def migrate_datasets_pre_v0_3(client):
"""Migrate datasets from Renku 0.3.x."""
for old_path in dataset_pre_0_3(client):
name = str(old_path.parent.relative_to(client.path / 'data'))
dataset = Dataset.from_yaml(old_path, client=client)
new_path = (client.renku_datasets_path / dataset.uid / client.METADATA)
new_path.parent.mkdir(parents=True, exist_ok=True)
with client.with_metadata(read_only=True) as meta:
for module in client.repo.submodules:
if Path(module.url).name == meta.name:
module.remove()
for file_ in dataset.files:
if not Path(file_.path).exists():
expected_path = (
client.path / 'data' / dataset.name / file_.path
)
if expected_path.exists():
file_.path = expected_path.relative_to(client.path)
dataset.__reference__ = new_path
dataset.to_yaml()
Path(old_path).unlink()
ref = LinkReference.create(
client=client,
name='datasets/{0}'.format(name),
force=True,
)
ref.set_reference(new_path)
def _migrate_datasets_pre_v0_3(client):
"""Migrate datasets from Renku 0.3.x."""
for old_path in get_pre_0_3_4_datasets_metadata(client):
name = str(old_path.parent.relative_to(client.path / DATA_DIR))
dataset = Dataset.from_yaml(old_path, client=client)
dataset.title = name
dataset.name = name
new_path = client.renku_datasets_path / dataset.identifier / client.METADATA
new_path.parent.mkdir(parents=True, exist_ok=True)
with client.with_metadata(read_only=True) as meta:
for module in client.repo.submodules:
if Path(module.url).name == meta.name:
module.remove()
for file_ in dataset.files:
if not Path(file_.path).exists():
expected_path = client.path / DATA_DIR / dataset.name / file_.path
if expected_path.exists():
file_.path = expected_path.relative_to(client.path)
dataset.to_yaml(new_path)
Path(old_path).unlink()
ref = LinkReference.create(
client=client,
name="datasets/{0}".format(name),
force=True,
)
ref.set_reference(new_path)
def create_dataset(
self,
short_name=None,
title=None,
description=None,
creators=None,
keywords=None,
):
"""Create a dataset."""
if not short_name:
raise errors.ParameterError('Dataset short_name must be provided.')
if not is_dataset_short_name_valid(short_name):
raise errors.ParameterError(
'Dataset short_name "{}" is not valid.'.format(short_name))
if self.load_dataset(short_name=short_name):
raise errors.DatasetExistsError(
'Dataset exists: "{}".'.format(short_name))
if not title:
title = short_name
identifier = str(uuid.uuid4())
path = self.renku_datasets_path / identifier / self.METADATA
if path.exists():
raise errors.DatasetExistsError(
'Dataset with reference {} exists'.format(path))
path.parent.mkdir(parents=True, exist_ok=True)
if creators is None:
creators = [Person.from_git(self.repo)]
keywords = keywords or ()
with with_reference(path):
dataset = Dataset(
client=self,
identifier=identifier,
short_name=short_name,
name=title,
description=description,
creator=creators,
keywords=keywords,
)
dataset_ref = LinkReference.create(client=self,
name='datasets/' + short_name)
dataset_ref.set_reference(path)
dataset.path = Path(dataset.path).relative_to(self.path)
dataset.to_yaml()
return dataset, path, dataset_ref
def _migrate_broken_dataset_paths(client):
"""Ensure all paths are using correct directory structure."""
for dataset in client.datasets.values():
dataset_path = Path(dataset.path)
expected_path = (client.path / client.renku_datasets_path /
Path(quote(dataset.identifier, safe='')))
# migrate the refs
ref = LinkReference.create(
client=client,
name='datasets/{0}'.format(dataset.short_name),
force=True,
)
ref.set_reference(expected_path / client.METADATA)
if not dataset_path.exists():
dataset_path = (client.path / client.renku_datasets_path /
uuid.UUID(dataset.identifier).hex)
if not expected_path.exists():
shutil.move(str(dataset_path), str(expected_path))
dataset.path = expected_path
dataset.__reference__ = expected_path / client.METADATA
for file_ in dataset.files:
file_path = Path(file_.path)
if not file_path.exists() and file_.path.startswith('..'):
new_path = (client.path / client.renku_datasets_path /
dataset.uid / file_path).resolve().relative_to(
client.path)
file_.path = new_path
_, commit, _ = client.resolve_in_submodules(
client.find_previous_commit(file_.path, revision='HEAD'),
file_.path,
)
host = client.remote.get('host') or 'localhost'
host = os.environ.get('RENKU_DOMAIN') or host
# always set the id by the identifier
file_._id = urllib.parse.urljoin(
'https://{host}'.format(host=host),
posixpath.join('/blob/{hexsha}/{path}'.format(
hexsha=commit.hexsha, path=new_path)))
file_._label = '{}@{}'.format(new_path, commit.hexsha)
dataset.to_yaml()
def migrate_broken_dataset_paths(client):
"""Ensure all paths are using correct directory structure."""
