Delete indexes

Author: stsewd

readthedocs/projects/models.py

@@ -480,6 +480,7 @@ def delete(self, *args, **kwargs):  # pylint: disable=arguments-differ
             args=[(self.doc_path,)],
         )
 
+        # Remove extra resources
         tasks.clean_project_resources(self)
 
         super().delete(*args, **kwargs)

readthedocs/projects/tasks.py

@@ -1621,8 +1621,9 @@ def _sync_imported_files(version, build, changed_files):
     # Remove old HTMLFiles from ElasticSearch
     remove_indexed_files(
         model=HTMLFile,
-        version=version,
-        build=build,
+        project_slug=version.project.slug,
+        version_slug=version.slug,
+        build_id=build,
     )
 
     # Delete SphinxDomain objects from previous versions
@@ -1844,6 +1845,17 @@ def remove_build_storage_paths(paths):
         storage.delete_directory(storage_path)
 
 
+@app.task(queue='web')
+def remove_search_indexes(project_slug, version_slug=None):
+    """Wrapper around `remove_indexed_files to make it a task."""
+    # Remove ES indexes
+    remove_indexed_files(
+        model=HTMLFile,
+        project_slug=project_slug,
+        version_slug=version_slug,
+    )
+
+
 def clean_project_resources(project, version=None):
     """
     Delete all extra resources used by `version` of `project`.
@@ -1853,16 +1865,27 @@ def clean_project_resources(project, version=None):
     - Artifacts from storage.
     - Search indexes from ES.
 
-    :param version: Version instance. if isn't given,
+    :param version: Version instance. If isn't given,
                     all resources of `project` will be deleted.
+
+    .. note::
+       This function is usually called just before deleting project.
+       Make sure to not depend on the project object inside the tasks.
     """
+    # Remove storage paths
     storage_paths = []
     if version:
         version.get_storage_paths()
     else:
         storage_paths = project.get_storage_paths()
     remove_build_storage_paths.delay(storage_paths)
 
+    # Remove indexes
+    remove_search_indexes.delay(
+        project_slug=project.slug,
+        version_slug=version.slug if version else None,
+    )
+
 
 @app.task(queue='web')
 def sync_callback(_, version_pk, commit, build, *args, **kwargs):

readthedocs/search/utils.py

@@ -43,35 +43,41 @@ def index_new_files(model, version, build):
         log.exception('Unable to index a subset of files. Continuing.')
 
 
-def remove_indexed_files(model, version, build):
+def remove_indexed_files(model, project_slug, version_slug=None, build_id=None):
     """
-    Remove files from the version from the search index.
+    Remove files from `version_slug` of `project_slug` from the search index.
 
-    This excludes files from the current build.
+    :param model: Class of the model to be deleted.
+    :param project_slug: Project slug.
+    :param version_slug: Version slug. If isn't given,
+                    all index from `project` are deleted.
+    :param build_id: Build id. If isn't given, all index from `version` are deleted.
     """
 
     if not DEDConfig.autosync_enabled():
         log.info(
             'Autosync disabled, skipping removal from the search index for: %s:%s',
-            version.project.slug,
-            version.slug,
+            project_slug,
+            version_slug,
         )
         return
 
     try:
         document = list(registry.get_documents(models=[model]))[0]
         log.info(
             'Deleting old files from search index for: %s:%s',
-            version.project.slug,
-            version.slug,
+            project_slug,
+            version_slug,
         )
-        (
+        documents = (
             document().search()
-            .filter('term', project=version.project.slug)
-            .filter('term', version=version.slug)
-            .exclude('term', build=build)
-            .delete()
+            .filter('term', project=project_slug)
         )
+        if version_slug:
+            documents.filter('term', version=version_slug)
+        if build_id:
+            documents.exclude('term', build=build_id)
+        documents.delete()
     except Exception:
         log.exception('Unable to delete a subset of files. Continuing.')