vicrodh c21a9393a7 fix(text): decode HTML entities in API prose — including malformed bare forms ...

Qobuz prose fields carry HTML entities the plain-text Slint renderer
showed literally — the artist bio credit rendered '&copy John Book
/TiVo' (the API sample itself ends '&copy  Mariano Prunes /TiVo': a
malformed entity with NO semicolon, which the old decoder never
matched). New qbz_text_utils::decode_html_entities: ~90 named
entities, full numeric dec/hex with cp1252 quirk range, and a
word-boundary tolerance for the bare no-semicolon forms restricted to
copy/reg/amp/nbsp (documented tradeoff; 'AC&DC' never decodes). Wired
INTO the existing strip_html pipeline so every treated funnel (album/
label/playlist descriptions, bios) inherits it, plus four new direct
sites: bio source credit, magazine story title/excerpt, and the
track/album copyright lines in the info modals. 19 unit tests.

cc

2026-07-08 11:08:01 -06:00

..

src

fix(text): decode HTML entities in API prose — including malformed bare forms

2026-07-08 11:08:01 -06:00

Cargo.toml

refactor(slint): extract strip_html + dates into qbz-text-utils crate

2026-06-25 18:13:55 -06:00