def main(argv):
parser = argparse.ArgumentParser(description='Program to create a BDBag containing a set of Minids for remote content')
parser.add_argument('-m', '--minids', metavar='<minid file>',
help='File listing Minids for new bag', required=True)
parser.add_argument('-b', '--bagname', metavar='<bag name>',
help='Name of directory for new bag.', required=True)
parser.add_argument('-v', '--verify', action='store_true',
help='Validate bag after building it.', required=False)
parser.add_argument('-q', '--quiet', action="store_true", help="Suppress logging output.")
parser.add_argument('-d', '--debug', action="store_true", help="Enable debug logging output.")
parser.add_argument('-n', '--author-name', metavar="<person or entity name>",
help="Optional name of the person or entity responsible for the creation of this bag, "
"for inclusion in the bag metadata.")
parser.add_argument('-o', '--author-orcid', metavar="<orcid>",
help="Optional ORCID identifier of the bag creator, for inclusion in the bag metadata.")
args = parser.parse_args()
bdb.configure_logging(level=logging.ERROR if args.quiet else (logging.DEBUG if args.debug else logging.INFO))
# Create the directory that will hold the new BDBag
bdb.ensure_bag_path_exists(args.bagname)
# For each supplied minid, fetch sub-bag to determine its properties
minid_fields = extract_fields(args.minids)
# Create 'README' file in the newly created bag directory. (moved to 'data' when bag is created)
write_readme(args.bagname, minid_fields)
# Create remote_file_manifest_file, to be used by make_bag
working_dir = temp_path = tempfile.mkdtemp(prefix='encode2bag_')
remote_file_manifest_file = osp.abspath(osp.join(working_dir, 'remote-file-manifest.json'))
generate_remote_manifest_file(minid_fields, remote_file_manifest_file)
# Create the new bag based on the supplied remote manifest file
bag = bdb.make_bag(args.bagname,
algs=['md5', 'sha256'],
remote_file_manifest=remote_file_manifest_file)
# Create metadata/manifest.json file with Research Object JSON object
ro_manifest = ro.init_ro_manifest(author_name=args.author_name, author_orcid=args.author_orcid,
creator_name = 'bagofbags using BDBag version: %s (Bagit version: %s)' % (VERSION, BAGIT_VERSION),
creator_uri='https://github.com/fair-research/bdbag/examples/bagofbags/')
add_remote_file_manifest_to_ro(ro_manifest, minid_fields)
ro.write_bag_ro_metadata(ro_manifest, args.bagname, 'manifest.json')
# Run make_bag again to include manifest.json in the checksums etc.
bdb.make_bag(args.bagname, update=True)
if args.verify:
bdb.resolve_fetch(args.bagname, force=True)
bdb.validate_bag(args.bagname, fast=False, callback=None)
def generate_ro_manifest(bag_path,
overwrite=False,
config_file=DEFAULT_CONFIG_FILE):
bag = bdbagit.BDBag(bag_path)
bag_ro_metadata_path = os.path.abspath(
os.path.join(bag_path, "metadata", "manifest.json"))
exists = os.path.isfile(bag_ro_metadata_path)
if exists and not overwrite:
logger.info("Auto-generating RO manifest: update existing file.")
ro_metadata = bdbro.read_bag_ro_metadata(bag_path)
else:
logger.info(
"Auto-generating RO manifest: %s." %
"creating new file" if not exists else "overwrite existing file")
ro_metadata = bdbro.init_ro_manifest(
author_name=bag.info.get("Contact-Name"),
author_orcid=bag.info.get("Contact-Orcid"),
creator_name=bdbro.BAG_CREATOR_NAME,
creator_uri=bdbro.BAG_CREATOR_URI)
config = read_config(config_file)
resolvers = config.get(
ID_RESOLVER_TAG,
DEFAULT_ID_RESOLVERS) if config else DEFAULT_ID_RESOLVERS
fetched = bag.fetch_entries()
local = bag.payload_files()
for url, length, filename in fetched:
if url.startswith("minid:") or url.startswith("ark:"):
url = "".join(["http://", resolvers[0], "/", url])
bdbro.add_file_metadata(ro_metadata,
source_url=url,
bundled_as=bdbro.make_bundled_as(
folder=os.path.dirname(filename),
filename=os.path.basename(filename)),
update_existing=True)
for path in local:
bdbro.add_file_metadata(ro_metadata,
local_path=path.replace("\\", "/"),
bundled_as=bdbro.make_bundled_as(),
update_existing=True)
bdbro.write_bag_ro_metadata(ro_metadata, bag_path)
profile = bag.info.get(BAG_PROFILE_TAG)
if profile == BDBAG_PROFILE_ID:
bag.info.update({BAG_PROFILE_TAG: BDBAG_RO_PROFILE_ID})
bag.save()
def download(self, **kwargs):
if not self.config:
raise DerivaDownloadConfigurationError(
"No configuration specified!")
if self.config.get("catalog") is None:
raise DerivaDownloadConfigurationError(
"Catalog configuration error!")
