Engineer to Adapt the Current Zanata Translations Infrastructure to Weblate¶

Summary¶

The OpenStack community is seeking an engineer with system administrator skills (or a system administrator with engineering skills) to adapt the current translations infrastructure that connects gerrit/zuul to a self-hosted Zanata server, to connect instead to a third-party hosted Weblate server.

Business Case¶

The official language of OpenStack is United States English (en-us). End-user-facing strings in both the web UI and in responses from the OpenStack APIs appear by default in en-us. This can be a barrier to non-native English readers who would otherwise like to use an OpenStack cloud. Sponsoring an engineer to adapt the current translations infrastructure from Zanata to Weblate is a way to visibly demonstrate your commitment to the OpenStack community and investment in support for its translation to other languages.

Technical Details¶

There are two Zuul CI jobs that exist today: upstream-translation-update and propose-translation-update [0].

upstream-translation-update operates in a source repository and generates a gettext .pot translations file for that repo then uploads those to Zanata [1]. This is the piece that takes information from the git repos and puts it in Zanata so that translators know what needs to be translated.

propose-translation-update operates in a source repository and fetches .po files containing translated strings for various languages and synchronizes them into the git repositories then pushes the result to Gerrit for review [2].

Unfortunately, the end result here is some spaghetti code. A possibly incomplete list of bits that come into play here: [3][4][5]. Much of this isn’t actually Zanata specific and can be reused. We suspect that the bulk of the work in switching to Weblate is updating these scripts [1][2] to talk to Weblate. Additionally, reorganizing things to be more centralized and easier to maintain would be helpful.

Another approach would be to build new jobs from scratch copying what is useful from the old Zanata stuff. Then we can run both sets of jobs in parallel as part of the transition ; this also has the upside of allowing you to refactor from the start.

[0] https://opendev.org/openstack/project-config/src/commit/b5e236a37ece2a31959142b282cb24186439921e/zuul.d/jobs.yaml#L1168-L1223

[1] https://opendev.org/openstack/openstack-zuul-jobs/src/commit/ab0fad7f5878961884394a4c552256947a106fdc/roles/prepare-zanata-client/files/upstream_translation_update.sh (Note the extract_messages_* functions that are called; these create .pot files for various types of sources.)

[2] https://opendev.org/openstack/openstack-zuul-jobs/src/commit/ab0fad7f5878961884394a4c552256947a106fdc/roles/prepare-zanata-client/files/propose_translation_update.sh

[3] https://opendev.org/openstack/project-config/src/commit/b5e236a37ece2a31959142b282cb24186439921e/playbooks/translation/

[4] https://opendev.org/openstack/project-config/src/commit/b5e236a37ece2a31959142b282cb24186439921e/roles/copy-proposal-common-scripts

[5] https://opendev.org/openstack/openstack-zuul-jobs/src/commit/ab0fad7f5878961884394a4c552256947a106fdc/roles/prepare-zanata-client/