feat: use corpus from cache for typo checking (#17812)

* fix: handle missing keys in cache

Don't try to deserialize is nothing came back.

Signed-off-by: Mike Fiedler <[email protected]>

* feat: use corpus from cache for typo checking

Signed-off-by: Mike Fiedler <[email protected]>

---------

Signed-off-by: Mike Fiedler <[email protected]>

Author: miketheman

pyproject.toml

@@ -82,7 +82,7 @@ ignore_missing_imports = true
 [tool.pytest.ini_options]
 addopts = [
   "--disable-socket",
-  "--allow-hosts=localhost,::1,stripe",
+  "--allow-hosts=localhost,::1,stripe,redis",
   "--durations=20",
   "--numprocesses=auto",
   # Disable ddtrace for tests

tests/conftest.py

@@ -320,7 +320,7 @@ def get_app_config(database, nondefaults=None):
         "database.url": database,
         "docs.url": "http://docs.example.com/",
         "ratelimit.url": "memory://",
-        "db_results_cache.url": "redis://localhost:0/",
+        "db_results_cache.url": "redis://redis:0/",
         "opensearch.url": "https://localhost/warehouse",
         "files.backend": "warehouse.packaging.services.LocalFileStorage",
         "archive_files.backend": "warehouse.packaging.services.LocalArchiveFileStorage",

tests/unit/cache/test_services.py

@@ -34,6 +34,12 @@ def test_create_service(self):
 
         assert isinstance(service, RedisQueryResults)
 
+    def test_get_missing(self, query_results_cache_service):
+        # Attempt to get a value that doesn't exist in the cache
+        result = query_results_cache_service.get("missing_key")
+
+        assert result is None
+
     def test_set_get_simple(self, query_results_cache_service):
         # Set a value in the cache
         query_results_cache_service.set("test_key", {"foo": "bar"})

warehouse/cache/services.py

@@ -49,7 +49,7 @@ def get(self, key: str) -> list | dict | None:
         """Get a cached result by key."""
         result = self.redis_client.get(key)
         # deserialize the value as a JSON object
-        return orjson.loads(result)
+        return orjson.loads(result) if result else None
 
     def set(self, key: str, value) -> None:
         """Set a cached result by key."""

warehouse/packaging/services.py

@@ -34,6 +34,7 @@
 from zope.interface import implementer
 
 from warehouse.admin.flags import AdminFlagValue
+from warehouse.cache import IQueryResultsCache
 from warehouse.email import send_pending_trusted_publisher_invalidated_email
 from warehouse.events.tags import EventTag
 from warehouse.helpdesk.interfaces import IAdminNotificationService
@@ -411,13 +412,18 @@ def create_service(cls, context, request):
 
 @implementer(IProjectService)
 class ProjectService:
-    def __init__(self, session, metrics=None, ratelimiters=None) -> None:
+    def __init__(
+        self, session, metrics=None, ratelimiters=None, query_results_cache=None
+    ) -> None:
         if ratelimiters is None:
             ratelimiters = {}
+        if query_results_cache is None:
+            query_results_cache = {}
 
         self.db = session
         self.ratelimiters = collections.defaultdict(DummyRateLimiter, ratelimiters)
         self._metrics = metrics
+        self._query_results_cache = query_results_cache
 
     def _check_ratelimits(self, request, creator):
         # First we want to check if a single IP is exceeding our rate limiter.
@@ -486,7 +492,10 @@ def check_project_name(self, name: str) -> None:
             raise ProjectNameUnavailableSimilarError(similar_project_name)
 
         # Check for typo-squatting.
-        if typo_check_match := typo_check_name(canonicalize_name(name)):
+        cached_corpus = self._query_results_cache.get("top_dependents_corpus")
+        if typo_check_match := typo_check_name(
+            canonicalize_name(name), corpus=cached_corpus
+        ):
             raise ProjectNameUnavailableTypoSquattingError(
                 check_name=typo_check_match[0],
                 existing_project_name=typo_check_match[1],
@@ -718,4 +727,10 @@ def project_service_factory(context, request):
             IRateLimiter, name="project.create.ip", context=None
         ),
     }
-    return ProjectService(request.db, metrics=metrics, ratelimiters=ratelimiters)
+    query_results_cache = request.find_service(IQueryResultsCache)
+    return ProjectService(
+        request.db,
+        metrics=metrics,
+        ratelimiters=ratelimiters,
+        query_results_cache=query_results_cache,
+    )