async def create_pulp3_content(self):
"""
Create a Pulp 3 Package content for saving it later in a bulk operation.
"""
cr_package = await get_cr_obj(self)
pkg_dict = Package.createrepo_to_dict(cr_package)
return Package(**pkg_dict)
def create_pulp3_content(self):
"""
Create a Pulp 3 Package content for saving it later in a bulk operation.
"""
cr_package = get_cr_obj(self)
pkg_dict = Package.createrepo_to_dict(cr_package)
pkg_dict['is_modular'] = self.is_modular
return (Package(**pkg_dict), None)
def deferred_validate(self, data):
"""
Validate the rpm package data.
Args:
data (dict): Data to be validated
Returns:
dict: Data that has been validated
"""
data = super().deferred_validate(data)
# export META from rpm and prepare dict as saveable format
try:
new_pkg = Package.createrepo_to_dict(
read_crpackage_from_artifact(data["artifact"]))
except OSError:
log.info(traceback.format_exc())
raise NotAcceptable(
detail="RPM file cannot be parsed for metadata")
attrs = {key: new_pkg[key] for key in Package.natural_key_fields()}
package = Package.objects.filter(**attrs)
if package.exists():
keywords = (
"name",
"epoch",
"version",
"release",
"arch",
"checksum_type",
"pkgId",
)
error_data = ", ".join([
"=".join(item) for item in new_pkg.items()
if item[0] in keywords
])
package.get().touch()
raise serializers.ValidationError(
_("There is already a package with: {values}.").format(
values=error_data))
new_pkg["location_href"] = (format_nevra_short(
new_pkg["name"],
new_pkg["epoch"],
new_pkg["version"],
new_pkg["release"],
new_pkg["arch"],
) + ".rpm")
if not data.get("relative_path"):
data["relative_path"] = new_pkg["location_href"]
data.update(new_pkg)
return data
def publish_artifacts(self, content, prefix=""):
"""
Publish artifacts.
Args:
content (pulpcore.plugin.models.Content): content set.
prefix (str): a relative path prefix for the published artifact
"""
published_artifacts = []
# Special case for Packages
contentartifact_qs = ContentArtifact.objects.filter(
content__in=content).filter(
content__pulp_type=Package.get_pulp_type())
for content_artifact in contentartifact_qs.values(
"pk", "relative_path").iterator():
relative_path = content_artifact["relative_path"]
relative_path = os.path.join(prefix, PACKAGES_DIRECTORY,
relative_path.lower()[0],
relative_path)
published_artifacts.append(
PublishedArtifact(
relative_path=relative_path,
publication=self.publication,
content_artifact_id=content_artifact["pk"],
))
# Handle everything else
is_treeinfo = Q(relative_path__in=["treeinfo", ".treeinfo"])
unpublishable_types = Q(content__pulp_type__in=[
RepoMetadataFile.get_pulp_type(),
Modulemd.get_pulp_type(),
ModulemdDefaults.get_pulp_type(),
# already dealt with
Package.get_pulp_type(),
])
contentartifact_qs = (ContentArtifact.objects.filter(
content__in=content).exclude(unpublishable_types).exclude(
is_treeinfo))
for content_artifact in contentartifact_qs.values(
"pk", "relative_path").iterator():
published_artifacts.append(
PublishedArtifact(
relative_path=content_artifact["relative_path"],
publication=self.publication,
content_artifact_id=content_artifact["pk"],
))
PublishedArtifact.objects.bulk_create(published_artifacts,
batch_size=2000)
async def _parse_packages(self, packages):
progress_data = {
"message": "Parsed Packages",
"code": "sync.parsing.packages",
"total": len(packages),
}
with ProgressReport(**progress_data) as packages_pb:
while True:
try:
(_, pkg) = packages.popitem(last=False)
except KeyError:
break
package = Package(**Package.createrepo_to_dict(pkg))
del pkg
artifact = Artifact(size=package.size_package)
checksum_type = getattr(CHECKSUM_TYPES,
package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urlpath_sanitize(self.data.remote_url,
package.location_href)
filename = os.path.basename(package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download,
)
dc = DeclarativeContent(content=package, d_artifacts=[da])
dc.extra_data = defaultdict(list)
# find if a package relates to a modulemd
if dc.content.nevra in self.data.nevra_to_module.keys():
dc.content.is_modular = True
for dc_modulemd in self.data.nevra_to_module[
dc.content.nevra]:
dc.extra_data["modulemd_relation"].append(dc_modulemd)
dc_modulemd.extra_data["package_relation"].append(dc)
if dc.content.name in self.data.pkgname_to_groups.keys():
for dc_group in self.data.pkgname_to_groups[
dc.content.name]:
dc.extra_data["group_relations"].append(dc_group)
dc_group.extra_data["related_packages"].append(dc)
packages_pb.increment()
await self.put(dc)
def _prepare_package(artifact, filename):
"""
Helper function for creating package.
Copy file to a temp directory under
the user provided filename and
parsing it into a saveable format.
Returns: artifact model as dict
Args:
artifact: inited and validated artifact to save
filename: name of file uploaded by user
"""
# Copy file to a temp directory under the user provided filename
with tempfile.TemporaryDirectory() as td:
temp_path = os.path.join(td, filename)
shutil.copy2(artifact.file.path, temp_path)
cr_pkginfo = createrepo_c.package_from_rpm(temp_path)
package = Package.createrepo_to_dict(cr_pkginfo)
package['location_href'] = filename
# parsing it into a saveable format
new_pkg = {}
for key, value in package.items():
if isinstance(value, list):
new_pkg[key] = json.dumps(value)
else:
new_pkg[key] = value
return new_pkg
def publish_artifacts(self, content):
"""
Publish artifacts.
Args:
content (pulpcore.plugin.models.Content): content set.
"""
published_artifacts = []
for content_artifact in ContentArtifact.objects.filter(
content__in=content.exclude(pulp_type__in=[
RepoMetadataFile.get_pulp_type(),
Modulemd.get_pulp_type(),
ModulemdDefaults.get_pulp_type()
]).distinct()).iterator():
relative_path = content_artifact.relative_path
if content_artifact.content.pulp_type == Package.get_pulp_type():
relative_path = os.path.join(PACKAGES_DIRECTORY,
relative_path.lower()[0],
content_artifact.relative_path)
published_artifacts.append(
PublishedArtifact(relative_path=relative_path,
publication=self.publication,
content_artifact=content_artifact))
PublishedArtifact.objects.bulk_create(published_artifacts,
batch_size=2000)
def deferred_validate(self, data):
"""
Validate the rpm package data.
Args:
data (dict): Data to be validated
Returns:
dict: Data that has been validated
"""
data = super().deferred_validate(data)
# export META from rpm and prepare dict as saveable format
try:
new_pkg = _prepare_package(data["artifact"], data["relative_path"])
except OSError:
raise NotAcceptable(detail='RPM file cannot be parsed for metadata.')
attrs = {key: new_pkg[key] for key in Package.natural_key_fields()}
package = Package.objects.filter(**attrs)
if package.exists():
keywords = ('name', 'epoch', 'version', 'release', 'arch', 'checksum_type', 'pkgId')
error_data = ", ".join(
["=".join(item) for item in new_pkg.items() if item[0] in keywords]
)
raise serializers.ValidationError(
_(
"There is already a package with: {values}."
).format(values=error_data)
)
data.update(new_pkg)
return data
def _prepare_package(artifact, filename):
"""
Helper function for creating package.
Copy file to a temp directory under
the user provided filename.
Returns: artifact model as dict
Args:
artifact: inited and validated artifact to save
filename: name of file uploaded by user
"""
artifact_file = storage.open(artifact.file.name)
with tempfile.NamedTemporaryFile("wb", dir=".",
suffix=filename) as temp_file:
shutil.copyfileobj(artifact_file, temp_file)
temp_file.flush()
cr_pkginfo = createrepo_c.package_from_rpm(
temp_file.name, changelog_limit=settings.KEEP_CHANGELOG_LIMIT)
package = Package.createrepo_to_dict(cr_pkginfo)
package["location_href"] = filename
artifact_file.close()
return package
def _apply_retention_policy(self, new_version):
"""Apply the repository's "retain_package_versions" settings to the new version.
Remove all non-modular packages that are older than the retention policy. A value of 0
for the package retention policy represents disabled. A value of 3 would mean that the
3 most recent versions of each package would be kept while older versions are discarded.
