def test_parent(self):
root_path = FigsharePath('/', _ids=('', ), folder=True, is_public=False)
folder_path = FigsharePath('/folder/', _ids=('', '142132'), folder=True, is_public=False)
file_path = FigsharePath('/folder/test.txt',
_ids=('', '142132', '1595252'), folder=False, is_public=False)
assert root_path.parent is None
assert folder_path.parent == root_path
assert file_path.parent == folder_path
path = file_path.parent
assert path.identifier_path == '142132/'
assert path.identifier == '142132'
assert path.is_dir is True
assert path.is_folder is True
assert path.kind == 'folder'
assert path.name == 'folder'
assert path.ext == ''
assert path.path == 'folder/'
assert str(path) == '/folder/'
assert path.raw_path == 'folder/'
assert path.full_path == 'folder/'
assert path.materialized_path == '/folder/'
assert path.extra == {}
async def validate_v1_path(self, path, **kwargs):
"""Take a string path from the url and attempt to map it to an entity within this article.
If the entity is found, returns a FigsharePath object with the entity identifiers included.
Otherwise throws a 404 Not Found. Will also assert that the entity type inferred from the
path matches the type of the entity at that url.
:param str path: entity path from the v1 API
:rtype FigsharePath:
"""
if path == '/':
return FigsharePath('/', _ids=('', ), folder=True, is_public=False)
path_parts = self._path_split(path)
if len(path_parts) != 2:
raise exceptions.InvalidPathError(
'{} is not a valid Figshare path.'.format(path))
file_id = path_parts[1]
resp = await self.make_request(
'GET',
self.build_url(False, *self.root_path_parts, 'files', file_id),
expects=(200, ),
)
file_json = await resp.json()
return FigsharePath('/' + file_json['name'],
_ids=('', file_id),
folder=False,
is_public=False)
async def upload(self, stream, path, conflict='replace', **kwargs):
"""Upload a file to provider root or to an article whose defined_type is
configured to represent a folder.
:param asyncio.StreamReader stream: stream to upload
:param FigsharePath path: FigsharePath to upload the file to.
:param dict \*\*kwargs: Will be passed to returned metadata object
"""
path, exists = await self.handle_name_conflict(path, conflict=conflict)
if not path.parent.is_root:
parent_resp = await self.make_request(
'GET',
self.build_url(False, *self.root_path_parts, 'articles',
path.parent.identifier),
expects=(200, ),
)
parent_json = await parent_resp.json()
if not parent_json['defined_type'] in settings.FOLDER_TYPES:
del path._parts[1]
stream.add_writer('md5', streams.HashStreamWriter(hashlib.md5))
file_id = await self._upload_file(self.container_id, path.name, stream)
# Build new file path and return metadata
path = FigsharePath('/' + file_id,
_ids=('', file_id),
folder=False,
is_public=False)
metadata = await self.metadata(path, **kwargs)
if stream.writers['md5'].hexdigest != metadata.extra['hashes']['md5']:
raise exceptions.UploadChecksumMismatchError()
return metadata, True
async def validate_path(self, path, **kwargs):
"""Take a string path from the url and attempt to map it to an entity within this article.
If the entity is found, returns a FigsharePath object with the entity identifiers included.
Otherwise returns a FigsharePath with empty identifiers.
:param str path: identifier path URN as passed through the v0 API
:rtype FigsharePath:
Quirks:
* v0 may pass an identifier_path whose last part is a name and not an identifier, in the
case of file/folder creation calls.
* validate_path validates parent and returns a FigsharePath as accurately as possible.
"""
if path == '/':
return FigsharePath('/', _ids=('', ), folder=True, is_public=False)
path_parts = self._path_split(path)
if len(path_parts) != 2:
raise exceptions.InvalidPathError(
'{} is not a valid Figshare path.'.format(path))
file_id = path_parts[1]
resp = await self.make_request(
'GET',
self.build_url(False, *self.root_path_parts, 'files', file_id),
expects=(
200,
404,
),
)
if resp.status == 200:
file_json = await resp.json()
file_name = file_json['name']
return FigsharePath('/' + file_name,
_ids=('', file_id),
folder=False,
is_public=False)
# catch for create file in article root
await resp.release()
return FigsharePath('/' + file_id,
_ids=('', ''),
folder=False,
is_public=False)
async def upload(self, stream, path, conflict='replace', **kwargs):
"""Upload a file to provider root or to an article whose defined_type is
configured to represent a folder.
