Line breaking for Chinese, Japanese, Korean and Thai, in the browser and in WeasyPrint #1052

New issue

Closed

opened 2026-10-25 16:29:13 +00:00 by tiagoagueda · 0 comments

tiagoagueda commented

2026-10-25 16:29:13 +00:00

Owner

Spun out of #269 (its "From #71" point 4).

Chinese (zh-Hans, zh-Hant) is selected for 0.5, and Japanese, Korean and Thai are in the ordering for later releases. None of these scripts puts spaces between words, so line breaking cannot rely on spaces.

Chinese and Japanese: a line may break between almost any two characters. The breaks are governed by kinsoku rules: no closing punctuation such as ，。）」 at the start of a line, and no opening punctuation such as （「 at its end.
Korean: uses spaces, but browsers default to breaking inside words (word-break: normal treats Hangul like CJK). Korean text usually wants word-break: keep-all.
Thai: has no spaces and no character-level breaks; lines may break only at word boundaries, which needs a dictionary.

What to check

Browser: tables, cards, chips and button labels should wrap at legal break points and never overflow. The rules that keep long Latin words from overflowing (overflow-wrap: anywhere, wrap-anywhere) are fine for CJK but wrong for Korean.
WeasyPrint (CVs, letters, portfolios): Pango breaks lines itself and handles CJK and Thai via ICU/libthai. Check that the image's Pango build actually has Thai word breaking. Check that hyphens: auto and text-align: justify in the themes do nothing harmful to CJK.

Done when

A browser test renders the review page, a table and a CV preview in zh-Hans and th at 320 px wide, with no horizontal scroll and no line starting with closing punctuation.
A test renders a CV in zh-Hans and th through WeasyPrint and checks that no line of the text layer starts with closing punctuation, and that Thai breaks only between words.
ko uses word-break: keep-all (:lang(ko)).

Spun out of #269 (its "From #71" point 4). Chinese (`zh-Hans`, `zh-Hant`) is selected for 0.5, and Japanese, Korean and Thai are in the ordering for later releases. None of these scripts puts spaces between words, so line breaking cannot rely on spaces. - **Chinese and Japanese:** a line may break between almost any two characters. The breaks are governed by kinsoku rules: no closing punctuation such as `，。）」` at the start of a line, and no opening punctuation such as `（「` at its end. - **Korean:** uses spaces, but browsers default to breaking inside words (`word-break: normal` treats Hangul like CJK). Korean text usually wants `word-break: keep-all`. - **Thai:** has no spaces and no character-level breaks; lines may break only at word boundaries, which needs a dictionary. ## What to check - **Browser:** tables, cards, chips and button labels should wrap at legal break points and never overflow. The rules that keep long Latin words from overflowing (`overflow-wrap: anywhere`, `wrap-anywhere`) are fine for CJK but wrong for Korean. - **WeasyPrint (CVs, letters, portfolios):** Pango breaks lines itself and handles CJK and Thai via ICU/libthai. Check that the image's Pango build actually has Thai word breaking. Check that `hyphens: auto` and `text-align: justify` in the themes do nothing harmful to CJK. ## Done when - A browser test renders the review page, a table and a CV preview in `zh-Hans` and `th` at 320 px wide, with no horizontal scroll and no line starting with closing punctuation. - A test renders a CV in `zh-Hans` and `th` through WeasyPrint and checks that no line of the text layer starts with closing punctuation, and that Thai breaks only between words. - `ko` uses `word-break: keep-all` (`:lang(ko)`).

tiagoagueda referenced this issue

2026-10-25 16:29:35 +00:00

Languages after 0.5: take them up in speaker order (the ordering from #269) #1055

tiagoagueda added the

internationalisation

label

2026-10-25 16:29:36 +00:00

tiagoagueda added this to the 0.5.0 milestone

2026-10-25 17:04:12 +00:00

tiagoagueda referenced this issue

2026-10-27 13:57:11 +00:00

Choose the next language by how many people speak it, not by which continent it is on #269