def intra_copy(self, dest_provider, src_path, dest_path):
if dest_path.identifier:
yield from dest_provider.delete(dest_path)
resp = yield from self.make_signed_request(
'POST',
self.build_url('hooks', 'copy'),
data=json.dumps({
'user': self.auth['id'],
'source': src_path.identifier,
'destination': {
'name': dest_path.name,
'node': dest_provider.nid,
'parent': dest_path.parent.identifier
}
}),
headers={'Content-Type': 'application/json'},
expects=(200, 201)
)
data = yield from resp.json()
if data['kind'] == 'file':
return OsfStorageFileMetadata(data, str(dest_path)), dest_path.identifier is None
return OsfStorageFolderMetadata(data, str(dest_path)), dest_path.identifier is None
async def intra_copy(self, dest_provider, src_path, dest_path):
created = True
if dest_path.identifier:
created = False
await dest_provider.delete(dest_path)
async with self.signed_request(
'POST',
self.build_url('hooks', 'copy'),
data=json.dumps({
'user': self.auth['id'],
'source': src_path.identifier,
'destination': {
'name': dest_path.name,
'node': dest_provider.nid,
'parent': dest_path.parent.identifier
}
}),
headers={'Content-Type': 'application/json'},
expects=(200, 201)) as resp:
data = await resp.json()
if data['kind'] == 'file':
return OsfStorageFileMetadata(
data, str(dest_path)), dest_path.identifier is None
folder_meta = OsfStorageFolderMetadata(data, str(dest_path))
dest_path = await dest_provider.validate_v1_path(data['path'])
folder_meta.children = await dest_provider._children_metadata(dest_path
)
return folder_meta, created
async def _item_metadata(self, path, revision=None):
resp = await self.make_signed_request('GET',
self.build_url(
path.identifier,
revision=revision),
expects=(200, ))
return OsfStorageFileMetadata((await resp.json()), str(path))
def _item_metadata(self, path):
resp = yield from self.make_signed_request(
'GET',
self.build_url(path.identifier),
expects=(200, )
)
return OsfStorageFileMetadata((yield from resp.json()), str(path))
def _children_metadata(self, path):
resp = yield from self.make_signed_request(
'GET',
self.build_url(path.identifier, 'children'),
expects=(200, )
)
resp_json = yield from resp.json()
ret = []
for item in resp_json:
if item['kind'] == 'folder':
ret.append(OsfStorageFolderMetadata(item, str(path.child(item['name']))))
else:
ret.append(OsfStorageFileMetadata(item, str(path.child(item['name']))))
return ret
async def upload(self, stream, path, **kwargs):
"""Upload a new file to osfstorage
When a file is uploaded to osfstorage, WB does a bit of a dance to make sure it gets there
reliably. First we take the stream and add several hash calculators that can determine the
hash of the file as it streams through. We then upload the file to a uuid-named file on the
remote storage provider. Once that's complete, we determine the file's final name, which
will be its sha256 hash. We then check to see if a file already exists at that path on the
remote storage provider. If it does, we can skip moving the file (since it has already been
uploaded) and instead delete the pending file. If it does not, we move the file on the
remote storage provider from the pending path to its final path.
Once this is done the file metadata is sent back to the metadata provider to be recorded.
Finally, WB constructs its metadata response and sends that back to the original request
issuer.
"""
metadata = await self._send_to_storage_provider(stream, path, **kwargs)
metadata = metadata.serialized()
data, created = await self._send_to_metadata_provider(
stream, path, metadata, **kwargs)
name = path.name
metadata.update({
'name':
name,
'md5':
data['data']['md5'],
'path':
data['data']['path'],
'sha256':
data['data']['sha256'],
'version':
data['data']['version'],
'downloads':
data['data']['downloads'],
'checkout':
data['data']['checkout'],
'modified':
data['data']['modified'],
'modified_utc':
utils.normalize_datetime(data['data']['modified']),
})
path._parts[-1]._id = metadata['path'].strip('/')
return OsfStorageFileMetadata(metadata, str(path)), created
async def _children_metadata(self, path):
async with self.signed_request('GET',
self.build_url(path.identifier,
'children'),
expects=(200, )) as resp:
resp_json = await resp.json()
ret = []
for item in resp_json:
if item['kind'] == 'folder':
ret.append(
OsfStorageFolderMetadata(
item, str(path.child(item['name'], folder=True))))
else:
ret.append(
OsfStorageFileMetadata(item,
str(path.child(item['name']))))
return ret
async def _do_intra_move_or_copy(self, action: str, dest_provider,
src_path, dest_path):
"""Update files and folders on osfstorage with a single request.
