Importing translations¶

Note

Please make sure that amaGama is installed and that the source language is already initialized in the database with the initdb command.

To populate the amaGama database the amagama-manage command build_tmdb subcommand should be used:

$ amagama-manage build_tmdb --verbose -s en -t ar -i foo.po
Importing foo.po
Imported 1 files: 1000 units (100 skipped).

This will parse foo.po, assuming that source language is English (en) and target language is Arabic (ar), and will populate the database accordingly.

The source and target language options only need to be specified if the file does not provide this information. But if source and target language options are specified they will override the languages metadata in the translation file.

All bilingual formats supported by the Translate Toolkit are supported, including PO, TMX and XLIFF. An XLIFF unit counts as translated when it is approved="yes" or its target’s state is translated, final or signed-off. A large PO, TMX or XLIFF file is read and committed in pieces, so memory doesn’t limit its size.

If a directory is passed to the -i option, then its content will be read recursively:

$ amagama-manage build_tmdb --verbose -s en -t gl -i translations/
Importing translations/foo.po
Importing translations/bar.po
Imported 2 files: 5000 units (100 skipped).

A file that can’t be imported (unreadable, or with no target language) is left out and listed at the end with the reason, and the command exits with status 1:

$ amagama-manage build_tmdb -s en -i translations/
Imported 1 files: 1000 units (100 skipped).
Couldn't import 1 files:
  translations/notes.po: missing source or target language

Recording imports¶

Every imported file is recorded in the imports table of its source language: the file name, its SHA-256 checksum (none for a snapshot), the target language, how many units were loaded and how many were skipped (untranslated or fuzzy), and when. Name the set of files with --block and give their licence as an SPDX licence expression with --licence, such as MPL-2.0 OR LGPL-3.0-or-later. Identifiers must be current ones from the SPDX License List (GPL-2.0-or-later, not the deprecated GPL-2.0+) and are stored in their normal case; anything else stops the import before anything is loaded:

$ amagama-manage build_tmdb -s en -i gnome-51/ --block gnome-51 \
    --licence GPL-2.0-or-later

tmdb_stats summarises the imports by block, and lookups can keep only translations under chosen licences with licence.

Each translation is linked to every import that supplied it, in the target_imports table, so a lookup lists its blocks in sources. A translation found again in a later import gets another link, not another copy. A loaded snapshot links its translations to the snapshot’s own import. A schema made before the links existed still serves, with empty sources.