From ef68982728a670652d6d95736c90c275064d51ec Mon Sep 17 00:00:00 2001
From: yackob03 <jacob.moshenko@gmail.com>
Date: Thu, 6 Feb 2014 14:13:35 -0500
Subject: [PATCH] Garbage collect repositories on push and on tag deletion.

---
 data/model.py      | 33 +++++++++++++++++++++++++++++++++
 endpoints/index.py |  5 ++++-
 endpoints/tags.py  | 16 +---------------
 3 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/data/model.py b/data/model.py
index c047e1602..336fc1f32 100644
--- a/data/model.py
+++ b/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,
diff --git a/endpoints/index.py b/endpoints/index.py
index 998a0f94d..41064d805 100644
--- a/endpoints/index.py
+++ b/endpoints/index.py
@@ -222,6 +222,8 @@ def update_images(namespace, repository):
       updated_tags[image['Tag']] = image['id']
       model.set_image_checksum(image['id'], repo, image['checksum'])
 
+    num_removed = model.garbage_collect_repository(namespace, repository)
+
     # Generate a job for each webhook that has been added to this repo
     webhooks = model.list_webhooks(namespace, repository)
     for webhook in webhooks:
@@ -237,7 +239,8 @@ def update_images(namespace, repository):
         'homepage': 'https://quay.io/repository/%s' % repo_string,
         'visibility': repo.visibility.name,
         'updated_tags': updated_tags,
-        'pushed_image_count': len(image_with_checksums), 
+        'pushed_image_count': len(image_with_checksums),
+        'pruned_image_count': num_removed,
       }
       webhook_queue.put(json.dumps(webhook_data))
 
diff --git a/endpoints/tags.py b/endpoints/tags.py
index f6b0e1163..cbfefda9f 100644
--- a/endpoints/tags.py
+++ b/endpoints/tags.py
@@ -73,21 +73,7 @@ def delete_tag(namespace, repository, tag):
 
   if permission.can():
     model.delete_tag(namespace, repository, tag)
-
-    return make_response('Deleted', 204)
-
-  abort(403)
-
-
-@tags.route('/repositories/<path:repository>/tags',
-           methods=['DELETE'])
-@process_auth
-@parse_repository_name
-def delete_repository_tags(namespace, repository):
-  permission = ModifyRepositoryPermission(namespace, repository)
-
-  if permission.can():
-    model.delete_all_repository_tags(namespace, repository)
+    model.garbage_collect_repository(namespace, repository)
 
     return make_response('Deleted', 204)