def CompareRepresentationXMLFiles(self):
Validation.objects.filter(task=self.get_processtask()).delete()
ip = InformationPackage.objects.get(pk=self.ip)
reps_path, reps_dir = find_destination("representations", ip.get_structure(), ip.object_path)
if reps_path is None:
return None
representations_dir = os.path.join(reps_path, reps_dir)
for p in find_pointers(os.path.join(ip.object_path, ip.content_mets_path)):
rep_mets_path = p.path
rep_mets_path = os.path.join(ip.object_path, rep_mets_path)
rep_path = os.path.relpath(rep_mets_path, representations_dir)
rep_path = PurePath(rep_path).parts[0]
rep_premis_path = get_premis_ref(etree.parse(rep_mets_path)).path
rep_premis_path = os.path.join(representations_dir, rep_path, rep_premis_path)
validator = XMLComparisonValidator(
context=rep_premis_path,
options={
'rootdir': os.path.join(representations_dir, rep_path),
'representation': rep_path,
'recursive': False,
},
task=self.get_processtask(),
ip=self.ip,
responsible=ip.responsible,
)
validator.validate(rep_mets_path)
msg = "All XML files in the representations have the same set of files"
self.create_success_event(msg)
def run(self):
Validation.objects.filter(task=self.get_processtask()).delete()
ip = InformationPackage.objects.get(pk=self.ip)
reps_path, reps_dir = find_destination("representations", ip.get_structure(), ip.object_path)
if reps_path is None:
return None
representations_dir = os.path.join(reps_path, reps_dir)
for p in find_pointers(ip.content_mets_path):
rep_mets_path = p.path
rep_mets_path = os.path.join(ip.object_path, rep_mets_path)
rep_path = os.path.relpath(rep_mets_path, representations_dir)
rep_path = PurePath(rep_path).parts[0]
rep_premis_path = get_premis_ref(etree.parse(rep_mets_path)).path
rep_premis_path = os.path.join(representations_dir, rep_path, rep_premis_path)
validator = XMLComparisonValidator(
context=rep_premis_path,
options={
'rootdir': os.path.join(representations_dir, rep_path),
'representation': rep_path,
},
task=self.get_processtask(),
ip=self.ip,
responsible=ip.responsible,
)
validator.validate(rep_mets_path)
def validate(self, path, expected=None):
xmlfile = self.context
objs = []
self._reset_dicts()
self._reset_counters()
logger.debug(u'Validating {path} against {xml}'.format(path=path,
xml=xmlfile))
checksum_in_context_file = self.checksums.get(path)
if checksum_in_context_file:
try:
self._pop_checksum_dict(self.deleted, checksum_in_context_file,
path)
self._pop_checksum_dict(self.present, checksum_in_context_file,
path)
except (KeyError, ValueError):
pass
skip_files = [os.path.relpath(xmlfile, self.rootdir)]
skip_files.extend([p.path for p in find_pointers(path)])
skip_files = list(map(normalize_path, skip_files))
for f in find_files(path, rootdir=self.rootdir, skip_files=skip_files):
if f in self.exclude:
continue
objs.append(self._validate(f))
delete_count = self._validate_deleted_files(objs)
self._validate_present_files(objs)
if checksum_in_context_file:
try:
self.deleted[checksum_in_context_file].append(path)
except KeyError:
self.deleted[checksum_in_context_file] = [path]
try:
self.present[checksum_in_context_file].append(path)
except KeyError:
self.present[checksum_in_context_file] = [path]
objs = [o for o in objs if o is not None]
Validation.objects.bulk_create(objs, batch_size=100)
if delete_count + self.added + self.changed + self.renamed > 0:
msg = (
'Comparison of {path} against {xml} failed: '
'{cfmd} confirmed, {a} added, {c} changed, {r} renamed, {d} deleted'
).format(path=path,
xml=self.context,
cfmd=self.confirmed,
a=self.added,
c=self.changed,
r=self.renamed,
d=delete_count)
logger.warning(msg)
raise ValidationError(msg)
logger.info(u"Successful comparison of {path} against {xml}".format(
path=path, xml=self.context))
def _get_files(self):
skip_files = [p.path for p in find_pointers(self.context)]
self.logical_files = find_files(
self.context,
rootdir=self.rootdir,
skip_files=skip_files,
recursive=self.recursive,
)