If the data of the file or the folder's children doesn't need to be copied to another
bucket, then doing an intra-move or intra-copy is just a matter of updating the entity
metadata in the OSF. If something already exists at ``dest_path``, it must be deleted
before relocating the source to the new path.
"""
created = True
if dest_path.identifier:
created = False
await dest_provider.delete(dest_path)
async with self.signed_request(
'POST',
self.build_url('hooks', action),
data=json.dumps({
'user': self.auth['id'],
'source': src_path.identifier,
'destination': {
'name': dest_path.name,
'node': dest_provider.nid,
'parent': dest_path.parent.identifier
}
}),
headers={'Content-Type': 'application/json'},
expects=(200, 201)) as resp:
data = await resp.json()
if data['kind'] == 'file':
return OsfStorageFileMetadata(
data, str(dest_path)), dest_path.identifier is None
folder_meta = OsfStorageFolderMetadata(data, str(dest_path))
dest_path = await dest_provider.validate_v1_path(data['path'])
folder_meta.children = await dest_provider._children_metadata(dest_path
)
return folder_meta, created
async def upload(self, stream, path, **kwargs):
"""Upload a new file to osfstorage
When a file is uploaded to osfstorage, WB does a bit of a dance to make sure it gets there
reliably. First we take the stream and add several hash calculators that can determine the
hash of the file as it streams through. We then tee the file so that it's written to a
"pending" directory on both local disk and the remote storage provider. Once that's
complete, we determine the file's final location, which will be in another directory (by
default called 'complete'), and renamed to its sha256 hash. We then check to see if a
file already exists at that path on the remote storage provider. If it does, we can skip
moving the file (since its already been uploaded) and instead delete the pending file. If
it does not, we move the file on the remote storage provider from the pending path to its
final path.
Once this is done the local copy of the file is moved from the pending directory to the
complete directory. The file metadata is sent back to the metadata provider to be recorded.
Finally, we schedule two futures to archive the locally complete file. One copies the file
into Amazon Glacier, the other calculates a parity archive, so that the file can be
reconstructed if any on-disk corruption happens. These tasks are scheduled via celery and
don't need to complete for the request to finish.
Finally, WB constructs its metadata response and sends that back to the original request
issuer.
The local file sitting in complete will be archived by the celery tasks at some point in
the future. The archivers do not signal when they have finished their task, so for the time
being the local complete files are allowed to accumulate and must be deleted by some
external process. COS currently uses a cron job to delete files older than X days. If the
system is being heavily used, it's possible that the files may be deleted before the
archivers are able to run. To get around this we have another script in the osf.io
repository that can audit our files on the remote storage and initiate any missing archives.
"""
self._create_paths()
pending_name = str(uuid.uuid4())
provider = self.make_provider(self.settings)
local_pending_path = os.path.join(settings.FILE_PATH_PENDING,
pending_name)
remote_pending_path = await provider.validate_path('/' + pending_name)
logger.debug(
'upload: local_pending_path::{}'.format(local_pending_path))
logger.debug(
'upload: remote_pending_path::{}'.format(remote_pending_path))
stream.add_writer('md5', streams.HashStreamWriter(hashlib.md5))
stream.add_writer('sha1', streams.HashStreamWriter(hashlib.sha1))
stream.add_writer('sha256', streams.HashStreamWriter(hashlib.sha256))
try:
with open(local_pending_path, 'wb') as file_pointer:
stream.add_writer('file', file_pointer)
await provider.upload(stream,
remote_pending_path,
check_created=False,
fetch_metadata=False,
**kwargs)
except Exception as exc:
# If we fail to upload to the remote storage provider, then delete the copy of the file
# from the local provider, too. The user will have to reupload the file to local
# anyway, and this will avoid filling up the local disk with unused pending files.
try:
os.remove(local_pending_path)
except OSError as os_exc:
raise exceptions.UploadFailedError(
'Upload failed, please try again.') from os_exc
raise exceptions.UploadFailedError(
'Upload failed, please try again.') from exc
complete_name = stream.writers['sha256'].hexdigest
local_complete_path = os.path.join(settings.FILE_PATH_COMPLETE,
complete_name)
remote_complete_path = await provider.validate_path('/' +
complete_name)
try:
metadata = await provider.metadata(remote_complete_path)
except exceptions.MetadataError as e:
if e.code != 404:
raise
metadata, _ = await provider.move(provider, remote_pending_path,
remote_complete_path)
else:
await provider.delete(remote_pending_path)
metadata = metadata.serialized()
