[Fixes #14097] Label thesaurus is not reloaded on thesaurus changes (#14098)

* [Fixes #14097] Label thesaurus is not reloaded on thesaurus changes

* Fix thread safety: guard set(), clear(), force_check() with _lock

Agent-Logs-Url: https://github.com/GeoNode/geonode/sessions/26a81ebe-aef6-4850-b356-885f6bdce0dc

Co-authored-by: etj <717359+etj@users.noreply.github.com>

* Add test for stale cache invalidation on thesaurus update

Agent-Logs-Url: https://github.com/GeoNode/geonode/sessions/cbf02dc8-5fb7-402a-9ab5-9f58f1a5378e

Co-authored-by: etj <717359+etj@users.noreply.github.com>

* Apply black formatting to test_i18n.py

Agent-Logs-Url: https://github.com/GeoNode/geonode/sessions/73015399-e501-43f4-8387-7807b3257567

Co-authored-by: etj <717359+etj@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: etj <717359+etj@users.noreply.github.com>

Author: etj

geonode/base/i18n.py

@@ -115,7 +115,8 @@ def get_entry(self, lang, data_key):
                     .first()
                 )
                 if cached_entry and cached_entry.date != thesaurus_date:
-                    logger.info(f"Cache for {lang}:{data_key} needs to be recreated")
+                    logger.info(f"Cache for {lang}:{data_key} dirty, clearing all caches")
+                    self.lang_cache.clear()
                     return thesaurus_date, None
                 if not cached_entry:
                     logger.info(f"Cache for {lang}:{data_key} needs to be created")
@@ -126,31 +127,35 @@ def get_entry(self, lang, data_key):
 
     def set(self, lang: str, data_key: str, data, request_date: str):
         # TODO: check if lang is allowed
-        cached_entry: I18nCacheEntry = self.lang_cache.setdefault(lang, I18nCacheEntry())
-
+        # Perform DB query outside the lock to avoid holding it during I/O
         latest_date = (
             Thesaurus.objects.filter(identifier=I18N_THESAURUS_IDENTIFIER).values_list("date", flat=True).first()
         )
 
-        if request_date == latest_date:
-            # no changes after processing, set the info right away
-            logger.debug(f"Caching lang:{lang} key:{data_key} date:{request_date}")
-            cached_entry.date = latest_date
-            cached_entry.caches[data_key] = data
-            return True
-        else:
-            logger.warning(
-                f"Cache will not be updated for lang:{lang} key:{data_key} reqdate:{request_date} latest:{latest_date}"
-            )
-            return False
+        with self._lock:
+            cached_entry: I18nCacheEntry = self.lang_cache.setdefault(lang, I18nCacheEntry())
+
+            if request_date == latest_date:
+                # no changes after processing, set the info right away
+                logger.debug(f"Caching lang:{lang} key:{data_key} date:{request_date}")
+                cached_entry.date = latest_date
+                cached_entry.caches[data_key] = data
+                return True
+            else:
+                logger.warning(
+                    f"Cache will not be updated for lang:{lang} key:{data_key} reqdate:{request_date} latest:{latest_date}"
+                )
+                return False
 
     def clear(self):
         logger.info("Clearing i18n cache")
-        self.lang_cache.clear()
+        with self._lock:
+            self.lang_cache.clear()
 
     def force_check(self):
         """For testing: forces a check against the DB on the next get_entry call."""
-        self._last_check = 0
+        with self._lock:
+            self._last_check = 0
 
 
 class LabelResolver:

geonode/metadata/tests/test_i18n.py

@@ -23,7 +23,7 @@
 from geonode.metadata.handlers.sparse import SparseHandler, SparseFieldRegistry
 from geonode.metadata.manager import MetadataManager
 
-from geonode.base.i18n import I18N_THESAURUS_IDENTIFIER, i18nCache
+from geonode.base.i18n import I18N_THESAURUS_IDENTIFIER, i18nCache, labelResolver
 from geonode.base.models import (
     ThesaurusKeyword,
     ThesaurusKeywordLabel,
@@ -147,3 +147,39 @@ def test_schema_i18n_title_defined(self):
         self._add_label("field1__ovr", "en", "f1_ovr_en")
         schema = self.mm.build_schema(lang="en")
         self.assertEqual("f1_ovr_en", schema["properties"]["field1"]["title"])
+
+    def test_stale_cache_invalidated_on_thesaurus_update(self):
+        """
+        Ensure that all language caches (en and it) are invalidated when the Thesaurus date
+        changes, so that stale values are never served after a thesaurus update.
+        """
+        # Populate labels for two languages and warm up the cache
+        self._add_label("key1", "en", "key1_en_v1")
+        self._add_label("key1", "it", "key1_it_v1")
+
+        labels_en = labelResolver.get_labels("en")
+        labels_it = labelResolver.get_labels("it")
+
+        self.assertEqual("key1_en_v1", labels_en.get("key1"))
+        self.assertEqual("key1_it_v1", labels_it.get("key1"))
+
+        # Update label values in the DB to simulate a thesaurus edit
+        ThesaurusKeywordLabel.objects.filter(keyword__thesaurus_id=self.tid, keyword__about="key1", lang="en").update(
+            label="key1_en_v2"
+        )
+        ThesaurusKeywordLabel.objects.filter(keyword__thesaurus_id=self.tid, keyword__about="key1", lang="it").update(
+            label="key1_it_v2"
+        )
+
+        # Simulate a thesaurus date bump (what happens when the thesaurus is updated)
+        Thesaurus.objects.filter(id=self.tid).update(date="2024-01-01")
+
+        # Force cache freshness check to bypass CHECK_INTERVAL
+        i18nCache.force_check()
+
+        # Both language caches must be rebuilt with the new values
+        labels_en = labelResolver.get_labels("en")
+        labels_it = labelResolver.get_labels("it")
+
+        self.assertEqual("key1_en_v2", labels_en.get("key1"))
+        self.assertEqual("key1_it_v2", labels_it.get("key1"))