Garbage collect repositories on push and on tag deletion.

data/model.py

@@ -973,6 +973,39 @@ def delete_tag_and_images(namespace_name, repository_name, tag_name):
     store.remove(repository_path)
 
 
+def garbage_collect_repository(namespace_name, repository_name):
+  # Get a list of all images used by tags in the repository
+  tag_query = (RepositoryTag
+    .select(RepositoryTag, Image)
+    .join(Image)
+    .switch(RepositoryTag)
+    .join(Repository)
+    .where(Repository.name == repository_name,
+           Repository.namespace == namespace_name))
+
+  referenced_anscestors = set()
+  for tag in tag_query:
+    ancestor_list = [int(img) for img in tag.image.ancestors.split('/')[1:-1]]
+    referenced_anscestors = referenced_anscestors.union(set(ancestor_list))
+    referenced_anscestors.add(tag.image.id)
+
+  all_repo_images = get_repository_images(namespace_name, repository_name)
+  all_images = {int(img.id):img for img in all_repo_images}
+  to_remove = set(all_images.keys()).difference(referenced_anscestors)
+
+  logger.info('Cleaning up unreferenced images: %s', to_remove)
+
+  for image_id_to_remove in to_remove:
+    image_to_remove = all_images[image_id_to_remove]
+    image_path = store.image_path(namespace_name, repository_name,
+                                       image_to_remove.docker_image_id)
+    logger.debug('Recursively deleting image path: %s' % image_path)
+    image_to_remove.delete_instance()
+    store.remove(image_path)
+
+  return len(to_remove)
+
+
 def get_tag_image(namespace_name, repository_name, tag_name):
   joined = Image.select().join(RepositoryTag).join(Repository)
   fetched = list(joined.where(Repository.name == repository_name,