Args:
new_version (models.RepositoryVersion): Repository version to filter
"""
assert not new_version.complete, \
"Cannot apply retention policy to completed repository versions"
if self.retain_package_versions > 0:
# It would be more ideal if, instead of annotating with an age and filtering manually,
# we could use Django to filter the particular Package content we want to delete.
# Something like ".filter(F('age') > self.retain_package_versions)" would be better
# however this is not currently possible with Django. It would be possible with raw
# SQL but the repository version content membership subquery is currently
# django-managed and would be difficult to share.
#
# Instead we have to do the filtering manually.
nonmodular_packages = Package.objects.with_age().filter(
pk__in=new_version.content.filter(pulp_type=Package.get_pulp_type()),
is_modular=False, # don't want to filter out modular RPMs
).only('pk')
old_packages = []
for package in nonmodular_packages:
if package.age > self.retain_package_versions:
old_packages.append(package.pk)
new_version.remove_content(Content.objects.filter(pk__in=old_packages))
async def _parse_packages(self, packages):
progress_data = {
'message': 'Parsed Packages',
'code': 'parsing.packages',
'total': len(packages),
}
with ProgressReport(**progress_data) as packages_pb:
for pkg in packages.values():
package = Package(**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(CHECKSUM_TYPES,
package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(self.data.remote_url, package.location_href)
filename = os.path.basename(package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download)
dc = DeclarativeContent(content=package, d_artifacts=[da])
dc.extra_data = defaultdict(list)
# find if a package relates to a modulemd
if dc.content.nevra in self.data.nevra_to_module.keys():
dc.content.is_modular = True
for dc_modulemd in self.data.nevra_to_module[
dc.content.nevra]:
dc.extra_data['modulemd_relation'].append(dc_modulemd)
dc_modulemd.extra_data['package_relation'].append(dc)
if dc.content.name in self.data.pkgname_to_groups.keys():
for dc_group in self.data.pkgname_to_groups[
dc.content.name]:
dc.extra_data['group_relations'].append(dc_group)
dc_group.extra_data['related_packages'].append(dc)
packages_pb.increment()
await self.put(dc)
def create(self, request):
"""
Create a new Package from a request.
"""
try:
artifact = self.get_resource(request.data['_artifact'], Artifact)
except KeyError:
raise serializers.ValidationError(
detail={'_artifact': _('This field is required')})
try:
filename = request.data['filename']
except KeyError:
raise serializers.ValidationError(
detail={'filename': _('This field is required')})
# Copy file to a temp directory under the user provided filename
with tempfile.TemporaryDirectory() as td:
temp_path = os.path.join(td, filename)
shutil.copy2(artifact.file.path, temp_path)
cr_pkginfo = createrepo_c.package_from_rpm(temp_path)
package = Package.createrepo_to_dict(cr_pkginfo)
package['location_href'] = filename
# TODO: Clean this up, maybe make a new function for the purpose of parsing it into
# a saveable format
new_pkg = {}
new_pkg['_artifact'] = request.data['_artifact']
for key, value in package.items():
if isinstance(value, list):
new_pkg[key] = json.dumps(value)
else:
new_pkg[key] = value
serializer = self.get_serializer(data=new_pkg)
serializer.is_valid(raise_exception=True)
serializer.validated_data.pop('_artifact')
package = serializer.save()
if package.pk:
ContentArtifact.objects.create(artifact=artifact,
content=package,
relative_path=package.filename)
headers = self.get_success_headers(request.data)
return Response(serializer.data,
status=status.HTTP_201_CREATED,
headers=headers)
def _prepare_package(artifact, filename):
"""
Helper function for creating package.
Copy file to a temp directory under
the user provided filename.
Returns: artifact model as dict
Args:
artifact: inited and validated artifact to save
filename: name of file uploaded by user
"""
artifact_file = storage.open(artifact.file.name)
with tempfile.NamedTemporaryFile('wb', suffix=filename) as temp_file:
shutil.copyfileobj(artifact_file, temp_file)
temp_file.flush()
cr_pkginfo = createrepo_c.package_from_rpm(temp_file.name)
package = Package.createrepo_to_dict(cr_pkginfo)
package['location_href'] = filename
return package
def _prepare_package(artifact, filename):
"""
Helper function for creating package.
Copy file to a temp directory under
the user provided filename.
Returns: artifact model as dict
Args:
artifact: inited and validated artifact to save
filename: name of file uploaded by user
"""
# Copy file to a temp directory under the user provided filename
with tempfile.TemporaryDirectory() as td:
temp_path = os.path.join(td, filename)
shutil.copy2(artifact.file.path, temp_path)
cr_pkginfo = createrepo_c.package_from_rpm(temp_path)
package = Package.createrepo_to_dict(cr_pkginfo)
package['location_href'] = filename
return package
def find_children_of_content(content, repository_version):
"""Finds the content referenced directly by other content and returns it all together.
Finds RPMs referenced by Advisory/Errata content.
Args:
content (iterable): Content for which to resolve children
repository_version (pulpcore.models.RepositoryVersion): Source repo version
Returns: Queryset of Content objects that are children of the intial set of content
"""
# Advisories that were selected to be copied
advisory_ids = content.filter(
pulp_type=UpdateRecord.get_pulp_type()).only('pk')
# All packages in the source repository version
package_ids = repository_version.content.filter(
pulp_type=Package.get_pulp_type()).only('pk')
# All modules in the source repository version
module_ids = repository_version.content.filter(
pulp_type=Modulemd.get_pulp_type()).only('pk')
advisories = UpdateRecord.objects.filter(pk__in=advisory_ids)
packages = Package.objects.filter(pk__in=package_ids)
modules = Modulemd.objects.filter(pk__in=module_ids)
children = set()
for advisory in advisories:
# Find rpms referenced by Advisories/Errata
package_nevras = advisory.get_pkglist()
for nevra in package_nevras:
(name, epoch, version, release, arch) = nevra
try:
package = packages.get(name=name,
epoch=epoch,
version=version,
release=release,
arch=arch)
children.add(package.pk)
except Package.DoesNotExist:
raise
except MultipleObjectsReturned:
raise
module_nsvcas = advisory.get_module_list()
for nsvca in module_nsvcas:
(name, stream, version, context, arch) = nsvca
try:
module = modules.get(name=name,
stream=stream,
version=version,
context=context,
arch=arch)
children.add(module.pk)
except Modulemd.DoesNotExist:
raise
except MultipleObjectsReturned:
raise
# TODO: Find rpms referenced by PackageGroups,
# PackageGroups referenced by PackageCategories, etc.
return Content.objects.filter(pk__in=children)
def find_children_of_content(content, src_repo_version):
"""Finds the content referenced directly by other content and returns it all together.
Finds RPMs referenced by Advisory/Errata content.
Args:
content (iterable): Content for which to resolve children
src_repo_version (pulpcore.models.RepositoryVersion): Source repo version
Returns: Queryset of Content objects that are children of the intial set of content
"""
# Content that were selected to be copied
advisory_ids = content.filter(
pulp_type=UpdateRecord.get_pulp_type()).only('pk')
packagecategory_ids = content.filter(
pulp_type=PackageCategory.get_pulp_type()).only('pk')
packageenvironment_ids = content.filter(
pulp_type=PackageEnvironment.get_pulp_type()).only('pk')
packagegroup_ids = content.filter(
pulp_type=PackageGroup.get_pulp_type()).only('pk')
# Content in the source repository version
package_ids = src_repo_version.content.filter(
pulp_type=Package.get_pulp_type()).only('pk')
module_ids = src_repo_version.content.filter(
pulp_type=Modulemd.get_pulp_type()).only('pk')
advisories = UpdateRecord.objects.filter(pk__in=advisory_ids)
packages = Package.objects.filter(pk__in=package_ids)
packagecategories = PackageCategory.objects.filter(
pk__in=packagecategory_ids)
packageenvironments = PackageEnvironment.objects.filter(
pk__in=packageenvironment_ids)
packagegroups = PackageGroup.objects.filter(pk__in=packagegroup_ids)
modules = Modulemd.objects.filter(pk__in=module_ids)
children = set()
for advisory in advisories:
# Find rpms referenced by Advisories/Errata
package_nevras = advisory.get_pkglist()
for nevra in package_nevras:
(name, epoch, version, release, arch) = nevra
try:
package = packages.get(name=name,
epoch=epoch,
version=version,
release=release,
arch=arch)
children.add(package.pk)
except Package.DoesNotExist:
raise
except MultipleObjectsReturned:
raise
module_nsvcas = advisory.get_module_list()
for nsvca in module_nsvcas:
(name, stream, version, context, arch) = nsvca
try:
module = modules.get(name=name,
stream=stream,
version=version,
context=context,
arch=arch)
children.add(module.pk)
except Modulemd.DoesNotExist:
raise
except MultipleObjectsReturned:
raise
# PackageCategories & PackageEnvironments resolution must go before PackageGroups
# TODO: refactor to be more effecient (lower number of queries)
for packagecategory in packagecategories.iterator():
for category_package_group in packagecategory.group_ids:
category_package_groups = PackageGroup.objects.filter(
name=category_package_group['name'],
pk__in=src_repo_version.content)
children.update(
[pkggroup.pk for pkggroup in category_package_groups])
packagegroups = packagegroups.union(category_package_groups)
for packageenvironment in packageenvironments.iterator():
for env_package_group in packageenvironment.group_ids:
env_package_groups = PackageGroup.objects.filter(
name=env_package_group['name'],
pk__in=src_repo_version.content)
children.update([envgroup.pk for envgroup in env_package_groups])
packagegroups = packagegroups.union(env_package_groups)
for optional_env_package_group in packageenvironment.option_ids:
opt_env_package_groups = PackageGroup.objects.filter(
name=optional_env_package_group['name'],
pk__in=src_repo_version.content)
children.update(
[optpkggroup.pk for optpkggroup in opt_env_package_groups])
packagegroups = packagegroups.union(opt_env_package_groups)
# Find rpms referenced by PackageGroups
for packagegroup in packagegroups.iterator():
group_package_names = [pkg['name'] for pkg in packagegroup.packages]
for pkg in group_package_names:
packages_by_name = [
pkg for pkg in Package.objects.with_age().filter(
name=pkg, pk__in=src_repo_version.content) if pkg.age == 1
]
for pkg in packages_by_name:
children.add(pkg.pk)
return Content.objects.filter(pk__in=children)
async def run(self):
"""
Build `DeclarativeContent` from the repodata.
