def test_version_rest_publish_zarr(
api_client,
user: User,
draft_version: Version,
draft_asset_factory,
zarr_archive_factory,
):
assign_perm('owner', user, draft_version.dandiset)
api_client.force_authenticate(user=user)
zarr_archive = zarr_archive_factory(dandiset=draft_version.dandiset)
zarr_asset: Asset = draft_asset_factory(zarr=zarr_archive, blob=None)
normal_asset: Asset = draft_asset_factory()
draft_version.assets.add(zarr_asset)
draft_version.assets.add(normal_asset)
# Validate the metadata to mark the assets and version as `VALID`
tasks.validate_asset_metadata(zarr_asset.id)
tasks.validate_asset_metadata(normal_asset.id)
tasks.validate_version_metadata(draft_version.id)
draft_version.refresh_from_db()
assert draft_version.valid
resp = api_client.post(
f'/api/dandisets/{draft_version.dandiset.identifier}'
f'/versions/{draft_version.version}/publish/'
)
assert resp.status_code == 400
assert resp.json() == ['Cannot publish dandisets which contain zarrs']
def revalidate(assets: bool, versions: bool, revalidate_all: bool,
dry_run: bool):
"""
Revalidate all Versions and Assets.
This script will run the validation immediately in band without dispatching tasks to the queue.
"""
if assets:
asset_qs = Asset.objects
if not revalidate_all:
asset_qs = asset_qs.filter(status=Asset.Status.INVALID)
click.echo(f'Revalidating {asset_qs.count()} assets')
if not dry_run:
for asset in asset_qs.values('id'):
validate_asset_metadata(asset['id'])
if versions:
# Only revalidate draft versions
version_qs = Version.objects.filter(version='draft')
if not revalidate_all:
version_qs = version_qs.filter(status=Version.Status.INVALID, )
click.echo(f'Revalidating {version_qs.count()} versions')
if not dry_run:
for version in version_qs.values('id'):
validate_version_metadata(version['id'])
def create_dev_dandiset(name: str, owner: str):
owner = User.objects.get(email=owner)
# Create a new dandiset
dandiset = Dandiset()
dandiset.save()
dandiset.add_owner(owner)
# Create the draft version
version_metadata = {
'schemaVersion':
settings.DANDI_SCHEMA_VERSION,
'schemaKey':
'Dandiset',
'description':
'An informative description',
'license': ['spdx:CC0-1.0'],
'contributor': [
{
'name': f'{owner.last_name}, {owner.first_name}',
'email': owner.email,
'roleName': ['dcite:ContactPerson'],
'schemaKey': 'Person',
'affiliation': [],
'includeInCitation': True,
},
],
}
draft_version = Version(
dandiset=dandiset,
name=name,
metadata=version_metadata,
version='draft',
)
draft_version.save()
uploaded_file = SimpleUploadedFile(name='foo/bar.txt', content=b'A' * 20)
etag = '76d36e98f312e98ff908c8c82c8dd623-0'
try:
asset_blob = AssetBlob.objects.get(etag=etag)
except AssetBlob.DoesNotExist:
asset_blob = AssetBlob(
blob_id=uuid4(),
blob=uploaded_file,
etag=etag,
size=20,
)
asset_blob.save()
asset_metadata = {
'schemaVersion': settings.DANDI_SCHEMA_VERSION,
'encodingFormat': 'text/plain',
'schemaKey': 'Asset',
}
asset = Asset(blob=asset_blob, metadata=asset_metadata, path='foo/bar.txt')
asset.save()
draft_version.assets.add(asset)
calculate_sha256(blob_id=asset_blob.blob_id)
validate_asset_metadata(asset_id=asset.id)
validate_version_metadata(version_id=draft_version.id)
def test_unembargo_dandiset_existing_blobs(
dandiset_factory,
draft_version_factory,
asset_factory,
asset_blob_factory,
embargoed_asset_blob_factory,
storage_tuple,
):
# Pretend like AssetBlob/EmbargoedAssetBlob were defined with the given storage
storage, embargoed_storage = storage_tuple
AssetBlob.blob.field.storage = storage
EmbargoedAssetBlob.blob.field.storage = embargoed_storage
# Create dandiset and version
dandiset: Dandiset = dandiset_factory(embargo_status=Dandiset.EmbargoStatus.EMBARGOED)
draft_version: Version = draft_version_factory(dandiset=dandiset)
# Create embargoed assets
embargoed_asset_blob: EmbargoedAssetBlob = embargoed_asset_blob_factory()
