def _bag_mets_files(self, workspace, bagdir, ocrd_manifestation_depth,
ocrd_mets, processes):
mets = workspace.mets
# TODO allow filtering by fileGrp@USE and such
oldpwd = getcwd()
chdir(workspace.directory)
for f in mets.find_files():
log.info("Resolving %s (%s)", f.url, ocrd_manifestation_depth)
if is_local_filename(f.url):
f.url = abspath(f.url)
# XXX cannot happen because chdir above
# elif is_local_filename(join(workspace.directory, 'data', f.url)):
# f.url = abspath(join(workspace.directory, 'data', f.url))
elif ocrd_manifestation_depth != 'full':
self._log_or_raise(
"Not fetching non-local files, skipping %s" % f.url,
oldpwd)
continue
elif not f.url.startswith('http'):
self._log_or_raise("Not an http URL: %s" % f.url, oldpwd)
continue
log.info("Resolved %s", f.url)
file_grp_dir = join(bagdir, 'data', f.fileGrp)
if not isdir(file_grp_dir):
makedirs(file_grp_dir)
self.resolver.download_to_directory(file_grp_dir,
f.url,
basename=f.ID)
f.url = join(f.fileGrp, f.ID)
# save mets.xml
with open(join(bagdir, 'data', ocrd_mets), 'wb') as f:
f.write(workspace.mets.to_xml())
chdir(bagdir)
total_bytes, total_files = make_manifests('data',
processes,
algorithms=['sha512'])
chdir(oldpwd)
return total_bytes, total_files
def download_file(self, f):
"""
Download a :py:mod:`ocrd.model.ocrd_file.OcrdFile` to the workspace.
"""
# os.chdir(self.directory)
# log.info('f=%s' % f)
oldpwd = os.getcwd()
try:
os.chdir(self.directory)
if is_local_filename(f.url):
f.local_filename = abspath(f.url)
else:
if f.local_filename:
log.debug("Already downloaded: %s", f.local_filename)
else:
f.local_filename = self.download_url(f.url,
basename='%s/%s' %
(f.fileGrp, f.ID))
finally:
os.chdir(oldpwd)
# print(f)
return f
def test_abspath(self):
self.assertEqual(abspath('file:///'), '/')