"""
remote_url = self.new_url or self.remote.url
remote_url = remote_url if remote_url[-1] == "/" else f"{remote_url}/"
optimize_sync = self.optimize
progress_data = dict(message='Downloading Metadata Files',
code='downloading.metadata')
with ProgressReport(**progress_data) as metadata_pb:
downloader = self.remote.get_downloader(
url=urljoin(remote_url, 'repodata/repomd.xml'))
# TODO: decide how to distinguish between a mirror list and a normal repo
result = await downloader.run()
metadata_pb.increment()
repomd_path = result.path
repomd = cr.Repomd(repomd_path)
# Caution: we are not storing when the remote was last updated, so the order of this
# logic must remain in this order where we first check the version number as other
# changes than sync could have taken place such that the date or repo version will be
# different from last sync
if (optimize_sync and self.repository.last_sync_remote
and self.remote.pk == self.repository.last_sync_remote.pk
and (self.repository.last_sync_repo_version
== self.repository.latest_version().number)
and (self.remote.pulp_last_updated <=
self.repository.latest_version().pulp_created)
and is_previous_version(
repomd.revision,
self.repository.last_sync_revision_number)):
optimize_data = dict(message='Optimizing Sync',
code='optimizing.sync')
with ProgressReport(**optimize_data) as optimize_pb:
optimize_pb.done = 1
optimize_pb.save()
return
self.repository.last_sync_revision_number = repomd.revision
if self.treeinfo:
d_artifacts = [
DeclarativeArtifact(
artifact=Artifact(),
url=urljoin(remote_url, self.treeinfo["filename"]),
relative_path=".treeinfo",
remote=self.remote,
deferred_download=False,
)
]
for path, checksum in self.treeinfo["download"][
"images"].items():
artifact = Artifact(**checksum)
da = DeclarativeArtifact(
artifact=artifact,
url=urljoin(remote_url, path),
relative_path=path,
remote=self.remote,
deferred_download=self.deferred_download)
d_artifacts.append(da)
distribution_tree = DistributionTree(
**self.treeinfo["distribution_tree"])
dc = DeclarativeContent(content=distribution_tree,
d_artifacts=d_artifacts)
dc.extra_data = self.treeinfo
await self.put(dc)
package_repodata_urls = {}
downloaders = []
modulemd_list = list()
dc_groups = []
dc_categories = []
dc_environments = []
nevra_to_module = defaultdict(dict)
pkgname_to_groups = defaultdict(list)
group_to_categories = defaultdict(list)
group_to_environments = defaultdict(list)
optionalgroup_to_environments = defaultdict(list)
modulemd_results = None
comps_downloader = None
main_types = set()
checksums = {}
for record in repomd.records:
checksums[record.type] = record.checksum_type.upper()
if record.type in PACKAGE_REPODATA:
main_types.update([record.type])
package_repodata_urls[record.type] = urljoin(
remote_url, record.location_href)
elif record.type in UPDATE_REPODATA:
updateinfo_url = urljoin(remote_url, record.location_href)
downloader = self.remote.get_downloader(url=updateinfo_url)
downloaders.append([downloader.run()])
elif record.type in COMPS_REPODATA:
comps_url = urljoin(remote_url, record.location_href)
comps_downloader = self.remote.get_downloader(
url=comps_url)
elif record.type in SKIP_REPODATA:
continue
elif '_zck' in record.type:
continue
elif record.type in MODULAR_REPODATA:
modules_url = urljoin(remote_url, record.location_href)
modulemd_downloader = self.remote.get_downloader(
url=modules_url)
modulemd_results = await modulemd_downloader.run()
elif record.type not in PACKAGE_DB_REPODATA:
file_data = {
record.checksum_type: record.checksum,
"size": record.size
}
da = DeclarativeArtifact(
artifact=Artifact(**file_data),
url=urljoin(remote_url, record.location_href),
relative_path=record.location_href,
remote=self.remote,
deferred_download=False)
repo_metadata_file = RepoMetadataFile(
data_type=record.type,
checksum_type=record.checksum_type,
checksum=record.checksum,
)
dc = DeclarativeContent(content=repo_metadata_file,
d_artifacts=[da])
await self.put(dc)
missing_type = set(PACKAGE_REPODATA) - main_types
if missing_type:
raise FileNotFoundError(
_("XML file(s): {filename} not found").format(
filename=", ".join(missing_type)))
self.repository.original_checksum_types = checksums
# we have to sync module.yaml first if it exists, to make relations to packages
if modulemd_results:
modulemd_index = mmdlib.ModuleIndex.new()
open_func = gzip.open if modulemd_results.url.endswith(
'.gz') else open
with open_func(modulemd_results.path, 'r') as moduleyaml:
content = moduleyaml.read()
module_content = content if isinstance(
content, str) else content.decode()
modulemd_index.update_from_string(module_content, True)
modulemd_names = modulemd_index.get_module_names() or []
modulemd_all = parse_modulemd(modulemd_names, modulemd_index)
# Parsing modules happens all at one time, and from here on no useful work happens.
# So just report that it finished this stage.
modulemd_pb_data = {
'message': 'Parsed Modulemd',
'code': 'parsing.modulemds'
}
with ProgressReport(**modulemd_pb_data) as modulemd_pb:
modulemd_total = len(modulemd_all)
modulemd_pb.total = modulemd_total
modulemd_pb.done = modulemd_total
for modulemd in modulemd_all:
artifact = modulemd.pop('artifact')
relative_path = '{}{}{}{}{}snippet'.format(
modulemd[PULP_MODULE_ATTR.NAME],
modulemd[PULP_MODULE_ATTR.STREAM],
modulemd[PULP_MODULE_ATTR.VERSION],
modulemd[PULP_MODULE_ATTR.CONTEXT],
modulemd[PULP_MODULE_ATTR.ARCH])
da = DeclarativeArtifact(artifact=artifact,
relative_path=relative_path,
url=modules_url)
modulemd_content = Modulemd(**modulemd)
dc = DeclarativeContent(content=modulemd_content,
d_artifacts=[da])
dc.extra_data = defaultdict(list)
# dc.content.artifacts are Modulemd artifacts
for artifact in dc.content.artifacts:
nevra_to_module.setdefault(artifact, set()).add(dc)
modulemd_list.append(dc)
# delete list now that we're done with it for memory savings
del modulemd_all
modulemd_default_names = parse_defaults(modulemd_index)