:param asyncio.StreamReader stream: stream to upload
:param FigsharePath path: FigsharePath to upload the file to.
:param dict \*\*kwargs: Will be passed to returned metadata object
"""
if path.identifier and conflict == 'replace':
raise exceptions.UnsupportedOperationError(
'Files in Figshare cannot be updated')
path, exists = await self.handle_name_conflict(path, conflict=conflict)
if not path.parent.is_root:
parent_resp = await self.make_request(
'GET',
self.build_url(False, *self.root_path_parts, 'articles',
path.parent.identifier),
expects=(200, ),
)
parent_json = await parent_resp.json()
if not parent_json['defined_type'] in settings.FOLDER_TYPES:
del path._parts[1]
# Create article or retrieve article_id from existing article
if not path.parent.is_root:
article_id = path.parent.identifier
else:
article_name = json.dumps({'title': path.name})
if self.container_type == 'project':
article_id = await self._create_article(article_name)
elif self.container_type == 'collection':
# TODO don't think this is correct. Probably should POST to /accounts/articles
article_id = await self._create_article(article_name)
article_list = json.dumps({'articles': [article_id]})
await self.make_request(
'POST',
self.build_url(False, *self.root_path_parts, 'articles'),
data=article_list,
expects=(201, ),
)
stream.add_writer('md5', streams.HashStreamWriter(hashlib.md5))
file_id = await self._upload_file(article_id, path.name, stream)
# Build new file path and return metadata
path = FigsharePath('/' + article_id + '/' + file_id,
_ids=(self.container_id, article_id, file_id),
folder=False,
is_public=False)
metadata = await self.metadata(path, **kwargs)
if stream.writers['md5'].hexdigest != metadata.extra['hashes']['md5']:
raise exceptions.UploadChecksumMismatchError()
return metadata, True
def test_folder_path(self):
path = FigsharePath('/folder/', _ids=('', '142132'), folder=True, is_public=False)
assert path.identifier_path == '142132/'
assert path.identifier == '142132'
assert path.is_dir is True
assert path.is_folder is True
assert path.kind == 'folder'
assert path.name == 'folder'
assert path.ext == ''
assert path.path == 'folder/'
assert str(path) == '/folder/'
assert path.raw_path == 'folder/'
assert path.full_path == 'folder/'
assert path.materialized_path == '/folder/'
assert path.extra == {}
def test_file_path(self):
path = FigsharePath('/folder/test.txt',
_ids=('', '142132', '1595252'), folder=False, is_public=False)
assert path.identifier_path == '142132/1595252'
assert path.identifier == '1595252'
assert path.is_dir is False
assert path.is_folder is False
assert path.kind == 'file'
assert path.name == 'test.txt'
assert path.ext == '.txt'
assert path.path == 'folder/test.txt'
assert str(path) == '/folder/test.txt'
assert path.raw_path == 'folder/test.txt'
assert path.full_path == 'folder/test.txt'
assert path.materialized_path == '/folder/test.txt'
assert path.extra == {}
def test_child(self):
root_path = FigsharePath('/', _ids=('', ), folder=True, is_public=False)
folder_path = FigsharePath('/folder/', _ids=('', '142132'), folder=True, is_public=False)
file_path = FigsharePath('/folder/test.txt',
_ids=('', '142132', '1595252'), folder=False, is_public=False)
child_root_path = root_path.child('folder/', _id='142132', folder=True)
child_folder_path = folder_path.child('test.txt', _id='1595252')
assert child_root_path == folder_path
assert child_folder_path == file_path
assert child_root_path.identifier_path == '142132/'
assert child_root_path.identifier == '142132'
assert child_root_path.is_dir is True
assert child_root_path.is_folder is True
assert child_root_path.kind == 'folder'
assert child_root_path.name == 'folder'
assert child_root_path.ext == ''
assert child_root_path.path == 'folder/'
assert child_root_path.raw_path == 'folder/'
assert child_root_path.full_path == 'folder/'
assert child_root_path.materialized_path == '/folder/'
assert child_root_path.extra == {}
assert child_folder_path.identifier_path == '142132/1595252'
assert child_folder_path.identifier == '1595252'
assert child_folder_path.is_dir is False
assert child_folder_path.is_folder is False
assert child_folder_path.kind == 'file'
assert child_folder_path.name == 'test.txt'
assert child_folder_path.ext == '.txt'
assert child_folder_path.path == 'folder/test.txt'
assert child_folder_path.raw_path == 'folder/test.txt'
assert child_folder_path.full_path == 'folder/test.txt'
assert child_folder_path.materialized_path == '/folder/test.txt'
assert child_folder_path.extra == {}
def test_root_path(self):
root_path = FigsharePath('/', _ids=('', ), folder=True, is_public=False)
assert root_path.identifier_path == ''
async def validate_path(self, path, **kwargs):
"""Take a string path from the url and attempt to map it to an entity within this project.
