def pipeline_stages(self, new_version):
"""
Build a list of stages feeding into the ContentUnitAssociation stage.
This defines the "architecture" of the entire sync.
Args:
new_version (:class:`~pulpcore.plugin.models.RepositoryVersion`): The
new repository version that is going to be built.
Returns:
list: List of :class:`~pulpcore.plugin.stages.Stage` instances
"""
pipeline = [
self.first_stage,
QueryExistingArtifacts(),
ArtifactDownloader(),
ArtifactSaver(),
QueryExistingContents(),
RpmContentSaver(),
RemoteArtifactSaver(),
]
for dupe_query_dict in self.remove_duplicates:
pipeline.append(RemoveDuplicates(new_version, **dupe_query_dict))
return pipeline
def synchronize(remote_pk, repository_pk):
"""
Sync content from the remote repository.
Create a new version of the repository that is synchronized with the remote.
Args:
remote_pk (str): The remote PK.
repository_pk (str): The repository PK.
Raises:
ValueError: If the remote does not specify a url to sync.
"""
remote = RpmRemote.objects.get(pk=remote_pk)
repository = Repository.objects.get(pk=repository_pk)
dupe_criteria = {'model': Package,
'field_names': ['name', 'epoch', 'version', 'release', 'arch']}
if not remote.url:
raise ValueError(_('A remote must have a url specified to synchronize.'))
log.info(_('Synchronizing: repository={r} remote={p}').format(
r=repository.name, p=remote.name))
download_artifacts = (remote.policy == Remote.IMMEDIATE)
first_stage = RpmFirstStage(remote)
with WorkingDirectory():
with RepositoryVersion.create(repository) as new_version:
loop = asyncio.get_event_loop()
remove_duplicates_stage = RemoveDuplicates(new_version, **dupe_criteria)
stages = [first_stage]
if download_artifacts:
stages.extend([QueryExistingArtifacts(), ArtifactDownloader(), ArtifactSaver()])
stages.extend([
QueryExistingContents(), ErratumContentSaver(), remove_duplicates_stage,
ContentAssociation(new_version), EndStage()
])
pipeline = create_pipeline(stages)
loop.run_until_complete(pipeline)