def _status_data(upload_db_data: Upload,
upload_workspace: UploadWorkspace) -> dict:
return {
'upload_id': upload_db_data.upload_id,
'upload_total_size': upload_workspace.total_upload_size,
'upload_compressed_size': upload_workspace.content_package_size,
'created_datetime': upload_db_data.created_datetime,
'modified_datetime': upload_db_data.modified_datetime,
'start_datetime': upload_db_data.lastupload_start_datetime,
'completion_datetime': upload_db_data.lastupload_completion_datetime,
'files': json.loads(upload_db_data.lastupload_file_summary),
'errors': json.loads(upload_db_data.lastupload_logs),
'upload_status': upload_db_data.lastupload_upload_status,
'workspace_state': upload_db_data.state,
'lock_state': upload_db_data.lock,
'source_format': upload_workspace.source_format,
'checksum': upload_workspace.content_checksum()
}
def check_upload_content_exists(upload_id: int) -> Response:
"""
Verify that the package content exists/is available.
Parameters
----------
upload_id : int
The unique identifier for upload workspace.
Returns
-------
Standard Response tuple containing response content, HTTP status, and HTTP headers.
"""
try:
upload_db_data: Optional[Upload] = uploads.retrieve(upload_id)
except IOError:
logger.error(
"%s: ContentExistsCheck: There was a problem connecting "
"to database.", upload_id)
raise InternalServerError(UPLOAD_DB_CONNECT_ERROR)
if upload_db_data is None:
raise NotFound(UPLOAD_NOT_FOUND)
logger.info("%s: Upload content summary request.", upload_id)
upload_workspace = UploadWorkspace(upload_id)
# This will potentially build content package if it does not exist
checksum = upload_workspace.content_checksum()
modified = ''
size = 0
# Double check package exists
if upload_workspace.content_package_exists:
modified = upload_workspace.content_package_modified
size = upload_workspace.content_package_size
return {}, status.OK, {
'ETag': checksum,
'Content-Length': size,
'Last-Modified': modified
}
headers = {'ARXIV-OWNER': upload_db_data.owner_user_id, 'ETag': checksum}
return {}, status.OK, headers
def get_upload_content(upload_id: int) -> Response:
"""
Package up files for downloading as a compressed gzipped tar file.
Parameters
----------
upload_id : int
The unique identifier for upload workspace.
Returns
-------
Standard Response tuple containing compressed content, HTTP status, and HTTP headers.
"""
try:
upload_db_data: Optional[Upload] = uploads.retrieve(upload_id)
except IOError:
logger.error(
"%s: ContentDownload: There was a problem connecting "
"to database.", upload_id)
raise InternalServerError(UPLOAD_DB_CONNECT_ERROR)
if upload_db_data is None:
raise NotFound(UPLOAD_NOT_FOUND)
upload_workspace = UploadWorkspace(upload_id)
checksum = upload_workspace.content_checksum()
try:
filepointer = upload_workspace.get_content()
except FileNotFoundError as e:
raise NotFound("No content in workspace") from e
headers = {
"Content-disposition": f"filename={filepointer.name}",
'ETag': checksum,
'ARXIV-OWNER': upload_db_data.owner_user_id
}
return filepointer, status.OK, headers
def client_delete_all_files(upload_id: int) -> Response:
"""
Delete all files uploaded by client from specified workspace.
This request is being received from API so we need to be extra careful.
Parameters
----------
upload_id : int
The unique identifier for the upload_db_data in question.
public_file_path: str
relative path of file to be deleted.
Returns
-------
dict
Complete summary of upload processing.
int
An HTTP status code.
dict
Some extra headers to add to the response.
"""
logger.info("%s: Deleting all uploaded files from this workspace.",
upload_id)
try:
# Make sure we have an upload_db_data to work with
upload_db_data: Optional[Upload] = uploads.retrieve(upload_id)
if upload_db_data is None:
# Invalid workspace identifier
raise NotFound(UPLOAD_NOT_FOUND)
if upload_db_data.state != Upload.ACTIVE:
# Do we log anything for these requests
raise Forbidden(UPLOAD_NOT_ACTIVE)
if upload_db_data.lock == Upload.LOCKED:
raise Forbidden(UPLOAD_WORKSPACE_LOCKED)
# Create Upload object
upload_workspace = UploadWorkspace(upload_id)
upload_workspace.client_remove_all_files()
except IOError:
logger.error("%s: Delete all files request failed ", upload_id)
raise InternalServerError(CANT_DELETE_ALL_FILES)
except NotFound as nf:
logger.info("%s: DeleteAllFiles: '%s'", upload_id, nf)
raise
except Forbidden as forb:
logger.info("%s: Upload failed: '%s'.", upload_id, forb)
raise forb
except Exception as ue:
logger.info(
"Unknown error in delete all files. "
" Add except clauses for '%s'. DO IT NOW!", ue)
raise InternalServerError(UPLOAD_UNKNOWN_ERROR)
response_data = _status_data(upload_db_data, upload_workspace)
response_data.update({
'reason': UPLOAD_DELETED_ALL_FILES,
'checksum': upload_workspace.content_checksum()
}) # Get rid of pylint error
headers = {'ARXIV-OWNER': upload_db_data.owner_user_id}
return response_data, status.OK, headers
def client_delete_file(upload_id: int, public_file_path: str) -> Response:
"""Delete a single file.
This request is being received from API so we need to be extra careful.
Parameters
----------
upload_id : int
The unique identifier for the upload_db_data in question.
public_file_path: str
relative path of file to be deleted.
Returns
-------
dict
Complete summary of upload processing.
int
An HTTP status code.
dict
Some extra headers to add to the response.
"""
logger.info("%s: Delete file '%s'.", upload_id, public_file_path)
try:
# Make sure we have an upload_db_data to work with
upload_db_data: Optional[Upload] = uploads.retrieve(upload_id)
if upload_db_data is None:
# Invalid workspace identifier
raise NotFound(UPLOAD_NOT_FOUND)
if upload_db_data.state != Upload.ACTIVE:
# Do we log anything for these requests
raise Forbidden(UPLOAD_NOT_ACTIVE)
if upload_db_data.lock == Upload.LOCKED:
raise Forbidden(UPLOAD_WORKSPACE_LOCKED)
# Create Upload object
upload_workspace = UploadWorkspace(upload_id)
# Call routine that will do the actual work
upload_workspace.client_remove_file(public_file_path)
except IOError:
logger.error("%s: Delete file request failed ", upload_id)
raise InternalServerError(CANT_DELETE_FILE)
except NotFound as nf:
logger.info("%s: DeleteFile: %s", upload_id, nf)
raise nf
except SecurityError as secerr:
logger.info("%s: %s", upload_id, secerr.description)
# TODO: Should this be BadRequest or NotFound. I'm leaning towards
# NotFound in order to provide as little feedback as posible to client.
raise NotFound(UPLOAD_FILE_NOT_FOUND)
except Forbidden as forb:
logger.info("%s: Delete file forbidden: %s.", upload_id, forb)
raise forb
except Exception as ue:
logger.info(
"Unknown error in delete file. "
" Add except clauses for '%s'. DO IT NOW!", ue)
raise InternalServerError(UPLOAD_UNKNOWN_ERROR)
response_data = _status_data(upload_db_data, upload_workspace)
response_data.update({
'reason': UPLOAD_DELETED_FILE,
'checksum': upload_workspace.content_checksum()
}) # Get rid of pylint errorT
headers = {'ARXIV-OWNER': upload_db_data.owner_user_id}
return response_data, status.OK, headers