# Parsing module-defaults happens all at one time, and from here on no useful
# work happens. So just report that it finished this stage.
modulemd_defaults_pb_data = {
'message': 'Parsed Modulemd-defaults',
'code': 'parsing.modulemd_defaults'
}
with ProgressReport(
**modulemd_defaults_pb_data) as modulemd_defaults_pb:
modulemd_defaults_total = len(modulemd_default_names)
modulemd_defaults_pb.total = modulemd_defaults_total
modulemd_defaults_pb.done = modulemd_defaults_total
for default in modulemd_default_names:
artifact = default.pop('artifact')
relative_path = '{}{}snippet'.format(
default[PULP_MODULEDEFAULTS_ATTR.MODULE],
default[PULP_MODULEDEFAULTS_ATTR.STREAM])
da = DeclarativeArtifact(artifact=artifact,
relative_path=relative_path,
url=modules_url)
default_content = ModulemdDefaults(**default)
dc = DeclarativeContent(content=default_content,
d_artifacts=[da])
await self.put(dc)
# delete list now that we're done with it for memory savings
del modulemd_default_names
if comps_downloader:
comps_result = await comps_downloader.run()
comps = libcomps.Comps()
comps.fromxml_f(comps_result.path)
with ProgressReport(message='Parsed Comps',
code='parsing.comps') as comps_pb:
comps_total = (len(comps.groups) + len(comps.categories) +
len(comps.environments))
comps_pb.total = comps_total
comps_pb.done = comps_total
if comps.langpacks:
langpack_dict = PackageLangpacks.libcomps_to_dict(
comps.langpacks)
packagelangpack = PackageLangpacks(
matches=strdict_to_dict(comps.langpacks),
digest=dict_digest(langpack_dict))
dc = DeclarativeContent(content=packagelangpack)
dc.extra_data = defaultdict(list)
await self.put(dc)
if comps.categories:
for category in comps.categories:
category_dict = PackageCategory.libcomps_to_dict(
category)
category_dict['digest'] = dict_digest(category_dict)
packagecategory = PackageCategory(**category_dict)
dc = DeclarativeContent(content=packagecategory)
dc.extra_data = defaultdict(list)
if packagecategory.group_ids:
for group_id in packagecategory.group_ids:
group_to_categories[group_id['name']].append(
dc)
dc_categories.append(dc)
if comps.environments:
for environment in comps.environments:
environment_dict = PackageEnvironment.libcomps_to_dict(
environment)
environment_dict['digest'] = dict_digest(
environment_dict)
packageenvironment = PackageEnvironment(
**environment_dict)
dc = DeclarativeContent(content=packageenvironment)
dc.extra_data = defaultdict(list)
if packageenvironment.option_ids:
for option_id in packageenvironment.option_ids:
optionalgroup_to_environments[
option_id['name']].append(dc)
if packageenvironment.group_ids:
for group_id in packageenvironment.group_ids:
group_to_environments[group_id['name']].append(
dc)
dc_environments.append(dc)
if comps.groups:
for group in comps.groups:
group_dict = PackageGroup.libcomps_to_dict(group)
group_dict['digest'] = dict_digest(group_dict)
packagegroup = PackageGroup(**group_dict)
dc = DeclarativeContent(content=packagegroup)
dc.extra_data = defaultdict(list)
if packagegroup.packages:
for package in packagegroup.packages:
pkgname_to_groups[package['name']].append(dc)
if dc.content.id in group_to_categories.keys():
for dc_category in group_to_categories[
dc.content.id]:
dc.extra_data['category_relations'].append(
dc_category)
dc_category.extra_data['packagegroups'].append(
dc)
if dc.content.id in group_to_environments.keys():
for dc_environment in group_to_environments[
dc.content.id]:
dc.extra_data['environment_relations'].append(
dc_environment)
dc_environment.extra_data[
'packagegroups'].append(dc)
if dc.content.id in optionalgroup_to_environments.keys(
):
for dc_environment in optionalgroup_to_environments[
dc.content.id]:
dc.extra_data['env_relations_optional'].append(
dc_environment)
dc_environment.extra_data[
'optionalgroups'].append(dc)
dc_groups.append(dc)
for dc_category in dc_categories:
await self.put(dc_category)
for dc_environment in dc_environments:
await self.put(dc_environment)
# delete lists now that we're done with them for memory savings
del dc_environments
del dc_categories
# to preserve order, downloaders are created after all repodata urls are identified
package_repodata_downloaders = []
for repodata_type in PACKAGE_REPODATA:
downloader = self.remote.get_downloader(
url=package_repodata_urls[repodata_type])
package_repodata_downloaders.append(downloader.run())
downloaders.append(package_repodata_downloaders)
# asyncio.gather is used to preserve the order of results for package repodata
pending = [
asyncio.gather(*downloaders_group)
for downloaders_group in downloaders
]
while pending:
done, pending = await asyncio.wait(
pending, return_when=asyncio.FIRST_COMPLETED)
for downloader in done:
try:
results = downloader.result()
except ClientResponseError as exc:
raise HTTPNotFound(
reason=_("File not found: {filename}").format(
filename=exc.request_info.url))
if results[0].url == package_repodata_urls['primary']:
primary_xml_path = results[0].path
filelists_xml_path = results[1].path
other_xml_path = results[2].path
metadata_pb.done += 3
metadata_pb.save()
packages = await RpmFirstStage.parse_repodata(
primary_xml_path, filelists_xml_path,
other_xml_path)
# skip SRPM if defined
if 'srpm' in self.skip_types:
packages = {
pkgId: pkg
for pkgId, pkg in packages.items()
if pkg.arch != 'src'
}
progress_data = {
'message': 'Parsed Packages',
'code': 'parsing.packages',
'total': len(packages),
}
with ProgressReport(**progress_data) as packages_pb:
for pkg in packages.values():
package = Package(
**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(
CHECKSUM_TYPES,
package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(remote_url,
package.location_href)
filename = os.path.basename(
package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download)
dc = DeclarativeContent(content=package,
d_artifacts=[da])
dc.extra_data = defaultdict(list)
# find if a package relates to a modulemd
if dc.content.nevra in nevra_to_module.keys():
dc.content.is_modular = True
for dc_modulemd in nevra_to_module[
dc.content.nevra]:
dc.extra_data[
'modulemd_relation'].append(
dc_modulemd)
dc_modulemd.extra_data[
'package_relation'].append(dc)
if dc.content.name in pkgname_to_groups.keys():
for dc_group in pkgname_to_groups[
dc.content.name]:
dc.extra_data[
'group_relations'].append(dc_group)
dc_group.extra_data[
'related_packages'].append(dc)
packages_pb.increment()
await self.put(dc)
elif results[0].url == updateinfo_url:
updateinfo_xml_path = results[0].path
metadata_pb.increment()
updates = await RpmFirstStage.parse_updateinfo(
updateinfo_xml_path)
progress_data = {
'message': 'Parsed Advisories',
'code': 'parsing.advisories',
'total': len(updates),
}
with ProgressReport(**progress_data) as advisories_pb:
for update in updates:
update_record = UpdateRecord(
**UpdateRecord.createrepo_to_dict(update))
update_record.digest = hash_update_record(
update)
future_relations = {
'collections': defaultdict(list),
'references': []
}
for collection in update.collections:
coll_dict = UpdateCollection.createrepo_to_dict(
collection)
coll = UpdateCollection(**coll_dict)
for package in collection.packages:
pkg_dict = UpdateCollectionPackage.createrepo_to_dict(
package)
pkg = UpdateCollectionPackage(
**pkg_dict)
future_relations['collections'][
coll].append(pkg)
for reference in update.references:
reference_dict = UpdateReference.createrepo_to_dict(
reference)
ref = UpdateReference(**reference_dict)
future_relations['references'].append(ref)
advisories_pb.increment()
dc = DeclarativeContent(content=update_record)
dc.extra_data = future_relations
await self.put(dc)
# now send modules down the pipeline since all relations have been set up
for modulemd in modulemd_list:
await self.put(modulemd)
for dc_group in dc_groups:
await self.put(dc_group)
def publish(
repository_version_pk,
gpgcheck_options=None,
metadata_signing_service=None,
checksum_types=None,
sqlite_metadata=False,
):
"""
Create a Publication based on a RepositoryVersion.
Args:
repository_version_pk (str): Create a publication from this repository version.
gpgcheck_options (dict): GPG signature check options.
metadata_signing_service (pulpcore.app.models.AsciiArmoredDetachedSigningService):
A reference to an associated signing service.
checksum_types (dict): Checksum types for metadata and packages.
sqlite_metadata (bool): Whether to generate metadata files in sqlite format.