for dataset in client.datasets.values():
dataset_path = Path(dataset.path)
expected_path = (
client.renku_datasets_path /
Path(quote(dataset.identifier, safe=''))
)
# migrate the refs
ref = LinkReference.create(
client=client,
name='datasets/{0}'.format(dataset.short_name),
force=True,
)
ref.set_reference(expected_path / client.METADATA)
if not dataset_path.exists():
dataset_path = (
client.renku_datasets_path / uuid.UUID(dataset.identifier).hex
)
if not expected_path.exists():
shutil.move(str(dataset_path), str(expected_path))
dataset.path = expected_path
dataset.__reference__ = expected_path / client.METADATA
for file_ in dataset.files:
file_path = Path(file_.path)
if not file_path.exists() and file_.path.startswith('..'):
new_path = (
client.renku_datasets_path / dataset.uid / file_path
).resolve().relative_to(client.path)
file_.path = new_path
file_._label = new_path
_, commit, _ = client.resolve_in_submodules(
client.find_previous_commit(file_.path, revision='HEAD'),
file_.path,
)
id_format = 'blob/{commit}/{path}'
file_._id = id_format.format(
commit=commit.hexsha, path=new_path
)
dataset.to_yaml()
def create_dataset(self,
name,
short_name=None,
description='',
creators=None):
"""Create a dataset."""
if not name:
raise errors.ParameterError('Dataset name must be provided.')
if not short_name:
short_name = generate_default_short_name(name, None)
if not is_dataset_name_valid(short_name):
raise errors.ParameterError(
'Dataset name "{}" is not valid.'.format(short_name))
if self.load_dataset(name=short_name):
raise errors.DatasetExistsError(
'Dataset exists: "{}".'.format(short_name))
identifier = str(uuid.uuid4())
path = self.renku_datasets_path / identifier / self.METADATA
if path.exists():
raise errors.DatasetExistsError(
'Dataset with reference {} exists'.format(path))
path.parent.mkdir(parents=True, exist_ok=True)
if creators is None:
creators = [Person.from_git(self.repo)]
with with_reference(path):
dataset = Dataset(client=self,
identifier=identifier,
name=name,
short_name=short_name,
description=description,
creator=creators)
dataset_ref = LinkReference.create(client=self,
name='datasets/' + short_name)
dataset_ref.set_reference(path)
dataset.to_yaml()
return dataset, path, dataset_ref
def _migrate_datasets_pre_v0_3(client):
"""Migrate datasets from Renku 0.3.x."""
def _dataset_pre_0_3(client):
"""Return paths of dataset metadata for pre 0.3.4."""
project_is_pre_0_3 = int(client.project.version) < 2
if project_is_pre_0_3:
return (client.path / DATA_DIR).rglob(client.METADATA)
return []
for old_path in _dataset_pre_0_3(client):
name = str(old_path.parent.relative_to(client.path / DATA_DIR))
dataset = Dataset.from_yaml(old_path, client=client)
new_path = (client.renku_datasets_path / dataset.uid / client.METADATA)
new_path.parent.mkdir(parents=True, exist_ok=True)
with client.with_metadata(read_only=True) as meta:
for module in client.repo.submodules:
if Path(module.url).name == meta.name:
module.remove()
for file_ in dataset.files:
if not Path(file_.path).exists():
expected_path = (client.path / DATA_DIR / dataset.name /
file_.path)
if expected_path.exists():
file_.path = expected_path.relative_to(client.path)
dataset.__reference__ = new_path.relative_to(client.path)
dataset.to_yaml()
Path(old_path).unlink()
ref = LinkReference.create(
client=client,
name='datasets/{0}'.format(name),
force=True,
)
ref.set_reference(new_path)
def _migrate_broken_dataset_paths(client):
"""Ensure all paths are using correct directory structure."""
for dataset in get_client_datasets(client):
expected_path = client.renku_datasets_path / dataset.identifier
if not dataset.name:
dataset.name = dataset.title
# migrate the refs
ref = LinkReference.create(
client=client,
name="datasets/{0}".format(dataset.name),
force=True,
)
ref.set_reference(expected_path / client.METADATA)
old_dataset_path = client.renku_datasets_path / uuid.UUID(
dataset.identifier).hex
dataset.path = os.path.relpath(expected_path, client.path)
if not expected_path.exists():
shutil.move(old_dataset_path, expected_path)
for file_ in dataset.files:
file_path = Path(file_.path)
if not file_path.exists() or file_.path.startswith(".."):
new_path = Path(
os.path.abspath(client.renku_datasets_path /
dataset.identifier /
file_path)).relative_to(client.path)
file_.path = new_path
file_.name = os.path.basename(file_.path)
dataset.to_yaml(expected_path / client.METADATA)
def set_name(client, name, path, force):
"""Sets the <name> for remote <path>."""
from renku.core.models.refs import LinkReference
LinkReference.create(client=client, name=_ref(name),
force=force).set_reference(path)
