def migrate_version(idx, node_file, node_settings, node=None, dry_run=True):
"""Migrate a legacy file version to OSF Storage. If `node` is provided, use
instead of the `Node` attached to `node_settings`; used when the git repo
for the current node is missing or corrupt.
:param int idx: Version index (zero-based)
:param NodeFile node_file: Legacy file record
:param OsfStorageNodeSettings node_settings: Node settings
:param Node node: Optional source node
"""
node = node or node_settings.owner
logger.info('Migrating version {0} from NodeFile {1} on node {2}'.format(
idx,
node_file._id,
node._id,
))
if not dry_run:
record = model.OsfStorageFileRecord.get_or_create(
node_file.path, node_settings)
if len(record.versions) > idx:
return
content = scripts_settings.SPECIAL_CASES.get((node._id, node_file._id))
if content is None:
content, _ = node.read_file_object(node_file)
logger.info('Loaded content with length {0}: {1}...'.format(
len(content), content[:10]))
if dry_run:
return
md5 = hashlib.md5(content).hexdigest()
file_pointer = StringIO(content)
hash_str = scripts_settings.UPLOAD_PRIMARY_HASH(content).hexdigest()
obj = container.get_or_upload_file(file_pointer, hash_str)
if obj.size != len(content):
raise SizeMismatchError
if obj.md5 != md5:
raise HashMismatchError
metadata = {
'size': obj.size,
'content_type': obj.content_type,
'date_modified': obj.date_modified.isoformat(),
'md5': md5,
}
try:
record.create_pending_version(node_file.uploader,
'{}-{}'.format(hash_str, idx))
except errors.OsfStorageError:
latest_version = record.get_version(required=True)
record.remove_version(latest_version)
record = OsfStorageFileRecord.get_or_create(node_file.path,
node_settings)
record.create_pending_version(node_file.uploader,
'{}-{}'.format(hash_str, idx))
record.resolve_pending_version(
'{}-{}'.format(hash_str, idx),
obj.location,
metadata,
log=False,
)
def test_migrate_incomplete(self):
node_settings = self.project.get_or_add_addon('osfstorage', auth=None, log=False)
record = model.OsfStorageFileRecord.get_or_create('pizza.md', node_settings)
node_file = NodeFile.load(self.project.files_versions['pizza_md'][0])
content, _ = self.project.read_file_object(node_file)
file_pointer = StringIO(content)
hash_str = scripts_settings.UPLOAD_PRIMARY_HASH(content).hexdigest()
record.create_pending_version(node_file.uploader, hash_str)
main(dry_run=False)