def bag(self,
workspace,
ocrd_identifier,
dest=None,
ocrd_mets='mets.xml',
ocrd_manifestation_depth='full',
ocrd_base_version_checksum=None,
processes=1,
skip_zip=False,
in_place=False,
tag_files=None):
"""
Bag a workspace
See https://ocr-d.github.com/ocrd_zip#packing-a-workspace-as-ocrd-zip
Arguments:
workspace (ocrd.Workspace): workspace to bag
ord_identifier (string): Ocrd-Identifier in bag-info.txt
dest (string): Path of the generated OCRD-ZIP.
ord_mets (string): Ocrd-Mets in bag-info.txt
ord_manifestation_depth (string): Ocrd-Manifestation-Depth in bag-info.txt
ord_base_version_checksum (string): Ocrd-Base-Version-Checksum in bag-info.txt
processes (integer): Number of parallel processes checksumming
skip_zip (boolean): Whether to leave directory unzipped
in_place (boolean): Whether to **replace** the workspace with its BagIt variant
tag_files (list<string>): Path names of additional tag files to be bagged at the root of the bag
"""
if ocrd_manifestation_depth not in ('full', 'partial'):
raise Exception("manifestation_depth must be 'full' or 'partial'")
if in_place and (dest is not None):
raise Exception("Setting 'dest' and 'in_place' is a contradiction")
if in_place and not skip_zip:
raise Exception(
"Setting 'skip_zip' and not 'in_place' is a contradiction")
if tag_files is None:
tag_files = []
# create bagdir
bagdir = mkdtemp(prefix=TMP_BAGIT_PREFIX)
if dest is None:
if in_place:
dest = workspace.directory
elif not skip_zip:
dest = '%s.ocrd.zip' % workspace.directory
else:
dest = '%s.ocrd' % workspace.directory
log.info("Bagging %s to %s (temp dir %s)", workspace.directory,
'(in-place)' if in_place else dest, bagdir)
# create data dir
makedirs(join(bagdir, 'data'))
# create bagit.txt
with open(join(bagdir, 'bagit.txt'), 'wb') as f:
f.write(BAGIT_TXT.encode('utf-8'))
# create manifests
total_bytes, total_files = self._bag_mets_files(
workspace, bagdir, ocrd_manifestation_depth, ocrd_mets, processes)
# create bag-info.txt
bag = Bag(bagdir)
self._set_bag_info(bag, total_bytes, total_files, ocrd_identifier,
ocrd_manifestation_depth,
ocrd_base_version_checksum)
for tag_file in tag_files:
copyfile(tag_file, join(bagdir, basename(tag_file)))
# save bag
bag.save()
# ZIP it
self._serialize_bag(workspace, bagdir, dest, in_place, skip_zip)
log.info('Created bag at %s', dest)
return dest
class ImageBag:
def __init__(self, path: str, auto_make: bool = False) -> None:
self.path = realpath(expanduser(expandvars(normpath(path))))
if auto_make:
try:
makedirs(self.path, exist_ok=False)
except OSError:
raise OSError('{} already exists, but auto_make = True'
''.format(self.path))
else:
self.bag = make_bag(self.path)
else:
try:
self.bag = Bag(self.path)
except BagError as e:
raise OSError(
'{} does not seem to be a valid Moondog Image bag: {}'
''.format(self.path, str(e)))
self.components = {}
return None
def accession(self, path: str):
self._import_original(path)
self._generate_master()
def _import_original(self, path: str):
d = self.components['original'] = {}
d['accession_path'] = realpath(expanduser(expandvars(normpath(path))))
d['filename'] = basename(d['accession_path'])
fn, ext = splitext(d['filename'])
target_path = join(self.path, 'data', d['filename'])
shutil.copy2(d['accession_path'], target_path)
with ExifTool() as et:
meta = et.get_metadata(target_path)
pprint(meta)
xmp = XMPFiles(file_path=target_path).get_xmp()
pprint(xmp)
self._update(manifests=True)
def _generate_master(self):
infn = self.components['original']['filename']
d = self.components['master'] = {}
d['filename'] = 'master.tif'
Image.open(join(self.path, 'data',
infn)).save(join(self.path, 'data', d['filename']))
self._update(manifests=True)
def _update(self, manifests=False):
"""Update the bag."""
for fn, fmeta in self.components.items():
for term, value in fmeta.items():
bag_term = '{}-{}'.format(
fn.title(),
term.replace('_', ' ').title().replace(' ', '-'))
try:
prior_value = self.bag.info[bag_term]
except KeyError:
self.bag.info[bag_term] = value
else:
if prior_value != value:
self.bag.info[bag_term] = value
self.bag.save(manifests=manifests)
