feat: clean unused glossary languages (#12331)

* cleanup stale glossaries translation delete and glossary languages sync

* changelog update

* test fix

* Update weblate/glossary/tasks.py

Co-authored-by: Michal Čihař <michal@cihar.com>

* update glossary.tasks.cleanup_stale_glossaries

* sentence correction

* fix missing argument TypeError

* only cleanup glossaries on translation delete

* update tests

* cleanup task trigger update

* adjust test to changed condition

* Update weblate/glossary/tasks.py

Co-authored-by: Michal Čihař <michal@cihar.com>

---------

Co-authored-by: Michal Čihař <michal@cihar.com>
Co-authored-by: Benjamin Alan Jamie <benjamin@weblate.org>

docs/changes.rst

@@ -19,6 +19,7 @@ Not yet released.
 * :ref:`autofix` for Devanagari danda now better handles latin script.
 * :ref:`autofix` for French and Breton now uses a non-breaking space before colons instead of a narrow one.
 * :ref:`api` now has a preview OpenAPI specification.
+* Stale, empty glossaries are now automatically removed.
 * :kbd:`?` now displays available :ref:`keyboard`.
 * Translation and language view in the project now include basic information about the language and plurals.
 * :ref:`bulk-edit` shows a preview of matched strings.

weblate/api/tests.py

@@ -3327,7 +3327,11 @@ def test_add_plural(self) -> None:
         )
 
     def test_delete(self) -> None:
-        start_count = Translation.objects.count()
+        def _translation_count():
+            # exclude glossaries because stale glossaries are also cleaned out
+            return Translation.objects.filter(component__is_glossary=False).count()
+
+        start_count = _translation_count()
         self.do_request(
             "api:translation-detail", self.translation_kwargs, method="delete", code=403
         )
@@ -3338,7 +3342,8 @@ def test_delete(self) -> None:
             superuser=True,
             code=204,
         )
-        self.assertEqual(Translation.objects.count(), start_count - 1)
+
+        self.assertEqual(_translation_count(), start_count - 1)
 
 
 class UnitAPITest(APIBaseTest):

weblate/glossary/tasks.py

@@ -4,10 +4,15 @@
 
 from __future__ import annotations
 
+from django.db import transaction
+from django.db.models import F
+
+from weblate.auth.models import get_anonymous
 from weblate.lang.models import Language
-from weblate.trans.models import Component
+from weblate.trans.models import Component, Project, Translation
 from weblate.utils.celery import app
 from weblate.utils.lock import WeblateLockTimeoutError
+from weblate.utils.stats import prefetch_stats
 
 
 @app.task(
@@ -40,6 +45,66 @@ def sync_glossary_languages(pk: int, component: Component | None = None) -> None
         component.create_translations_task()
 
 
+@app.task(trail=False, autoretry_for=(Project.DoesNotExist, WeblateLockTimeoutError))
+def cleanup_stale_glossaries(project: int | Project) -> None:
+    """
+    Delete stale glossaries.
+
+    A glossary translation is considered stale when it meets the following conditions:
+    - glossary.language is not used in any other non-glossary components
+    - glossary.language is different from glossary.component.source_language
+    - It has no translation
+
+    Stale glossary is not removed if:
+    - the component only has one glossary component
+    - if is managed outside weblate (i.e repo != 'local:')
+    """
+    if isinstance(project, int):
+        project = Project.objects.get(pk=project)
+
+    languages_in_non_glossary_components: set[int] = set(
+        Translation.objects.filter(
+            component__project=project, component__is_glossary=False
+        ).values_list("language_id", flat=True)
+    )
+
+    glossary_translations = prefetch_stats(
+        Translation.objects.filter(
+            component__project=project, component__is_glossary=True
+        )
+        .prefetch()
+        .exclude(language__id__in=languages_in_non_glossary_components)
+        .exclude(language=F("component__source_language"))
+    )
+
+    component_to_check = []
+
+    def can_delete(_glossary: Translation) -> bool:
+        """
+        Check if a glossary can be deleted.
+
+        It is possible to delete a glossary if:
+        - it has no translations
+        - it is not the only glossary in the project
+        - it is managed by Weblate (i.e. repo == 'local:')
+        """
+        return all(
+            [
+                _glossary.stats.translated == 0,
+                _glossary.component.repo == "local:",
+            ]
+        )
+
+    for glossary in glossary_translations:
+        if can_delete(glossary):
+            glossary.remove(get_anonymous())
+            if glossary.component not in component_to_check:
+                component_to_check.append(glossary.component)
+
+    for component in component_to_check:
+        transaction.on_commit(component.schedule_update_checks)
+
+
 @app.task(
     trail=False,
     autoretry_for=(Component.DoesNotExist, WeblateLockTimeoutError),

weblate/glossary/tests.py

@@ -11,7 +11,11 @@
 from django.urls import reverse
 
 from weblate.glossary.models import get_glossary_terms, get_glossary_tsv
-from weblate.glossary.tasks import sync_terminology
+from weblate.glossary.tasks import (
+    cleanup_stale_glossaries,
+    sync_terminology,
+)
+from weblate.lang.models import Language
 from weblate.trans.models import Unit
 from weblate.trans.tests.test_views import ViewTestCase
 from weblate.trans.tests.utils import get_test_file
@@ -472,3 +476,31 @@ def test_tsv(self) -> None:
         lines = list(reader)
         self.assertEqual(len(lines), 163)
         self.assertTrue(all(len(line) == 2 for line in lines))
+
+    def test_stale_glossaries_cleanup(self) -> None:
+        # setup: make glossary managed outside weblate
+        self.glossary_component.repo = "git://example.com/test/project.git"
+        self.glossary_component.save()
+
+        initial_count = self.glossary_component.translation_set.count()
+
+        # check glossary not deleted because it has a valid translation
+        cleanup_stale_glossaries(self.project.id)
+        self.assertEqual(self.glossary_component.translation_set.count(), initial_count)
+
+        # delete translation: should trigger cleanup_stale_glossary task
+        german = Language.objects.get(code="de")
+        self.component.translation_set.get(language=german).remove(self.user)
+
+        cleanup_stale_glossaries(self.project.id)
+        self.assertEqual(self.glossary_component.translation_set.count(), initial_count)
+
+        # make glossary managed by weblate
+        self.glossary_component.repo = "local:"
+        self.glossary_component.save()
+
+        # check that one glossary has been deleted
+        cleanup_stale_glossaries(self.project.id)
+        self.assertEqual(
+            self.glossary_component.translation_set.count(), initial_count - 1
+        )

weblate/trans/models/translation.py

@@ -1360,7 +1360,9 @@ def get_export_url(self):
         return self.component.get_export_url()
 
     def remove(self, user: User) -> None:
-        """Remove translation from the VCS."""
+        """Remove translation from the Database and VCS."""
+        from weblate.glossary.tasks import cleanup_stale_glossaries
+
         author = user.get_author_name()
         # Log
         self.log_info("removing %s as %s", self.filenames, author)
@@ -1389,6 +1391,8 @@ def remove(self, user: User) -> None:
             user=user,
             author=user,
         )
+        if not self.component.is_glossary:
+            cleanup_stale_glossaries.delay(self.component.project.id)
 
     def handle_store_change(
         self,