"""
repository_version = RepositoryVersion.objects.get(pk=repository_version_pk)
repository = repository_version.repository.cast()
checksum_types = checksum_types or {}
if metadata_signing_service:
metadata_signing_service = AsciiArmoredDetachedSigningService.objects.get(
pk=metadata_signing_service
)
checksum_types["original"] = repository.original_checksum_types
log.info(
_("Publishing: repository={repo}, version={version}").format(
repo=repository.name,
version=repository_version.number,
)
)
with tempfile.TemporaryDirectory("."):
with RpmPublication.create(repository_version) as publication:
kwargs = {}
first_package = repository_version.content.filter(
pulp_type=Package.get_pulp_type()
).first()
if first_package:
kwargs["default"] = first_package.cast().checksum_type
publication.metadata_checksum_type = get_checksum_type(
"primary", checksum_types, **kwargs
)
publication.package_checksum_type = (
checksum_types.get("package") or publication.metadata_checksum_type
)
if gpgcheck_options is not None:
publication.gpgcheck = gpgcheck_options.get("gpgcheck")
publication.repo_gpgcheck = gpgcheck_options.get("repo_gpgcheck")
if sqlite_metadata:
publication.sqlite_metadata = True
publication_data = PublicationData(publication)
publication_data.populate()
total_repos = 1 + len(publication_data.sub_repos)
pb_data = dict(
message="Generating repository metadata",
code="publish.generating_metadata",
total=total_repos,
)
with ProgressReport(**pb_data) as publish_pb:
content = publication.repository_version.content
# Main repo
generate_repo_metadata(
content,
publication,
checksum_types,
publication_data.repomdrecords,
metadata_signing_service=metadata_signing_service,
)
publish_pb.increment()
for sub_repo in publication_data.sub_repos:
name = sub_repo[0]
checksum_types["original"] = getattr(publication_data, f"{name}_checksums")
content = getattr(publication_data, f"{name}_content")
extra_repomdrecords = getattr(publication_data, f"{name}_repomdrecords")
generate_repo_metadata(
content,
publication,
checksum_types,
extra_repomdrecords,
name,
metadata_signing_service=metadata_signing_service,
)
publish_pb.increment()
log.info(_("Publication: {publication} created").format(publication=publication.pk))
return publication
def publish_artifacts(self, content, prefix=""):
"""
Publish artifacts.
Args:
content (pulpcore.plugin.models.Content): content set.
prefix (str): a relative path prefix for the published artifact
"""
published_artifacts = []
# Special case for Packages
contentartifact_qs = ContentArtifact.objects.filter(content__in=content).filter(
content__pulp_type=Package.get_pulp_type()
)
paths = set()
duplicated_paths = []
for content_artifact in contentartifact_qs.values("pk", "relative_path").iterator():
relative_path = content_artifact["relative_path"]
relative_path = os.path.join(
prefix, PACKAGES_DIRECTORY, relative_path.lower()[0], relative_path
)
#
# Some Suboptimal Repos have the 'same' artifact living in multiple places.
# Specifically, the same NEVRA, in more than once place, **with different checksums**
# (since if all that was different was location_href there would be only one
# ContentArtifact in the first place).
#
# pulp_rpm wants to publish a 'canonical' repository-layout, under which an RPM
# "name-version-release-arch" appears at "Packages/n/name-version-release-arch.rpm".
# Because the assumption is that Packages don't "own" their path, only the filename
# is kept as relative_path.
#
# In this case, we have to pick one - which is essentially what the rest of the RPM
# Ecosystem does when faced with the impossible. This code takes the first-found. We
# could implement something more complicated, if there are better options
# (choose by last-created maybe?)
#
# Note that this only impacts user-created publications, which produce the "standard"
# RPM layout of repo/Packages/f/foo.rpm. A publication created by mirror-sync retains
# whatever layout their "upstream" repo-metadata dictates.
#
if relative_path in paths:
duplicated_paths.append(f'{relative_path}:{content_artifact["pk"]}')
continue
else:
paths.add(relative_path)
published_artifacts.append(
PublishedArtifact(
relative_path=relative_path,
publication=self.publication,
content_artifact_id=content_artifact["pk"],
)
)
if duplicated_paths:
log.warning(
_("Duplicate paths found at publish : {problems} ").format(
problems="; ".join(duplicated_paths)
)
)
# Handle everything else
is_treeinfo = Q(relative_path__in=["treeinfo", ".treeinfo"])
unpublishable_types = Q(
content__pulp_type__in=[
RepoMetadataFile.get_pulp_type(),
Modulemd.get_pulp_type(),
ModulemdDefaults.get_pulp_type(),
# already dealt with
Package.get_pulp_type(),
]
)
contentartifact_qs = (
ContentArtifact.objects.filter(content__in=content)
.exclude(unpublishable_types)
.exclude(is_treeinfo)
)
for content_artifact in contentartifact_qs.values("pk", "relative_path").iterator():
published_artifacts.append(
PublishedArtifact(
relative_path=content_artifact["relative_path"],
publication=self.publication,
content_artifact_id=content_artifact["pk"],
)
)
PublishedArtifact.objects.bulk_create(published_artifacts, batch_size=2000)
def find_children_of_content(content, src_repo_version):
"""Finds the content referenced directly by other content and returns it all together.
Finds RPMs referenced by Advisory/Errata content.
Args:
content (Queryset): Content for which to resolve children
src_repo_version (pulpcore.models.RepositoryVersion): Source repo version
Returns: Queryset of Content objects that are children of the intial set of content
"""
# Content that were selected to be copied
advisory_ids = content.filter(
pulp_type=UpdateRecord.get_pulp_type()).only("pk")
packagecategory_ids = content.filter(
pulp_type=PackageCategory.get_pulp_type()).only("pk")
packageenvironment_ids = content.filter(
pulp_type=PackageEnvironment.get_pulp_type()).only("pk")
packagegroup_ids = content.filter(
pulp_type=PackageGroup.get_pulp_type()).only("pk")
# Content in the source repository version
package_ids = src_repo_version.content.filter(
pulp_type=Package.get_pulp_type()).only("pk")
module_ids = src_repo_version.content.filter(
pulp_type=Modulemd.get_pulp_type()).only("pk")
advisories = UpdateRecord.objects.filter(pk__in=advisory_ids)
packages = Package.objects.filter(pk__in=package_ids)
packagecategories = PackageCategory.objects.filter(
pk__in=packagecategory_ids)
packageenvironments = PackageEnvironment.objects.filter(
pk__in=packageenvironment_ids)
packagegroups = PackageGroup.objects.filter(pk__in=packagegroup_ids)
modules = Modulemd.objects.filter(pk__in=module_ids)
children = set()
for advisory in advisories.iterator():
# Find rpms referenced by Advisories/Errata
package_nevras = advisory.get_pkglist()
advisory_package_q = Q(pk__in=[])
for nevra in package_nevras:
(name, epoch, version, release, arch) = nevra
advisory_package_q |= Q(name=name,
epoch=epoch,
version=version,
release=release,
arch=arch)
children.update(
packages.filter(advisory_package_q).values_list("pk", flat=True))
module_nsvcas = advisory.get_module_list()
advisory_module_q = Q(pk__in=[])
for nsvca in module_nsvcas:
(name, stream, version, context, arch) = nsvca
advisory_module_q |= Q(name=name,
stream=stream,
version=version,
context=context,
arch=arch)
children.update(
modules.filter(advisory_module_q).values_list("pk", flat=True))
# PackageCategories & PackageEnvironments resolution must go before PackageGroups
packagegroup_names = set()
for packagecategory in packagecategories.iterator():
for group_id in packagecategory.group_ids:
packagegroup_names.add(group_id["name"])
for packageenvironment in packageenvironments.iterator():
for group_id in packageenvironment.group_ids:
packagegroup_names.add(group_id["name"])
for group_id in packageenvironment.option_ids:
packagegroup_names.add(group_id["name"])
child_package_groups = PackageGroup.objects.filter(
name__in=packagegroup_names, pk__in=src_repo_version.content)
children.update([pkggroup.pk for pkggroup in child_package_groups])
packagegroups = packagegroups.union(child_package_groups)
# Find rpms referenced by PackageGroups
packagegroup_package_names = set()
for packagegroup in packagegroups.iterator():
packagegroup_package_names |= set(pkg["name"]
for pkg in packagegroup.packages)
# TODO: do modular/nonmodular need to be taken into account?
existing_package_names = (Package.objects.filter(
name__in=packagegroup_package_names,
pk__in=content,
).values_list("name", flat=True).distinct())
missing_package_names = packagegroup_package_names - set(
existing_package_names)
needed_packages = Package.objects.with_age().filter(
name__in=missing_package_names, pk__in=src_repo_version.content)
# Pick the latest version of each package available which isn't already present
# in the content set.
for pkg in needed_packages.iterator():
if pkg.age == 1:
children.add(pkg.pk)
return Content.objects.filter(pk__in=children)
async def run(self):
"""
Build `DeclarativeContent` from the repodata.