ro_manifest = None
ro_author_name = None
ro_author_orcid = None
remote_file_manifest = os.path.abspath(''.join([
os.path.join(self.output_dir, 'remote-file-manifest_'),
str(uuid.uuid4()), ".json"
]))
catalog_config = self.config['catalog']
self.envars.update(self.config.get('env', dict()))
self.envars.update({"hostname": self.hostname})
# 1. If we don't have a client identity, we need to authenticate
identity = kwargs.get("identity")
if not identity:
try:
if not self.credentials:
self.set_credentials(get_credential(self.hostname))
logging.info("Validating credentials for host: %s" %
self.hostname)
attributes = self.catalog.get_authn_session().json()
identity = attributes["client"]
except HTTPError as he:
if he.response.status_code == 404:
logging.info(
"No existing login session found for host: %s" %
self.hostname)
except Exception as e:
raise DerivaDownloadAuthenticationError(
"Unable to validate credentials: %s" % format_exception(e))
wallet = kwargs.get("wallet", {})
# 2. Check for bagging config and initialize bag related variables
bag_path = None
bag_archiver = None
bag_algorithms = None
bag_config = self.config.get('bag')
create_bag = True if bag_config else False
if create_bag:
bag_name = bag_config.get(
'bag_name', ''.join([
"deriva_bag", '_',
time.strftime("%Y-%m-%d_%H.%M.%S")
])).format(**self.envars)
bag_path = os.path.abspath(os.path.join(self.output_dir, bag_name))
bag_archiver = bag_config.get('bag_archiver')
bag_algorithms = bag_config.get('bag_algorithms', ['sha256'])
bag_metadata = bag_config.get(
'bag_metadata',
{"Internal-Sender-Identifier": "deriva@%s" % self.server_url})
bag_ro = create_bag and stob(bag_config.get('bag_ro', "True"))
if create_bag:
bdb.ensure_bag_path_exists(bag_path)
bag = bdb.make_bag(bag_path,
algs=bag_algorithms,
metadata=bag_metadata)
if bag_ro:
ro_author_name = bag.info.get(
"Contact-Name", None if not identity else identity.get(
'full_name',
identity.get('display_name',
identity.get('id', None))))
ro_author_orcid = bag.info.get("Contact-Orcid")
ro_manifest = ro.init_ro_manifest(
author_name=ro_author_name,
author_orcid=ro_author_orcid)
bag_metadata.update({BAG_PROFILE_TAG: BDBAG_RO_PROFILE_ID})
# 3. Process the set of queries by locating, instantiating, and invoking the specified processor(s)
outputs = dict()
base_path = bag_path if bag_path else self.output_dir
for processor in catalog_config['query_processors']:
processor_name = processor["processor"]
processor_type = processor.get('processor_type')
processor_params = processor.get('processor_params')
try:
query_processor = find_query_processor(processor_name,
processor_type)
processor = query_processor(
self.envars,
inputs=outputs,
bag=create_bag,
catalog=self.catalog,
store=self.store,
base_path=base_path,
processor_params=processor_params,
remote_file_manifest=remote_file_manifest,
ro_manifest=ro_manifest,
ro_author_name=ro_author_name,
ro_author_orcid=ro_author_orcid,
identity=identity,
wallet=wallet)
outputs = processor.process()
except Exception as e:
logging.error(format_exception(e))
if create_bag:
bdb.cleanup_bag(bag_path)
raise
# 4. Execute anything in the transform processing pipeline, if configured
transform_processors = self.config.get('transform_processors', [])
if transform_processors:
for processor in transform_processors:
processor_name = processor["processor"]
processor_type = processor.get('processor_type')
processor_params = processor.get('processor_params')
try:
transform_processor = find_transform_processor(
processor_name, processor_type)
processor = transform_processor(
self.envars,
inputs=outputs,
processor_params=processor_params,
base_path=base_path,
bag=create_bag,
ro_manifest=ro_manifest,
ro_author_name=ro_author_name,
ro_author_orcid=ro_author_orcid,
identity=identity,
wallet=wallet)
outputs = processor.process()
except Exception as e:
logging.error(format_exception(e))
raise
# 5. Create the bag, and archive (serialize) if necessary
if create_bag:
try:
if ro_manifest:
ro.write_bag_ro_metadata(ro_manifest, bag_path)
if not os.path.isfile(remote_file_manifest):
remote_file_manifest = None
bdb.make_bag(
bag_path,
algs=bag_algorithms,
remote_file_manifest=remote_file_manifest if
(remote_file_manifest
and os.path.getsize(remote_file_manifest) > 0) else None,
update=True)
except Exception as e:
logging.fatal("Exception while updating bag manifests: %s" %
format_exception(e))
bdb.cleanup_bag(bag_path)
raise
finally:
if remote_file_manifest and os.path.isfile(
remote_file_manifest):
os.remove(remote_file_manifest)
logging.info('Created bag: %s' % bag_path)
if bag_archiver is not None:
try:
archive = bdb.archive_bag(bag_path, bag_archiver.lower())
bdb.cleanup_bag(bag_path)
outputs = {
os.path.basename(archive): {
LOCAL_PATH_KEY: archive
}
}
except Exception as e:
logging.error(
"Exception while creating data bag archive: %s" %
format_exception(e))
raise
else:
outputs = {
os.path.basename(bag_path): {
LOCAL_PATH_KEY: bag_path
}
}
# 6. Execute anything in the post processing pipeline, if configured
post_processors = self.config.get('post_processors', [])
if post_processors:
for processor in post_processors:
processor_name = processor["processor"]
processor_type = processor.get('processor_type')
processor_params = processor.get('processor_params')
try:
post_processor = find_post_processor(
processor_name, processor_type)
processor = post_processor(
self.envars,
inputs=outputs,
processor_params=processor_params,
identity=identity,
wallet=wallet)
outputs = processor.process()
except Exception as e:
logging.error(format_exception(e))
raise
return outputs
def download(self, identity=None):
if not self.config:
raise RuntimeError("No configuration specified!")
