Skip to content

Migrate Italian and Polish to LibreOffice sources - #6

Merged
fuadsaud merged 2 commits into
en-us-2026from
orphan-langs-to-lo
Jun 24, 2026
Merged

Migrate Italian and Polish to LibreOffice sources#6
fuadsaud merged 2 commits into
en-us-2026from
orphan-langs-to-lo

Conversation

@fuadsaud

Copy link
Copy Markdown

Stacked on #5 — depends on the patches mechanism from that PR.

Summary

  • Italian (it) — re-enable the language. The previous source (PLIO) has been offline for years, so the shipped dictionary was a multi-year-old frozen file. Switch to LibreOffice's actively maintained `it_IT` build (LibreItalia, Marina Latini, GPL-3.0).
  • Polish (pl) — switch from the stale OpenOffice extension on a SourceForge mirror to the LibreOffice-bundled `pl_PL` build (~16% larger vocabulary, license expression unchanged).
  • Both share the existing `libreoffice` crawl entry, so no new download paths are introduced. The unused `crawl "polish"` entry is dropped.

Patch documentation cleanup

Each `.sed` patch file now leads with a comment describing the class of problem it addresses, not the specific instance that motivated it. The Italian apostrophe-ICONV patch is also re-anchored on the normalized `SET UTF-8` line so it survives any future Italian upstream the pipeline normalizes through `iconv`.

Languages considered but not migrated

When I actually pulled side-by-side dictionary content from LibreOffice for the other "orphan-ish" languages, the picture changed:

Lang Current LO Result
nb 334,170 lines 334,170 lines identical — LO is downstream of the same dead alioth source
nn 234,330 lines 234,330 lines identical — same reason
cs 260,461 lines 261,168 lines +0.3% — LO redistributes the same maintainer's (translatoblog.cz) work

Skipped those: a same-content URL move without a real upstream quality argument isn't worth the churn.

fuadsaud added 2 commits June 24, 2026 20:05
Each .sed patch file now leads with a comment describing the
class of problem it addresses — the harm path that justifies
the patch — rather than the specific instance that motivated it.

Also generalize the Italian apostrophe ICONV patch to anchor on
the normalized SET UTF-8 line instead of a source-specific
#Nota comment. The original anchor only matched the (now
offline) PLIO source; this one survives any future upstream
that produces a SET line, which the crawl pipeline always
normalizes.
Italian (it): the previous source (PLIO via SourceForge OXT)
has been offline for years and the crawl entry was commented
out, so the shipped dictionary was a multi-year-old frozen
file. Re-enable the language using LibreOffice's it_IT, which
is the actively maintained Italian Hunspell dictionary by
LibreItalia (Marina Latini), license GPL-3.0.

Polish (pl): the previous source (OpenOffice extension on a
SourceForge mirror, 2017-era) has been superseded by the
LibreOffice-bundled pl_PL build maintained from sjp.pl —
~16% larger vocabulary, license unchanged.

Both languages now share the existing libreoffice crawl entry
that we already use for de, ru, pt and sv. The standalone
crawl "polish" entry is removed since nothing references it
anymore.
@fuadsaud
fuadsaud merged commit dd79711 into en-us-2026 Jun 24, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant