The texts and intros come from Wikisource and Wikipedia (CC BY-SA), so the data cannot be MIT; LICENSE was the MIT
text, which hosts showed as the licence of the whole repository.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
divan/<work url>.dtx (readable source) and .json (generated by the Divan app) override the
Wikisource text in the export; such works are marked DivanOwned with who edited them and when.
Radif, matla and maqta are computed from Divan's verses.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Grouping and order now read the radif from the first couplet's second line, which carries it
even when the source lacks the matla.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Six Mir ghazals were stored as prose, so their couplets were merged into paragraphs.
Classification now also uses the section (غزل، نظم، رباعی…); a fix-up pass reclassifies
stored works on every sync.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- wikisource.py: record each work's position in the author page list (ord), refreshed on
every sync; children of index pages keep their parent's position
- export: walk works by ord; books in order of first appearance (Iqbal: Bang-e-Dra,
Bal-e-Jibril, Zarb-e-Kalim, Armaghan-e-Hijaz)
- ghazal sections in divan order: last letter of the radif (opening line), alif..ye;
ties keep source order; Iqbal keeps his books' published order
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- author pages linking the same article (after redirects) become one poet; aliases kept in
poet_aliases so later syncs map new works to the canonical poet
- normalise ':ur:X' Wikipedia links
- result: مرزا اسد اللہ خان غالب merged into مرزا غالب (316 works); 383 poets
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