embargoed_asset: Asset = asset_factory(embargoed_blob=embargoed_asset_blob, blob=None)
draft_version.assets.add(embargoed_asset)
# Create unembargoed asset with identical data
embargoed_asset_blob_data = embargoed_asset_blob.blob.read()
embargoed_asset_blob.blob.seek(0)
existing_asset_blob = asset_blob_factory(
blob=factory.django.FileField(data=embargoed_asset_blob_data)
)
# Assert properties before unembargo
assert embargoed_asset.embargoed_blob is not None
assert embargoed_asset.blob is None
assert embargoed_asset.embargoed_blob.etag != ''
assert existing_asset_blob.etag != ''
assert embargoed_asset_blob.etag == existing_asset_blob.etag
# Run unembargo
tasks.unembargo_dandiset(dandiset.pk)
tasks.validate_version_metadata(draft_version.pk)
dandiset.refresh_from_db()
draft_version.refresh_from_db()
# Assert correct changes took place
assert dandiset.embargo_status == Dandiset.EmbargoStatus.OPEN
assert draft_version.status == Version.Status.VALID
assert draft_version.metadata['access'] == [
{'schemaKey': 'AccessRequirements', 'status': 'dandi:OpenAccess'}
]
# Assert no new asset created
asset: Asset = draft_version.assets.first()
assert asset == embargoed_asset
# Check blobs
assert asset.embargoed_blob is None
assert asset.blob is not None
assert asset.blob.etag == embargoed_asset_blob.etag
assert asset.blob == existing_asset_blob
def test_validate_version_metadata(version: Version, asset: Asset):
version.assets.add(asset)
tasks.validate_version_metadata(version.id)
version.refresh_from_db()
assert version.status == Version.Status.VALID
assert version.validation_errors == []
def test_validate_version_metadata_malformed_schema_version(
version: Version, asset: Asset):
version.assets.add(asset)
version.metadata['schemaVersion'] = 'xxx'
version.save()
tasks.validate_version_metadata(version.id)
version.refresh_from_db()
assert version.status == Version.Status.INVALID
assert len(version.validation_errors) == 1
assert version.validation_errors[0]['message'].startswith(
'Metadata version xxx is not allowed.')
def test_unembargo_dandiset_normal_asset_blob(
dandiset_factory,
draft_version_factory,
asset_factory,
asset_blob_factory,
storage,
):
# Pretend like AssetBlob was defined with the given storage
AssetBlob.blob.field.storage = storage
# Create dandiset and version
dandiset: Dandiset = dandiset_factory(embargo_status=Dandiset.EmbargoStatus.EMBARGOED)
draft_version: Version = draft_version_factory(dandiset=dandiset)
# Create asset
asset_blob: AssetBlob = asset_blob_factory()
asset: Asset = asset_factory(blob=asset_blob, embargoed_blob=None)
draft_version.assets.add(asset)
# Assert properties before unembargo
assert asset.embargoed_blob is None
assert asset.blob is not None
# Run unembargo
tasks.unembargo_dandiset(dandiset.pk)
tasks.validate_version_metadata(draft_version.pk)
dandiset.refresh_from_db()
draft_version.refresh_from_db()
# Assert correct changes took place
assert dandiset.embargo_status == Dandiset.EmbargoStatus.OPEN
assert draft_version.status == Version.Status.VALID
assert draft_version.metadata['access'] == [
{'schemaKey': 'AccessRequirements', 'status': 'dandi:OpenAccess'}
]
# Assert no new asset created
fetched_asset: Asset = draft_version.assets.first()
assert asset == fetched_asset
# Check that blob is unchanged
assert fetched_asset.blob == asset_blob
assert asset.embargoed_blob is None
assert asset.blob is not None
assert asset.blob.etag
assert asset.blob
def test_validate_version_metadata_malformed_license(version: Version,
asset: Asset):
version.assets.add(asset)
version.metadata['license'] = 'foo'
version.save()
tasks.validate_version_metadata(version.id)
version.refresh_from_db()
assert version.status == Version.Status.INVALID
assert version.validation_errors == [{
'field':
'license',
'message':
"'foo' is not of type 'array'"
}]
def test_validate_version_metadata_no_description(version: Version,
asset: Asset):
version.assets.add(asset)