If the entity is found, returns a FigsharePath object with the entity identifiers included.
Otherwise returns a FigsharePath with empty identifiers.
:param str path: identifier_path URN as passed through the v0 API
:rtype FigsharePath:
Quirks:
* v0 may pass an identifier_path whose last part is a name and not an identifier, in the
case of file/folder creation calls.
* validate_path validates parent and returns a FigsharePath as accurately as possible.
"""
if path == '/':
return FigsharePath('/', _ids=('', ), folder=True, is_public=False)
path_parts = self._path_split(path)
if len(path_parts) not in (2, 3):
raise exceptions.InvalidPathError(
'{} is not a valid Figshare path.'.format(path))
article_id = path_parts[1]
file_id = path_parts[2] if len(path_parts) == 3 else None
articles = await self._get_all_articles()
# TODO: need better way to get public/private
# This call's return value is currently busted at figshare for collections. Figshare always
# returns private-looking urls.
is_public = False
for item in articles:
if '/articles/' + article_id in item['url']:
article_name = item['title']
if settings.PRIVATE_IDENTIFIER not in item['url']:
is_public = True
article_segments = (*self.root_path_parts, 'articles', article_id)
if file_id:
file_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments, 'files', file_id),
expects=(
200,
404,
),
)
if file_response.status == 200:
file_response_json = await file_response.json()
file_name = file_response_json['name']
return FigsharePath('/' + article_name + '/' + file_name,
_ids=(self.container_id, article_id,
file_id),
folder=False,
is_public=is_public)
await file_response.release()
article_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments),
expects=(
200,
404,
),
)
if article_response.status == 200:
article_json = await article_response.json()
if article_json['defined_type'] in settings.FOLDER_TYPES:
# Case of v0 file creation
if file_id:
ids = ('', article_id, '')
folder = False
path_urn = '/' + article_name + '/' + file_id
else:
ids = ('', article_id)
folder = True
path_urn = '/' + article_name + '/'
return FigsharePath(path_urn,
_ids=ids,
folder=folder,
is_public=is_public)
else:
await article_response.release()
if file_id:
# Catch for if neither file nor article exist
raise exceptions.NotFoundError(path)
# Return for v0 folder creation
return FigsharePath(path, _ids=('', ''), folder=True, is_public=False)
async def validate_v1_path(self, path, **kwargs):
"""Take a string path from the url and attempt to map it to an entity within this project.
If the entity is found, returns a FigsharePath object with the entity identifiers included.
Otherwise throws a 404 Not Found. Will also assert that the entity type inferred from the
path matches the type of the entity at that url.