if self.config.get("catalog") is None:
raise RuntimeError("Catalog configuration error!")
if not identity:
logging.info("Validating credentials")
try:
if not self.credentials:
self.setCredentials(get_credential(self.hostname))
attributes = self.catalog.get_authn_session().json()
identity = attributes["client"]
except Exception as e:
raise RuntimeError("Unable to validate credentials: %s" % format_exception(e))
ro_manifest = None
ro_author_name = None
ro_author_orcid = None
remote_file_manifest = os.path.abspath(
''.join([os.path.join(self.output_dir, 'remote-file-manifest_'), str(uuid.uuid4()), ".json"]))
catalog_config = self.config['catalog']
self.envars.update(self.config.get('env', dict()))
bag_path = None
bag_archiver = None
bag_algorithms = None
bag_config = self.config.get('bag')
create_bag = True if bag_config else False
if create_bag:
bag_name = bag_config.get('bag_name', ''.join(["deriva_bag", '_', time.strftime("%Y-%m-%d_%H.%M.%S")]))
bag_path = os.path.abspath(os.path.join(self.output_dir, bag_name))
bag_archiver = bag_config.get('bag_archiver')
bag_algorithms = bag_config.get('bag_algorithms', ['sha256'])
bag_metadata = bag_config.get('bag_metadata', {"Internal-Sender-Identifier":
"deriva@%s" % self.server_url})
bag_ro = create_bag and stob(bag_config.get('bag_ro', "True"))
if create_bag:
bdb.ensure_bag_path_exists(bag_path)
bag = bdb.make_bag(bag_path, algs=bag_algorithms, metadata=bag_metadata)
if bag_ro:
ro_author_name = bag.info.get("Contact-Name",
identity.get('full_name',
identity.get('display_name',
identity.get('id', None))))
ro_author_orcid = bag.info.get("Contact-Orcid")
ro_manifest = ro.init_ro_manifest(author_name=ro_author_name, author_orcid=ro_author_orcid)
bag_metadata.update({BAG_PROFILE_TAG: BDBAG_RO_PROFILE_ID})
file_list = list()
base_path = bag_path if bag_path else self.output_dir
for query in catalog_config['queries']:
query_path = query['query_path']
output_format = query['output_format']
output_processor = query.get("output_format_processor")
format_args = query.get('output_format_params', None)
output_path = query.get('output_path', '')
try:
download_processor = findProcessor(output_format, output_processor)
processor = download_processor(self.envars,
bag=create_bag,
catalog=self.catalog,
store=self.store,
query=query_path,
base_path=base_path,
sub_path=output_path,
format_args=format_args,
remote_file_manifest=remote_file_manifest,
ro_manifest=ro_manifest,
ro_author_name=ro_author_name,
ro_author_orcid=ro_author_orcid)
file_list.extend(processor.process())
except Exception as e:
logging.error(format_exception(e))
if create_bag:
bdb.cleanup_bag(bag_path)
raise
if create_bag:
try:
if ro_manifest:
ro.write_bag_ro_metadata(ro_manifest, bag_path)
if not os.path.isfile(remote_file_manifest):
remote_file_manifest = None
bdb.make_bag(bag_path, algs=bag_algorithms, remote_file_manifest=remote_file_manifest, update=True)
except Exception as e:
logging.fatal("Exception while updating bag manifests: %s", format_exception(e))
bdb.cleanup_bag(bag_path)
raise
finally:
if remote_file_manifest and os.path.isfile(remote_file_manifest):
os.remove(remote_file_manifest)
logging.info('Created bag: %s' % bag_path)
if bag_archiver is not None:
try:
archive = bdb.archive_bag(bag_path, bag_archiver.lower())
bdb.cleanup_bag(bag_path)
return [archive]
except Exception as e:
logging.error("Exception while creating data bag archive:", format_exception(e))
raise
else:
return [bag_path]
return file_list