def parse_pid_dataobject_path(self, metadata, key='URL'):
""" Parse url / irods path """
url = metadata.get(key)
if url is None:
return url
# NOTE: this would only work until the protocol is unchanged
url = url.replace('irods://', '')
# path_pieces = url.split(path.os.sep)[1:]
path_pieces = url.split(path.os.sep)
path_pieces[0] = path.os.sep
# TEMPORARY FIX, waiting to decide final PID structure
try:
if path_pieces[3] == 'api' and path_pieces[4] == 'registered':
path_pieces[0] = "/"
path_pieces[1] = "/"
path_pieces[2] = "/"
path_pieces[3] = "/"
path_pieces[4] = "/"
except BaseException:
log.error("Error parsing URL, not enough tokens? %s", path_pieces)
# print("pieces", path_pieces)
ipath = str(path.build(path_pieces))
log.verbose("Data object: %s", ipath)
return ipath
def connect_client(self, force_no_credentials=False, disable_logs=False):
if getattr(self, '_handle_client', None) is None:
if disable_logs:
import logging
logging.getLogger('b2handle').setLevel(logging.WARNING)
# With credentials
if force_no_credentials:
self._handle_client = b2handle.instantiate_for_read_access()
log.debug("HANDLE client connected [w/out credentials]")
else:
found = False
file = os.environ.get('HANDLE_CREDENTIALS', None)
if file is not None:
from utilities import path
credentials_path = path.build(file)
found = path.file_exists_and_nonzero(credentials_path)
if not found:
log.warning("B2HANDLE credentials file not found %s",
file)
if found:
self._handle_client = \
b2handle.instantiate_with_credentials(
credentials.load_from_JSON(file)
)
log.debug("HANDLE client connected [w/ credentials]")
return self._handle_client, True
return self._handle_client, False
def get_ingestion_path_on_host(self, batch_id):
'''
Return the path where the data is located
on the Rancher host.
The parts of the path can be configured,
see: RESOURCES_LOCALPATH=/usr/share
see: SEADATA_WORKSPACE_INGESTION=ingestion
Example: /usr/share/ingestion/<batch_id>
'''
paths = [self._handle._localpath] # "/usr/share" (default)
paths.append(INGESTION_DIR) # "batches" (default)
paths.append(batch_id)
return str(path.build(paths))
def get_irods_path(self, irods_client, mypath, suffix=None):
'''
Helper to construct a path of a data object
inside irods.
Note: Helper, only used inside this file.
Note: The irods_client is of class
IrodsPythonClient, defined in module
rapydo/http-api/restapi/flask_ext/flask_irods/client
'''
paths = [mypath]
if suffix is not None:
paths.append(suffix)
from utilities import path
suffix_path = str(path.build(paths))
return irods_client.get_current_zone(suffix=suffix_path)
def get_ingestion_path_in_container(self):
'''
Return the path where the data is located
mounted inside the Rancher containers.
The start of the path can be configured,
see: RESOURCES_LOCALPATH=/usr/local
The directory name is fixed.
Note: The batch_id is not part of the path,
as every container only works on one batch
anyway. With every batch being mounted into
the same path, the programs inside the container
can easily operate on whichever data is inside
that directory.
Example: /usr/share/batch/
'''
paths = [FS_PATH_IN_CONTAINER] # "/usr/share/batch" (hard-coded)
return str(path.build(paths))
def connect_client(self, force_no_credentials=False):
found = False
# With credentials
if not force_no_credentials:
file = os.environ.get('HANDLE_CREDENTIALS', None)
if file is not None:
from utilities import path
credentials_path = path.build(file)
found = path.file_exists_and_nonzero(credentials_path)
if not found:
log.warning("B2HANDLE credentials file not found %s", file)
if found:
client = b2handle.instantiate_with_credentials(
credentials.load_from_JSON(file))
log.info("PID client connected: w/ credentials")
return client, True
client = b2handle.instantiate_for_read_access()
log.warning("PID client connected: NO credentials")
return client, False
def join_paths(self, paths):
return str(path.build(paths))
def get_batch_path(self, icom, batch_id):
from utilities import path
suffix_path = str(path.build(['batchs', batch_id]))
return icom.get_current_zone(suffix=suffix_path)