[moyogo]

This is true for other languages as well.

For example some other European languages like Sardinian ("agiuda·nos" capitalized as "Agiuda·nos"), in Gascon Occitan ("in·hèrn", "des·har" capitalized as "In·hèrn", "Des·har"), in one orthography of Arpitan ("ôtrich·ien", "g·eografia" capitalized "Ôtrich·ien", "G·eografia"), in one orthography of Gallo ("ag·rouée" capitalized "Ag·rouée") or even in French in some gender inclusive notations ("un·e" capitalized "Un·e").

Some American languages like Kutenai, Oneida, Onondaga, Tuscarora, Tsawwassen Halkomelem and others are typically not capitalized but borrowed words or names may appear in English or other languages that do capitalize. Sierra de Juárez Zapotec also has the grapheme l·l and does capitalize like Catalan.

[r12a]

I wonder whether we can make a general rule that a letter is never capitalised when it follows · word-medially?

[r12a]

This test shows that the various examples provided by @moyogo all work fine with Blink and WebKit browsers. And, if i understood the code change correctly, they will also all work as expected after Gecko is fixed.

[andjc]

It is a long standing issue, and isn't limited to casing in CSS. The interpunct isn't classified as a word forming character, so creates all sorts of interesting issues in many places not just casing, and in many programing languages.