del version.metadata['description']
version.save()
tasks.validate_version_metadata(version.id)
version.refresh_from_db()
assert version.status == Version.Status.INVALID
assert version.validation_errors == [{
'field':
'',
'message':
"'description' is a required property"
}]
def test_version_rest_publish_assets(
api_client,
user: User,
draft_version: Version,
draft_asset_factory,
published_asset_factory,
):
assign_perm('owner', user, draft_version.dandiset)
api_client.force_authenticate(user=user)
old_draft_asset: Asset = draft_asset_factory()
old_published_asset: Asset = published_asset_factory()
old_published_asset.publish()
old_published_asset.save()
assert not old_draft_asset.published
assert old_published_asset.published
draft_version.assets.add(old_draft_asset)
draft_version.assets.add(old_published_asset)
# Validate the metadata to mark the assets and version as `VALID`
tasks.validate_asset_metadata(old_draft_asset.id)
tasks.validate_asset_metadata(old_published_asset.id)
tasks.validate_version_metadata(draft_version.id)
draft_version.refresh_from_db()
assert draft_version.valid
resp = api_client.post(
f'/api/dandisets/{draft_version.dandiset.identifier}'
f'/versions/{draft_version.version}/publish/'
)
assert resp.status_code == 200
published_version = Version.objects.get(version=resp.data['version'])
assert published_version.assets.count() == 2
new_draft_asset: Asset = published_version.assets.get(asset_id=old_draft_asset.asset_id)
new_published_asset: Asset = published_version.assets.get(asset_id=old_published_asset.asset_id)
# The former draft asset should have been modified into a published asset
assert new_draft_asset.published
assert new_draft_asset.asset_id == old_draft_asset.asset_id
assert new_draft_asset.path == old_draft_asset.path
assert new_draft_asset.blob == old_draft_asset.blob
assert new_draft_asset.metadata == {
**old_draft_asset.metadata,
'datePublished': UTC_ISO_TIMESTAMP_RE,
'publishedBy': {
'id': URN_RE,
'name': 'DANDI publish',
'startDate': UTC_ISO_TIMESTAMP_RE,
# TODO endDate needs to be defined before publish is complete
'endDate': UTC_ISO_TIMESTAMP_RE,
'wasAssociatedWith': [
{
'id': URN_RE,
'identifier': 'RRID:SCR_017571',
'name': 'DANDI API',
'version': '0.1.0',
'schemaKey': 'Software',
}
],
'schemaKey': 'PublishActivity',
},
}
# The published_asset should be completely unchanged
assert new_published_asset.published
assert new_published_asset.asset_id == old_published_asset.asset_id
assert new_published_asset.path == old_published_asset.path
assert new_published_asset.blob == old_published_asset.blob
assert new_published_asset.metadata == old_published_asset.metadata
def test_version_rest_publish(api_client, user: User, draft_version: Version, asset: Asset):
assign_perm('owner', user, draft_version.dandiset)
api_client.force_authenticate(user=user)
draft_version.assets.add(asset)
# Validate the metadata to mark the version and asset as `VALID`
tasks.validate_version_metadata(draft_version.id)
tasks.validate_asset_metadata(asset.id)
draft_version.refresh_from_db()
assert draft_version.valid
resp = api_client.post(
f'/api/dandisets/{draft_version.dandiset.identifier}'
f'/versions/{draft_version.version}/publish/'
)
assert resp.data == {
'dandiset': {
'identifier': draft_version.dandiset.identifier,
'created': TIMESTAMP_RE,
'modified': TIMESTAMP_RE,
'contact_person': draft_version.metadata['contributor'][0]['name'],
'embargo_status': 'OPEN',
},
'version': VERSION_ID_RE,
'name': draft_version.name,
'created': TIMESTAMP_RE,
'modified': TIMESTAMP_RE,
'asset_count': 1,
'size': draft_version.size,
'status': 'Valid',
}
published_version = Version.objects.get(version=resp.data['version'])
assert published_version
assert draft_version.dandiset.versions.count() == 2
published_asset: Asset = published_version.assets.get()
assert published_asset.published
# The asset should be the same after publishing
assert asset.asset_id == published_asset.asset_id