"""
packages_pb = ProgressBar(message='Parsed Packages')
erratum_pb = ProgressBar(message='Parsed Erratum')
packages_pb.save()
erratum_pb.save()
with ProgressBar(message='Downloading Metadata Files') as metadata_pb:
downloader = self.remote.get_downloader(
url=urljoin(self.remote.url, 'repodata/repomd.xml'))
# TODO: decide how to distinguish between a mirror list and a normal repo
result = await downloader.run()
metadata_pb.increment()
repomd_path = result.path
repomd = cr.Repomd(repomd_path)
package_repodata_urls = {}
downloaders = []
for record in repomd.records:
if record.type in PACKAGE_REPODATA:
package_repodata_urls[record.type] = urljoin(
self.remote.url, record.location_href)
elif record.type in UPDATE_REPODATA:
updateinfo_url = urljoin(self.remote.url,
record.location_href)
downloader = self.remote.get_downloader(url=updateinfo_url)
downloaders.append([downloader.run()])
else:
log.info(
_('Unknown repodata type: {t}. Skipped.').format(
t=record.type))
# TODO: skip databases, save unknown types to publish them as-is
# to preserve order, downloaders are created after all repodata urls are identified
package_repodata_downloaders = []
for repodata_type in PACKAGE_REPODATA:
downloader = self.remote.get_downloader(
url=package_repodata_urls[repodata_type])
package_repodata_downloaders.append(downloader.run())
downloaders.append(package_repodata_downloaders)
# asyncio.gather is used to preserve the order of results for package repodata
pending = [
asyncio.gather(*downloaders_group)
for downloaders_group in downloaders
]
while pending:
done, pending = await asyncio.wait(
pending, return_when=asyncio.FIRST_COMPLETED)
for downloader in done:
results = downloader.result()
if results[0].url == package_repodata_urls['primary']:
primary_xml_path = results[0].path
filelists_xml_path = results[1].path
other_xml_path = results[2].path
metadata_pb.done += 3
metadata_pb.save()
packages = await RpmFirstStage.parse_repodata(
primary_xml_path, filelists_xml_path,
other_xml_path)
packages_pb.total = len(packages)
packages_pb.state = 'running'
packages_pb.save()
for pkg in packages.values():
package = Package(
**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(
CHECKSUM_TYPES, package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(self.remote.url,
package.location_href)
filename = os.path.basename(package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download)
dc = DeclarativeContent(content=package,
d_artifacts=[da])
packages_pb.increment()
await self.put(dc)
elif results[0].url == updateinfo_url:
updateinfo_xml_path = results[0].path
metadata_pb.increment()
updates = await RpmFirstStage.parse_updateinfo(
updateinfo_xml_path)
erratum_pb.total = len(updates)
erratum_pb.state = 'running'
erratum_pb.save()
for update in updates:
update_record = UpdateRecord(
**UpdateRecord.createrepo_to_dict(update))
update_record.digest = RpmFirstStage.hash_update_record(
update)
future_relations = {
'collections': defaultdict(list),
'references': []
}
for collection in update.collections:
coll_dict = UpdateCollection.createrepo_to_dict(
collection)
coll = UpdateCollection(**coll_dict)
for package in collection.packages:
pkg_dict = UpdateCollectionPackage.createrepo_to_dict(
package)
pkg = UpdateCollectionPackage(**pkg_dict)
future_relations['collections'][
coll].append(pkg)
for reference in update.references:
reference_dict = UpdateReference.createrepo_to_dict(
reference)
ref = UpdateReference(**reference_dict)
future_relations['references'].append(ref)
erratum_pb.increment()
dc = DeclarativeContent(content=update_record)
dc.extra_data = future_relations
await self.put(dc)
packages_pb.state = 'completed'
erratum_pb.state = 'completed'
packages_pb.save()
erratum_pb.save()
def publish_artifacts(self, content, prefix=""):
"""
Publish artifacts.
Args:
content (pulpcore.plugin.models.Content): content set.
prefix (str): a relative path prefix for the published artifact
"""
published_artifacts = []
# Special case for Packages
contentartifact_qs = (ContentArtifact.objects.filter(
content__in=content).filter(
content__pulp_type=Package.get_pulp_type()).select_related(
"content__rpm_package__time_build"))
rel_path_mapping = defaultdict(list)
# Some Suboptimal Repos have the 'same' artifact living in multiple places.
# Specifically, the same NEVRA, in more than once place, **with different checksums**
# (since if all that was different was location_href there would be only one
# ContentArtifact in the first place).
#
# pulp_rpm wants to publish a 'canonical' repository-layout, under which an RPM
# "name-version-release-arch" appears at "Packages/n/name-version-release-arch.rpm".
# Because the assumption is that Packages don't "own" their path, only the filename
# is kept as relative_path.
#
# In this case, we have to pick one - which is essentially what the rest of the RPM
# Ecosystem does when faced with the impossible. This code takes the one with the
# most recent build time which is the same heuristic used by Yum/DNF/Zypper.
#
# Note that this only impacts user-created publications, which produce the "standard"
# RPM layout of repo/Packages/f/foo.rpm. A publication created by mirror-sync retains
# whatever layout their "upstream" repo-metadata dictates.
fields = ["pk", "relative_path", "content__rpm_package__time_build"]
for content_artifact in contentartifact_qs.values(*fields).iterator():
relative_path = content_artifact["relative_path"]
time_build = content_artifact["content__rpm_package__time_build"]
relative_path = os.path.join(prefix, PACKAGES_DIRECTORY,
relative_path.lower()[0],
relative_path)
rel_path_mapping[relative_path].append(
(content_artifact["pk"], time_build))
for rel_path, content_artifacts in rel_path_mapping.items():
# sort the content artifacts by when the package was built
if len(content_artifacts) > 1:
content_artifacts.sort(key=lambda p: p[1], reverse=True)
log.warning(
"Duplicate packages found competing for {path}, selected the one with "
"the most recent build time, excluding {others} others.".
format(path=rel_path, others=len(content_artifacts[1:])))
# Only add the first one (the one with the highest build time)
published_artifacts.append(
PublishedArtifact(
relative_path=rel_path,
publication=self.publication,
content_artifact_id=content_artifacts[0][0],
))
# Handle everything else
is_treeinfo = Q(relative_path__in=["treeinfo", ".treeinfo"])
unpublishable_types = Q(content__pulp_type__in=[
RepoMetadataFile.get_pulp_type(),
Modulemd.get_pulp_type(),
ModulemdDefaults.get_pulp_type(),
# already dealt with
Package.get_pulp_type(),
])
contentartifact_qs = (ContentArtifact.objects.filter(
content__in=content).exclude(unpublishable_types).exclude(
is_treeinfo))
for content_artifact in contentartifact_qs.values(
"pk", "relative_path").iterator():
published_artifacts.append(
PublishedArtifact(
relative_path=content_artifact["relative_path"],
publication=self.publication,
content_artifact_id=content_artifact["pk"],
))
PublishedArtifact.objects.bulk_create(published_artifacts,
batch_size=2000)
async def __call__(self, in_q, out_q):
"""
Build `DeclarativeContent` from the repodata.
Args:
in_q (asyncio.Queue): Unused because the first stage doesn't read from an input queue.
out_q (asyncio.Queue): The out_q to send `DeclarativeContent` objects to
"""
with ProgressBar(message='Downloading and Parsing Metadata') as pb:
downloader = self.remote.get_downloader(
urljoin(self.remote.url, 'repodata/repomd.xml'))
# TODO: decide how to distinguish between a mirror list and a normal repo
result = await downloader.run()
pb.increment()
repomd_path = result.path
repomd = cr.Repomd(repomd_path)
package_repodata_urls = {}
downloaders = []
for record in repomd.records:
if record.type in PACKAGE_REPODATA:
package_repodata_urls[record.type] = urljoin(
self.remote.url, record.location_href)
elif record.type in UPDATE_REPODATA:
updateinfo_url = urljoin(self.remote.url,
record.location_href)
downloader = self.remote.get_downloader(updateinfo_url)
downloaders.append([downloader.run()])
else:
log.info(
_('Unknown repodata type: {t}. Skipped.').format(
t=record.type))
# TODO: skip databases, save unknown types to publish them as-is
# to preserve order, downloaders are created after all repodata urls are identified
package_repodata_downloaders = []
for repodata_type in PACKAGE_REPODATA:
downloader = self.remote.get_downloader(
package_repodata_urls[repodata_type])
package_repodata_downloaders.append(downloader.run())
downloaders.append(package_repodata_downloaders)
# asyncio.gather is used to preserve the order of results for package repodata
pending = [
asyncio.gather(*downloaders_group)
for downloaders_group in downloaders
]
while pending:
done, pending = await asyncio.wait(
pending, return_when=asyncio.FIRST_COMPLETED)
for downloader in done:
results = downloader.result()
if results[0].url == package_repodata_urls['primary']:
primary_xml_path = results[0].path
filelists_xml_path = results[1].path
other_xml_path = results[2].path
pb.done += 3
pb.save()
packages = await RpmFirstStage.parse_repodata(
primary_xml_path, filelists_xml_path,
other_xml_path)
for pkg in packages.values():
package = Package(
**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(
CHECKSUM_TYPES, package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(self.remote.url,
package.location_href)
da = DeclarativeArtifact(artifact, url,
package.location_href,
self.remote)
dc = DeclarativeContent(content=package,
d_artifacts=[da])
await out_q.put(dc)
elif results[0].url == updateinfo_url:
updateinfo_xml_path = results[0].path
pb.increment()
updates = await RpmFirstStage.parse_updateinfo(
updateinfo_xml_path)
for update in updates:
update_record = UpdateRecord(
**UpdateRecord.createrepo_to_dict(update))
update_record.digest = RpmFirstStage.hash_update_record(
update)
for collection in update.collections:
coll_dict = UpdateCollection.createrepo_to_dict(
collection)
coll = UpdateCollection(**coll_dict)
for package in collection.packages:
pkg_dict = UpdateCollectionPackage.createrepo_to_dict(
package)
pkg = UpdateCollectionPackage(**pkg_dict)
coll._packages.append(pkg)
update_record._collections.append(coll)
for reference in update.references:
reference_dict = UpdateReference.createrepo_to_dict(
reference)
update_record._references.append(
UpdateReference(**reference_dict))
dc = DeclarativeContent(content=update_record)
await out_q.put(dc)
await out_q.put(None)
async def run(self):
"""
Build `DeclarativeContent` from the repodata.