:param str path: entity path from the v1 API
:rtype FigsharePath:
"""
if path == '/':
return FigsharePath('/', _ids=('', ), folder=True, is_public=False)
path_parts = self._path_split(path)
if len(path_parts) not in (2, 3):
raise exceptions.InvalidPathError(
'{} is not a valid Figshare path.'.format(path))
article_id = path_parts[1]
file_id = path_parts[2] if len(path_parts) == 3 else None
articles = await self._get_all_articles()
# TODO: need better way to get public/private
# This call's return value is currently busted at figshare for collections. Figshare always
# returns private-looking urls.
is_public = False
for item in articles:
if '/articles/' + article_id in item['url']:
article_name = item['title']
if settings.PRIVATE_IDENTIFIER not in item['url']:
is_public = True
article_segments = (*self.root_path_parts, 'articles', article_id)
if file_id:
file_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments, 'files', file_id),
expects=(200, ),
)
file_json = await file_response.json()
file_name = file_json['name']
if path[-1] == '/':
raise exceptions.NotFoundError(
'File paths must not end with "/". '
'{} not found.'.format(path))
return FigsharePath('/' + article_name + '/' + file_name,
_ids=(self.container_id, article_id, file_id),
folder=False,
is_public=is_public)
article_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments),
expects=(200, ),
)
article_json = await article_response.json()
if article_json['defined_type'] in settings.FOLDER_TYPES:
if not path[-1] == '/':
raise exceptions.NotFoundError(
'Folder paths must end with "/". {} not found.'.format(
path))
return FigsharePath('/' + article_name + '/',
_ids=(self.container_id, article_id),
folder=True,
is_public=is_public)
raise exceptions.NotFoundError(
'This article is not configured as a folder defined_type. '
'{} not found.'.format(path))
async def validate_v1_path(self, path: str, **kwargs) -> FigsharePath:
"""Take a string path from the url and attempt to map it to an entity within this project.
If the entity is found, returns a FigsharePath object with the entity identifiers included.
Otherwise throws a 404 Not Found. Will also assert that the entity type inferred from the
path matches the type of the entity at that url.
:param str path: entity path from the v1 API
:rtype FigsharePath:
"""
if path == '/':
# Root path should always be private api-wise since project and collection itself must
# be owned by the fighsare-OSF OAuth user.
return FigsharePath('/', _ids=('', ), folder=True, is_public=False)
# Step 0: Preprocess the string path.
path_parts = self._path_split(path)
if len(path_parts) not in (2, 3):
raise exceptions.InvalidPathError('{} is not a valid Figshare path.'.format(path))
article_id = path_parts[1]
file_id = path_parts[2] if len(path_parts) == 3 else None
# Step 1: Get a list of all articles in the project.
articles = await self._get_all_articles()
# Step 2: Find the article; set `article_name`, `is_public`; and prepare `article_segments`.
is_public = False
article_name = None
for article in articles:
if '/articles/' + article_id in article['url']:
article_name = article['title']
is_public = article['published_date'] is not None
break
# Raise error earlier instead of on 404. Please note that this is different than V0.
if not article_name:
raise exceptions.NotFoundError('Path {} with article ID {} not found in the project\'s '
'article list'.format(path, article_id))
article_segments = (*self.root_path_parts, 'articles', article_id)
# Step 3.1: if the path is a file
if file_id:
file_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments, 'files', file_id),
expects=(200, ),
)
file_json = await file_response.json()
file_name = file_json['name']
if path[-1] == '/':
raise exceptions.NotFoundError('File paths must not end with "/". '
'{} not found.'.format(path))
return FigsharePath('/' + article_name + '/' + file_name,
_ids=(self.container_id, article_id, file_id),
folder=False,
is_public=is_public)
# Step 3.2: if the path is a folder
article_response = await self.make_request(
'GET',
self.build_url(is_public, *article_segments),
expects=(200, ),
)
article_json = await article_response.json()
if article_json['defined_type'] in pd_settings.FOLDER_TYPES:
if not path[-1] == '/':
raise exceptions.NotFoundError('Folder paths must end with "/". '
'{} not found.'.format(path))
return FigsharePath('/' + article_name + '/', _ids=(self.container_id, article_id),
folder=True, is_public=is_public)
raise exceptions.NotFoundError('This article is not configured as a folder defined_type. '
'{} not found.'.format(path))
