(superset) 01/01: fix(i18n): compile fuzzy translations into the backend .mo files

[ru...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

rusackas pushed a commit to branch fix/i18n-serve-fuzzy-backend
in repository https://gitbox.apache.org/repos/asf/superset.git

commit 7756ca84da3482370896c5e377e9412a4b21d7f8
Author: Amin Ghadersohi <am...@gmail.com>
AuthorDate: Wed Jul 1 17:10:27 2026 -0700

    fix(i18n): compile fuzzy translations into the backend .mo files
    
    The main Dockerfile compiled backend translations with a bare `pybabel compile`,
    which omits `#, fuzzy` entries — so fuzzy (machine-generated and other draft)
    translations fell back to English on the backend. This was inconsistent with
    every other translation build path in the repo:
    
    - the frontend build (`po2json --fuzzy`) includes fuzzy entries;
    - `flask fab babel-compile` (used by the RELEASING images and
      scripts/translations/generate_mo_files.sh) runs `pybabel compile -f`.
    
    Add `--use-fuzzy` to the Dockerfile's `pybabel compile` so the production image
    serves fuzzy translations consistently with the frontend and the other backend
    build paths. This makes the AI backfill effort (and any other draft
    translations) visible in the UI rather than falling back to English while they
    await native-speaker review. Docs updated to match.
    
    Co-Authored-By: Claude Opus 4.8 <no...@anthropic.com>
---
 Dockerfile                                 |  2 +-
 docs/developer_docs/contributing/howtos.md | 17 +++++++++++++++--
 2 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index ecdbb97c751..14d409e0207 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -141,7 +141,7 @@ RUN --mount=type=cache,target=/root/.cache/uv \
 
 COPY superset/translations/ /app/translations_mo/
 RUN if [ "${BUILD_TRANSLATIONS}" = "true" ]; then \
-        pybabel compile -d /app/translations_mo | true; \
+        pybabel compile --use-fuzzy -d /app/translations_mo | true; \
     fi; \
     rm -f /app/translations_mo/*/*/*.[po,json]
 
diff --git a/docs/developer_docs/contributing/howtos.md b/docs/developer_docs/contributing/howtos.md
index 405a6311c35..452f705ae57 100644
--- a/docs/developer_docs/contributing/howtos.md
+++ b/docs/developer_docs/contributing/howtos.md
@@ -332,15 +332,28 @@ cd superset-frontend
 npm run build-translation
 
 # Backend
-pybabel compile -d superset/translations
+pybabel compile --use-fuzzy -d superset/translations
 ```
 
+`--use-fuzzy` includes `#, fuzzy` entries in the compiled `.mo` files. Superset
+serves fuzzy translations on purpose: the frontend build (`po2json --fuzzy`)
+already includes them, `flask fab babel-compile` (used by the release images)
+compiles with `-f`, and the production `Dockerfile` compiles with `--use-fuzzy`
+as well. This keeps machine-generated (and other draft) translations visible in
+the UI rather than falling back to English while they await review.
+
 ### Backfilling missing translations with AI
 
 For languages with many untranslated strings, the repo includes a script that
 uses Claude AI to generate draft translations for any missing entries. All
 AI-generated strings are marked `#, fuzzy` and tagged with an attribution
-comment so that human reviewers know they need to be checked before merging.
+comment so that human reviewers know they need to be checked.
+
+Note that `#, fuzzy` marks a translation as *needing review*, not as *withheld*:
+both the frontend and backend builds serve fuzzy entries (see [Applying
+translations](#applying-translations) above), so an AI-generated string is shown
+in the UI as soon as it is built and deployed. Reviewers should verify each
+entry and remove the `#, fuzzy` flag to promote it to a confirmed translation.
 
 #### Prerequisites