def spill(self, src, dest):
"""
Spill a workspace, i.e. unpack it and turn it into a workspace.
See https://ocr-d.github.com/ocrd_zip#unpacking-ocrd-zip-to-a-workspace
Arguments:
src (string): Path to OCRD-ZIP
dest (string): Path to directory to unpack data folder to
"""
# print(dest)
if exists(dest) and not isdir(dest):
raise Exception("Not a directory: %s" % dest)
# If dest is an existing directory, try to derive its name from src
if isdir(dest):
workspace_name = re.sub(r'(\.ocrd)?\.zip$', '', basename(src))
new_dest = join(dest, workspace_name)
if exists(new_dest):
raise Exception("Directory exists: %s" % new_dest)
dest = new_dest
log.info("Spilling %s to %s", src, dest)
bagdir = mkdtemp(prefix=TMP_BAGIT_PREFIX)
unzip_file_to_dir(src, bagdir)
datadir = join(bagdir, 'data')
for root, _, files in walk(datadir):
for f in files:
srcfile = join(root, f)
destdir = join(dest, relpath(root, datadir))
destfile = join(destdir, f)
if not exists(destdir):
makedirs(destdir)
log.debug("Copy %s -> %s", srcfile, destfile)
copyfile(srcfile, destfile)
# TODO copy allowed tag files if present
# TODO validate bagit
# Drop tempdir
rmtree(bagdir)
# Create workspace
workspace = Workspace(self.resolver, directory=dest)
# TODO validate workspace
return workspace
def validate(self,
skip_checksums=False,
skip_bag=False,
skip_unzip=False,
skip_delete=False,
processes=2):
"""
Validate an OCRD-ZIP file for profile, bag and workspace conformance
Arguments:
skip_bag (boolean): Whether to skip all checks of manifests and files
skip_checksums (boolean): Whether to omit checksum checks but still check basic BagIt conformance
skip_unzip (boolean): Whether the OCRD-ZIP is unzipped, i.e. a directory
skip_delete (boolean): Whether to skip deleting the unpacked OCRD-ZIP dir after valdiation
processes (integer): Number of processes used for checksum validation
"""
if skip_unzip:
bagdir = self.path_to_zip
skip_delete = True
else:
# try:
self.profile_validator.validate_serialization(self.path_to_zip)
# except IOError as err:
# raise err
# except ProfileValidationError as err:
# self.report.add_error(err.value)
bagdir = mkdtemp(prefix=TMP_BAGIT_PREFIX)
unzip_file_to_dir(self.path_to_zip, bagdir)
try:
bag = Bag(bagdir)
self._validate_profile(bag)
if not skip_bag:
self._validate_bag(bag,
fast=skip_checksums,
processes=processes)
finally:
if not skip_delete:
# remove tempdir
rmtree(bagdir)
return self.report