assert published_version.metadata == {
**draft_version.metadata,
'publishedBy': {
'id': URN_RE,
'name': 'DANDI publish',
'startDate': UTC_ISO_TIMESTAMP_RE,
'endDate': UTC_ISO_TIMESTAMP_RE,
'wasAssociatedWith': [
{
'id': URN_RE,
'identifier': 'RRID:SCR_017571',
'name': 'DANDI API',
# TODO version the API
'version': '0.1.0',
'schemaKey': 'Software',
}
],
'schemaKey': 'PublishActivity',
},
'datePublished': UTC_ISO_TIMESTAMP_RE,
'manifestLocation': [
f'http://{settings.MINIO_STORAGE_ENDPOINT}/test-dandiapi-dandisets/test-prefix/dandisets/{draft_version.dandiset.identifier}/{published_version.version}/assets.yaml', # noqa: E501
],
'identifier': f'DANDI:{draft_version.dandiset.identifier}',
'version': published_version.version,
'id': f'DANDI:{draft_version.dandiset.identifier}/{published_version.version}',
'url': (
f'{settings.DANDI_WEB_APP_URL}/dandiset/{draft_version.dandiset.identifier}'
f'/{published_version.version}'
),
'citation': published_version.citation(published_version.metadata),
'doi': f'10.80507/dandi.{draft_version.dandiset.identifier}/{published_version.version}',
# Once the assets are linked, assetsSummary should be computed properly
'assetsSummary': {
'schemaKey': 'AssetsSummary',
'numberOfBytes': 100,
'numberOfFiles': 1,
'dataStandard': [
{
'schemaKey': 'StandardsType',
'identifier': 'RRID:SCR_015242',
'name': 'Neurodata Without Borders (NWB)',
}
],
'approach': [],
'measurementTechnique': [],
'variableMeasured': [],
'species': [],
},
}
draft_version.refresh_from_db()
assert draft_version.status == Version.Status.PUBLISHED
assert not draft_version.valid
def test_unembargo_dandiset(
dandiset_factory,
draft_version_factory,
asset_factory,
embargoed_asset_blob_factory,
storage_tuple,
file_size,
part_size,
monkeypatch,
):
# Pretend like AssetBlob/EmbargoedAssetBlob were defined with the given storage
storage, embargoed_storage = storage_tuple
monkeypatch.setattr(AssetBlob.blob.field, 'storage', storage)
monkeypatch.setattr(EmbargoedAssetBlob.blob.field, 'storage', embargoed_storage)
# Monkey patch PartGenerator so that upload and copy use a smaller part size
monkeypatch.setattr(PartGenerator, 'DEFAULT_PART_SIZE', part_size, raising=True)
# Create dandiset and version
dandiset: Dandiset = dandiset_factory(embargo_status=Dandiset.EmbargoStatus.EMBARGOED)
draft_version: Version = draft_version_factory(dandiset=dandiset)
# Create an embargoed asset blob
embargoed_asset_blob: EmbargoedAssetBlob = embargoed_asset_blob_factory(
size=file_size, blob=SimpleUploadedFile('test', content=os.urandom(file_size))
)
# Assert multiple parts were used
num_parts = math.ceil(file_size / part_size)
assert embargoed_asset_blob.etag.endswith(f'-{num_parts}')
# Create asset from embargoed blob
embargoed_asset: Asset = asset_factory(embargoed_blob=embargoed_asset_blob, blob=None)
draft_version.assets.add(embargoed_asset)
# Assert properties before unembargo
assert embargoed_asset.embargoed_blob is not None
assert embargoed_asset.blob is None
assert embargoed_asset.embargoed_blob.etag != ''
# Run unembargo and validate version metadata
tasks.unembargo_dandiset(dandiset.pk)
tasks.validate_version_metadata(draft_version.pk)
dandiset.refresh_from_db()
draft_version.refresh_from_db()
# Assert correct changes took place
assert dandiset.embargo_status == Dandiset.EmbargoStatus.OPEN
assert draft_version.status == Version.Status.VALID
assert draft_version.metadata['access'] == [
{'schemaKey': 'AccessRequirements', 'status': 'dandi:OpenAccess'}
]
# Assert no new asset created
asset: Asset = draft_version.assets.first()
assert asset == embargoed_asset
# Check blobs
assert asset.embargoed_blob is None
assert asset.blob is not None
assert asset.blob.etag == embargoed_asset_blob.etag
blob_id = str(asset.blob.blob_id)
assert asset.blob.blob.name == f'test-prefix/blobs/{blob_id[:3]}/{blob_id[3:6]}/{blob_id}'