# Due to cross volume movement in unix we leverage shutil.move which properly handles this case.
# http://bytes.com/topic/python/answers/41652-errno-18-invalid-cross-device-link-using-os-rename#post157964
shutil.move(local_pending_path, local_complete_path)
async with self.signed_request(
'POST',
self.build_url(path.parent.identifier, 'children'),
expects=(200, 201),
data=json.dumps({
'name': path.name,
'user': self.auth['id'],
'settings': self.settings['storage'],
'metadata': metadata,
'hashes': {
'md5': stream.writers['md5'].hexdigest,
'sha1': stream.writers['sha1'].hexdigest,
'sha256': stream.writers['sha256'].hexdigest,
},
'worker': {
'host': os.uname()[1],
# TODO: Include additional information
'address': None,
'version': self.__version__,
},
}),
headers={'Content-Type': 'application/json'},
) as response:
created = response.status == 201
data = await response.json()
if settings.RUN_TASKS and data.pop('archive', True):
parity.main(
local_complete_path,
data['version'],
self.build_url('hooks', 'metadata') + '/',
self.parity_credentials,
self.parity_settings,
)
backup.main(
local_complete_path,
data['version'],
self.build_url('hooks', 'metadata') + '/',
self.archive_credentials,
self.archive_settings,
)
name = path.name
metadata.update({
'name':
name,
'md5':
data['data']['md5'],
'path':
data['data']['path'],
'sha256':
data['data']['sha256'],
'version':
data['data']['version'],
'downloads':
data['data']['downloads'],
'checkout':
data['data']['checkout'],
'modified':
data['data']['modified'],
'modified_utc':
utils.normalize_datetime(data['data']['modified']),
})
path._parts[-1]._id = metadata['path'].strip('/')
return OsfStorageFileMetadata(metadata, str(path)), created
def file_metadata_object(file_metadata):
path = WaterButlerPath('/' + file_metadata['name'])
return OsfStorageFileMetadata(file_metadata, path)
def upload(self, stream, path, **kwargs):
self._create_paths()
pending_name = str(uuid.uuid4())
provider = self.make_provider(self.settings)
local_pending_path = os.path.join(settings.FILE_PATH_PENDING, pending_name)
remote_pending_path = yield from provider.validate_path('/' + pending_name)
stream.add_writer('md5', streams.HashStreamWriter(hashlib.md5))
stream.add_writer('sha1', streams.HashStreamWriter(hashlib.sha1))
stream.add_writer('sha256', streams.HashStreamWriter(hashlib.sha256))
with open(local_pending_path, 'wb') as file_pointer:
stream.add_writer('file', file_pointer)
yield from provider.upload(stream, remote_pending_path, check_created=False, fetch_metadata=False, **kwargs)
complete_name = stream.writers['sha256'].hexdigest
local_complete_path = os.path.join(settings.FILE_PATH_COMPLETE, complete_name)
remote_complete_path = yield from provider.validate_path('/' + complete_name)
try:
metadata = yield from provider.metadata(remote_complete_path)
except exceptions.MetadataError as e:
if e.code != 404:
raise
metadata, _ = yield from provider.move(provider, remote_pending_path, remote_complete_path)
else:
yield from provider.delete(remote_pending_path)
finally:
metadata = metadata.serialized()
# Due to cross volume movement in unix we leverage shutil.move which properly handles this case.
# http://bytes.com/topic/python/answers/41652-errno-18-invalid-cross-device-link-using-os-rename#post157964
shutil.move(local_pending_path, local_complete_path)
response = yield from self.make_signed_request(
'POST',
self.build_url(path.parent.identifier, 'children'),
expects=(200, 201),
data=json.dumps({
'name': path.name,
'user': self.auth['id'],
'settings': self.settings['storage'],
'metadata': metadata,
'hashes': {
'md5': stream.writers['md5'].hexdigest,
'sha1': stream.writers['sha1'].hexdigest,
'sha256': stream.writers['sha256'].hexdigest,
},
'worker': {
'host': os.uname()[1],
# TODO: Include additional information
'address': None,
'version': self.__version__,
},
}),
headers={'Content-Type': 'application/json'},
)
created = response.status == 201
data = yield from response.json()
if settings.RUN_TASKS and data.pop('archive', True):
parity.main(
local_complete_path,
self.parity_credentials,
self.parity_settings,
)
backup.main(
local_complete_path,
data['version'],
self.build_url('hooks', 'metadata') + '/',
self.archive_credentials,
self.archive_settings,
)
name = path.name
metadata.update({
'name': name,
'path': data['data']['path'],
'version': data['data']['version'],
'downloads': data['data']['downloads']
})
return OsfStorageFileMetadata(metadata, str(path)), created