def is_valid(path, completeness_only=False, printfn=print):
"""Return whether a BagIt package is valid given its ``path``."""
try:
bag = Bag(path)
bag.validate(processes=multiprocessing.cpu_count(),
completeness_only=completeness_only)
except BagError as err:
printfn("Error validating BagIt package:", err, file=sys.stderr)
return False
return True
def main():
parser = _make_parser()
args = parser.parse_args()
bags = []
_configure_logging(args)
checks = "Performing the following validations: Checking 0xums, Checking bag completeness"
if not args.slow:
checks += ", Recalculating hashes"
LOGGER.info(checks)
if args.directory:
directory_path = os.path.abspath(args.directory)
for path in os.listdir(directory_path):
path = os.path.join(directory_path, path)
if os.path.isdir(path):
bags.append(path)
if args.bagpath:
bags.append(os.path.abspath(args.bagpath))
LOGGER.info("Checking {} folder(s).".format(len(bags)))
for bagpath in tqdm(bags):
LOGGER.info("Checking: {}".format(bagpath))
try:
bag = Bag(bagpath)
except:
LOGGER.error("{}: Not a bag".format(bagpath))
else:
if bag.validate(fast=args.slow):
LOGGER.info("{}: valid".format(bagpath))
else:
LOGGER.error("{}: invalid".format(bagpath))
def verify_aip(job):
"""Verify the AIP was bagged correctly by extracting it and running
verification on its contents. This is also where we verify the checksums
now that the verifyPREMISChecksums_v0.0 ("Verify checksums generated on
ingest") micro-service has been removed. It was removed because verifying
checksums by calculating them in that MS and then having bagit calculate
them here was redundant.
job.args[1] = UUID
UUID of the SIP, which will become the UUID of the AIP
job.args[2] = current location
Full absolute path to the AIP's current location on the local filesystem
"""
sip_uuid = job.args[1] # %sip_uuid%
aip_path = job.args[2] # SIPDirectory%%sip_name%-%sip_uuid%.7z
temp_dir = mcpclient_settings.TEMP_DIRECTORY
is_uncompressed_aip = os.path.isdir(aip_path)
if is_uncompressed_aip:
bag_path = aip_path
else:
try:
extract_dir = os.path.join(temp_dir, sip_uuid)
bag_path = extract_aip(job, aip_path, extract_dir)
except Exception as err:
job.print_error(repr(err))
job.pyprint('Error extracting AIP at "{}"'.format(aip_path),
file=sys.stderr)
return 1
return_code = 0
try:
# Only validate completeness since we're going to verify checksums
# later against what we have in the database via `verify_checksums`.
bag = Bag(bag_path)
bag.validate(completeness_only=True)
except BagError as err:
job.print_error("Error validating BagIt package: {}".format(err))
return_code = 1
if return_code == 0:
try:
verify_checksums(job, bag, sip_uuid)
except VerifyChecksumsError:
return_code = 1
else:
job.pyprint("Not verifying checksums because other tests have already"
" failed.")
# cleanup
if not is_uncompressed_aip:
try:
shutil.rmtree(extract_dir)
except OSError as err:
job.pyprint(
"Failed to remove temporary directory at {extract_dir} which"
" contains the AIP extracted for verification."
" Error:\n{err}".format(extract_dir=extract_dir, err=err),
file=sys.stderr,
)
return return_code
