Previous to this change, repositories were looked up unfiltered in six different queries, and then filtered using the permissions model, which issued a query per repository found, making search incredibly slow. Instead, we now lookup a chunk of repositories unfiltered and then filter them via a single query to the database. By layering the filtering on top of the lookup, each as queries, we can minimize the number of queries necessary, without (at the same time) using a super expensive join.
Other changes:
- Remove the 5 page pre-lookup on V1 search and simply return that there is one more page available, until there isn't. While technically not correct, it is much more efficient, and no one should be using pagination with V1 search anyway.
- Remove the lookup for repos without entries in the RAC table. Instead, we now add a new RAC entry when the repository is created for *the day before*, with count 0, so that it is immediately searchable
- Remove lookup of results with a matching namespace; these aren't very relevant anyway, and it overly complicates sorting
Adds code to ensure we never GC CAS paths that are shared amongst multiple ImageStorage rows, as well as an associated pair of tests to catch the positive and negative cases.
Gitlab doesn't send any commit information for tagging events (because... reasons), and so we have to perform the lookup ourselves to have full metadata.
Fixes#1467
This API is still (apparently) being used by the Docker CLI for `docker search` (why?!) and we therefore have customers expecting this to work the same way as the DockerHub.
With this change, if all entitlements are valid, we sort to show the entitlement that will expire the farthest in the future, as that defines the point at which the user must act before the license becomes invalid.
The `> 0` check fails if the code was found first in the query string, which can occasionally happen under tox due to the `PYTHONHASHSEED` var changing. We simply change to use a proper parse and check to avoid this issue entirely.
