This splits translated code blocks into individual messages, thus
preserving the translation after #1460.
This removes ~6k unnecessary lines from the PO file.
This updates each PO file to the date when it was last touched:
for po_file in po/*.po; do
POT_DATE=$(git log -1 --date=short --format=%ad $po_file)
sed -i -e 's/"POT-Creation-Date:.*"/"POT-Creation-Date:
'$POT_DATE'\\n"/' $po_file
done
The dates will be updated with `msgmerge` in the future via
https://github.com/google/mdbook-i18n-helpers/pull/87.
The dates here can be adjusted by hand if needed — the starting point
here only serves as a rough anchor to let us freeze the translations
from further updates. So if you know that you ran `msgmerge` most
recently on a given date, please update the file to that date in a later
PR.
Before, po/da.po had these statistics:
279 translated messages, 26 fuzzy translations, 1476 untranslated
messages.
Afterwards, the statistics for po/da.po is:
244 translated messages, 22 fuzzy translations, 2201 untranslated
messages.
The number of translated messages changed from 16% to 10%.
With this change, it becomes important to use the latest version of
mdbook-i18n-helpers when viewing the translation locally. To update to
the latest version, run
cargo install mdbook-i18n-helpers
You will now be able to serve the translation locally.
Part of #330.
* da: Sync translation
This resulted in some rewrapped strings due to how `msgmerge` wraps
the strings compared to how `mdbook-xgettext` extracts them.
I also translated a few strings.
* Apply suggestions from code review
This copies a bunch of untranslatable page titles (such as crate
names) to the translation. I also translated a lot of small and fuzzy
entries. We should now be at 200 translated entries.
da: sync with c45b350
This is the result of running
MDBOOK_OUTPUT='{"xgettext": {"pot-file": "messages.pot"}}' mdbook build -d po
msgmerge --update po/da.po po/messages.pot
* Parse Markdown to support translation.
This upgrades from just splitting Markdown files on double-newlines, to
using a Markdown parser to break them into more appropriate chunks. The
upshot is that code samples are all in one message, lists are bundled
together, and generally it should be easier to translate.
* [WIP] helper to update po files for new translation
* process synthetic input file-by-file
* review comments
* remove temporary code
* fix msgfmt lints