"""
with ProgressBar(message='Downloading and Parsing Metadata') as pb:
downloader = self.remote.get_downloader(
url=urljoin(self.remote.url, 'repodata/repomd.xml')
)
# TODO: decide how to distinguish between a mirror list and a normal repo
result = await downloader.run()
pb.increment()
repomd_path = result.path
repomd = cr.Repomd(repomd_path)
package_repodata_urls = {}
downloaders = []
for record in repomd.records:
if record.type in PACKAGE_REPODATA:
package_repodata_urls[record.type] = urljoin(self.remote.url,
record.location_href)
elif record.type in UPDATE_REPODATA:
updateinfo_url = urljoin(self.remote.url, record.location_href)
downloader = self.remote.get_downloader(url=updateinfo_url)
downloaders.append([downloader.run()])
else:
log.info(_('Unknown repodata type: {t}. Skipped.').format(t=record.type))
# TODO: skip databases, save unknown types to publish them as-is
# to preserve order, downloaders are created after all repodata urls are identified
package_repodata_downloaders = []
for repodata_type in PACKAGE_REPODATA:
downloader = self.remote.get_downloader(url=package_repodata_urls[repodata_type])
package_repodata_downloaders.append(downloader.run())
downloaders.append(package_repodata_downloaders)
# asyncio.gather is used to preserve the order of results for package repodata
pending = [asyncio.gather(*downloaders_group) for downloaders_group in downloaders]
while pending:
done, pending = await asyncio.wait(pending, return_when=asyncio.FIRST_COMPLETED)
for downloader in done:
results = downloader.result()
if results[0].url == package_repodata_urls['primary']:
primary_xml_path = results[0].path
filelists_xml_path = results[1].path
other_xml_path = results[2].path
pb.done += 3
pb.save()
packages = await RpmFirstStage.parse_repodata(primary_xml_path,
filelists_xml_path,
other_xml_path)
for pkg in packages.values():
package = Package(**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(CHECKSUM_TYPES, package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(self.remote.url, package.location_href)
filename = os.path.basename(package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download
)
dc = DeclarativeContent(content=package, d_artifacts=[da])
await self.put(dc)
elif results[0].url == updateinfo_url:
updateinfo_xml_path = results[0].path
pb.increment()
updates = await RpmFirstStage.parse_updateinfo(updateinfo_xml_path)
for update in updates:
update_record = UpdateRecord(**UpdateRecord.createrepo_to_dict(update))
update_record.digest = RpmFirstStage.hash_update_record(update)
for collection in update.collections:
coll_dict = UpdateCollection.createrepo_to_dict(collection)
coll = UpdateCollection(**coll_dict)
for package in collection.packages:
pkg_dict = UpdateCollectionPackage.createrepo_to_dict(package)
pkg = UpdateCollectionPackage(**pkg_dict)
coll._packages.append(pkg)
update_record._collections.append(coll)
for reference in update.references:
reference_dict = UpdateReference.createrepo_to_dict(reference)
update_record._references.append(UpdateReference(**reference_dict))
dc = DeclarativeContent(content=update_record)
await self.put(dc)
async def run(self):
"""
Build `DeclarativeContent` from the repodata.
"""
packages_pb = ProgressReport(message='Parsed Packages',
code='parsing.packages')
errata_pb = ProgressReport(message='Parsed Erratum',
code='parsing.errata')
modulemd_pb = ProgressReport(message='Parse Modulemd',
code='parsing.modulemds')
modulemd_defaults_pb = ProgressReport(
message='Parse Modulemd-defaults', code='parsing.modulemddefaults')
comps_pb = ProgressReport(message='Parsed Comps', code='parsing.comps')
packages_pb.save()
errata_pb.save()
comps_pb.save()
remote_url = self.new_url or self.remote.url
remote_url = remote_url if remote_url[-1] == "/" else f"{remote_url}/"
progress_data = dict(message='Downloading Metadata Files',
code='downloading.metadata')
with ProgressReport(**progress_data) as metadata_pb:
downloader = self.remote.get_downloader(
url=urljoin(remote_url, 'repodata/repomd.xml'))
# TODO: decide how to distinguish between a mirror list and a normal repo
result = await downloader.run()
metadata_pb.increment()
if self.kickstart:
d_artifacts = []
for path, checksum in self.kickstart["download"][
"images"].items():
artifact = Artifact(**checksum)
da = DeclarativeArtifact(
artifact=artifact,
url=urljoin(remote_url, path),
relative_path=path,
remote=self.remote,
deferred_download=self.deferred_download)
d_artifacts.append(da)
distribution_tree = DistributionTree(
**self.kickstart["distribution_tree"])
dc = DeclarativeContent(content=distribution_tree,
d_artifacts=d_artifacts)
dc.extra_data = self.kickstart
await self.put(dc)
repomd_path = result.path
repomd = cr.Repomd(repomd_path)
package_repodata_urls = {}
downloaders = []
modulemd_list = list()
dc_groups = []
dc_categories = []
dc_environments = []
nevra_to_module = defaultdict(dict)
pkgname_to_groups = defaultdict(list)
group_to_categories = defaultdict(list)
group_to_environments = defaultdict(list)
optionalgroup_to_environments = defaultdict(list)
modulemd_results = None
comps_downloader = None
for record in repomd.records:
if record.type in PACKAGE_REPODATA:
package_repodata_urls[record.type] = urljoin(
remote_url, record.location_href)
elif record.type in UPDATE_REPODATA:
updateinfo_url = urljoin(remote_url, record.location_href)
downloader = self.remote.get_downloader(url=updateinfo_url)
downloaders.append([downloader.run()])
elif record.type in COMPS_REPODATA:
comps_url = urljoin(remote_url, record.location_href)
comps_downloader = self.remote.get_downloader(
url=comps_url)
elif record.type in SKIP_REPODATA:
continue
elif record.type in MODULAR_REPODATA:
modules_url = urljoin(remote_url, record.location_href)
modulemd_downloader = self.remote.get_downloader(
url=modules_url)
modulemd_results = await modulemd_downloader.run()
elif record.type not in PACKAGE_DB_REPODATA:
file_data = {
record.checksum_type: record.checksum,
"size": record.size
}
da = DeclarativeArtifact(
artifact=Artifact(**file_data),
url=urljoin(remote_url, record.location_href),
relative_path=record.location_href,
remote=self.remote,
deferred_download=False)
repo_metadata_file = RepoMetadataFile(
data_type=record.type,
checksum_type=record.checksum_type,
checksum=record.checksum,
)
dc = DeclarativeContent(content=repo_metadata_file,
d_artifacts=[da])
await self.put(dc)
# we have to sync module.yaml first if it exists, to make relations to packages
if modulemd_results:
modulemd_index = mmdlib.ModuleIndex.new()
open_func = gzip.open if modulemd_results.url.endswith(
'.gz') else open
with open_func(modulemd_results.path, 'r') as moduleyaml:
modulemd_index.update_from_string(
moduleyaml.read().decode(), True)
modulemd_names = modulemd_index.get_module_names() or []
modulemd_all = parse_modulemd(modulemd_names, modulemd_index)
modulemd_pb.total = len(modulemd_all)
modulemd_pb.state = 'running'
modulemd_pb.save()
for modulemd in modulemd_all:
artifact = modulemd.pop('artifact')
relative_path = '{}{}{}{}{}snippet'.format(
modulemd[PULP_MODULE_ATTR.NAME],
modulemd[PULP_MODULE_ATTR.STREAM],
modulemd[PULP_MODULE_ATTR.VERSION],
modulemd[PULP_MODULE_ATTR.CONTEXT],
modulemd[PULP_MODULE_ATTR.ARCH])
da = DeclarativeArtifact(artifact=artifact,
relative_path=relative_path,
url=modules_url)
modulemd_content = Modulemd(**modulemd)
dc = DeclarativeContent(content=modulemd_content,
d_artifacts=[da])
dc.extra_data = defaultdict(list)
# dc.content.artifacts are Modulemd artifacts
for artifact in json.loads(dc.content.artifacts):
nevra_to_module.setdefault(artifact, set()).add(dc)
modulemd_list.append(dc)
modulemd_default_names = parse_defaults(modulemd_index)
modulemd_defaults_pb.total = len(modulemd_default_names)
modulemd_defaults_pb.state = 'running'
modulemd_defaults_pb.save()
for default in modulemd_default_names:
artifact = default.pop('artifact')
relative_path = '{}{}snippet'.format(
default[PULP_MODULEDEFAULTS_ATTR.MODULE],
default[PULP_MODULEDEFAULTS_ATTR.STREAM])
da = DeclarativeArtifact(artifact=artifact,
relative_path=relative_path,
url=modules_url)
default_content = ModulemdDefaults(**default)
modulemd_defaults_pb.increment()
dc = DeclarativeContent(content=default_content,
d_artifacts=[da])
await self.put(dc)
if comps_downloader:
comps_result = await comps_downloader.run()
comps = libcomps.Comps()
comps.fromxml_f(comps_result.path)
comps_pb.total = (len(comps.groups) + len(comps.categories) +
len(comps.environments))
comps_pb.state = 'running'
comps_pb.save()
if comps.langpacks:
langpack_dict = PackageLangpacks.libcomps_to_dict(
comps.langpacks)
packagelangpack = PackageLangpacks(
matches=strdict_to_dict(comps.langpacks),
digest=dict_digest(langpack_dict))
dc = DeclarativeContent(content=packagelangpack)
dc.extra_data = defaultdict(list)
await self.put(dc)
if comps.categories:
for category in comps.categories:
category_dict = PackageCategory.libcomps_to_dict(
category)
category_dict['digest'] = dict_digest(category_dict)
packagecategory = PackageCategory(**category_dict)
dc = DeclarativeContent(content=packagecategory)
dc.extra_data = defaultdict(list)
if packagecategory.group_ids:
for group_id in packagecategory.group_ids:
group_to_categories[group_id['name']].append(
dc)
dc_categories.append(dc)
if comps.environments:
for environment in comps.environments:
environment_dict = PackageEnvironment.libcomps_to_dict(
environment)
environment_dict['digest'] = dict_digest(
environment_dict)
packageenvironment = PackageEnvironment(
**environment_dict)
dc = DeclarativeContent(content=packageenvironment)
dc.extra_data = defaultdict(list)
if packageenvironment.option_ids:
for option_id in packageenvironment.option_ids:
optionalgroup_to_environments[
option_id['name']].append(dc)
if packageenvironment.group_ids:
for group_id in packageenvironment.group_ids:
group_to_environments[group_id['name']].append(
dc)
dc_environments.append(dc)
if comps.groups:
for group in comps.groups:
group_dict = PackageGroup.libcomps_to_dict(group)
group_dict['digest'] = dict_digest(group_dict)
packagegroup = PackageGroup(**group_dict)
dc = DeclarativeContent(content=packagegroup)
dc.extra_data = defaultdict(list)
if packagegroup.packages:
for package in packagegroup.packages:
pkgname_to_groups[package['name']].append(dc)
if dc.content.id in group_to_categories.keys():
for dc_category in group_to_categories[
dc.content.id]:
dc.extra_data['category_relations'].append(
dc_category)
dc_category.extra_data['packagegroups'].append(
dc)
if dc.content.id in group_to_environments.keys():
for dc_environment in group_to_environments[
dc.content.id]:
dc.extra_data['environment_relations'].append(
dc_environment)
dc_environment.extra_data[
'packagegroups'].append(dc)
if dc.content.id in optionalgroup_to_environments.keys(
):
for dc_environment in optionalgroup_to_environments[
dc.content.id]:
dc.extra_data['env_relations_optional'].append(
dc_environment)
dc_environment.extra_data[
'optionalgroups'].append(dc)
dc_groups.append(dc)
for dc_category in dc_categories:
comps_pb.increment()
await self.put(dc_category)
for dc_environment in dc_environments:
comps_pb.increment()
await self.put(dc_environment)
# to preserve order, downloaders are created after all repodata urls are identified
package_repodata_downloaders = []
for repodata_type in PACKAGE_REPODATA:
downloader = self.remote.get_downloader(
url=package_repodata_urls[repodata_type])
package_repodata_downloaders.append(downloader.run())
downloaders.append(package_repodata_downloaders)
# asyncio.gather is used to preserve the order of results for package repodata
pending = [
asyncio.gather(*downloaders_group)
for downloaders_group in downloaders
]
while pending:
done, pending = await asyncio.wait(
pending, return_when=asyncio.FIRST_COMPLETED)
for downloader in done:
results = downloader.result()
if results[0].url == package_repodata_urls['primary']:
primary_xml_path = results[0].path
filelists_xml_path = results[1].path
other_xml_path = results[2].path
metadata_pb.done += 3
metadata_pb.save()
packages = await RpmFirstStage.parse_repodata(
primary_xml_path, filelists_xml_path,
other_xml_path)
packages_pb.total = len(packages)
packages_pb.state = 'running'
packages_pb.save()
for pkg in packages.values():
package = Package(
**Package.createrepo_to_dict(pkg))
artifact = Artifact(size=package.size_package)
checksum_type = getattr(
CHECKSUM_TYPES, package.checksum_type.upper())
setattr(artifact, checksum_type, package.pkgId)
url = urljoin(remote_url, package.location_href)
filename = os.path.basename(package.location_href)
da = DeclarativeArtifact(
artifact=artifact,
url=url,
relative_path=filename,
remote=self.remote,
deferred_download=self.deferred_download)
dc = DeclarativeContent(content=package,
d_artifacts=[da])
dc.extra_data = defaultdict(list)
# find if a package relates to a modulemd
if dc.content.nevra in nevra_to_module.keys():
dc.content.is_modular = True
for dc_modulemd in nevra_to_module[
dc.content.nevra]:
dc.extra_data['modulemd_relation'].append(
dc_modulemd)
dc_modulemd.extra_data[
'package_relation'].append(dc)
if dc.content.name in pkgname_to_groups.keys():
for dc_group in pkgname_to_groups[
dc.content.name]:
dc.extra_data['group_relations'].append(
dc_group)
dc_group.extra_data[
'related_packages'].append(dc)
packages_pb.increment()
await self.put(dc)
elif results[0].url == updateinfo_url:
updateinfo_xml_path = results[0].path
metadata_pb.increment()
updates = await RpmFirstStage.parse_updateinfo(
updateinfo_xml_path)
errata_pb.total = len(updates)
errata_pb.state = 'running'
errata_pb.save()
for update in updates:
update_record = UpdateRecord(
**UpdateRecord.createrepo_to_dict(update))
update_record.digest = RpmFirstStage.hash_update_record(
update)
future_relations = {
'collections': defaultdict(list),
'references': []
}
for collection in update.collections:
coll_dict = UpdateCollection.createrepo_to_dict(
collection)
coll = UpdateCollection(**coll_dict)
for package in collection.packages:
pkg_dict = UpdateCollectionPackage.createrepo_to_dict(
package)
pkg = UpdateCollectionPackage(**pkg_dict)
future_relations['collections'][
coll].append(pkg)
for reference in update.references:
reference_dict = UpdateReference.createrepo_to_dict(
reference)
ref = UpdateReference(**reference_dict)
future_relations['references'].append(ref)
errata_pb.increment()
dc = DeclarativeContent(content=update_record)
dc.extra_data = future_relations
await self.put(dc)
# now send modules down the pipeline since all relations have been set up
for modulemd in modulemd_list:
modulemd_pb.increment()
await self.put(modulemd)
for dc_group in dc_groups:
comps_pb.increment()
await self.put(dc_group)
packages_pb.state = 'completed'
errata_pb.state = 'completed'
modulemd_pb.state = 'completed'
modulemd_defaults_pb.state = 'completed'
comps_pb.state = 'completed'
packages_pb.save()
errata_pb.save()
modulemd_pb.save()
modulemd_defaults_pb.save()
comps_pb.save()
