def get_file_ctime_iso_date_str(file_path, fmt=DT_ISO_FORMAT, wd=None):
fp = remove_protocol(file_path)
path = fp if wd is None else os.path.join(wd, fp)
dt = pytz.timezone('Europe/Vienna').localize(
datetime.datetime.fromtimestamp(
os.path.getctime(path)).replace(microsecond=0))
return dt.strftime(fmt)
def package_sub_path_from_relative_path(root, containing_file_path,
relative_path):
containing_path, _ = os.path.split(containing_file_path)
return strip_prefixes(
os.path.abspath(
os.path.join(containing_path, remove_protocol(relative_path))),
root)
def validate_file(self, file):
'''
Validates every file found inside a Mets, so far: size, checksum, fixity. If a file exists, the counter for
self.total_files is diminished.
@param file: XML Element of a file that will be validated.
@return:
'''
err = []
log = []
# get information about the file
attr_path = file.getchildren()[0].attrib[q(XLINK_NS, 'href')]
attr_size = file.attrib['SIZE']
attr_checksum = file.attrib['CHECKSUM']
attr_checksumtype = file.attrib['CHECKSUMTYPE']
# mimetpye = file.attrib['MIMETYPE']
# check if file exists, if yes validate it
fitem = remove_protocol(attr_path)
file_path = os.path.join(self.rootpath, fitem).replace('\\', '/')
if not os.path.exists(file_path):
err.append(
"Unable to find file referenced in delivery METS file: %s" %
file_path)
else:
self.total_files -= 1
# check if file size is valid
# TODO: is this even needed?
file_size = os.path.getsize(file_path)
if not int(file_size) == int(attr_size):
err.append(
"Actual file size %s does not equal file size attribute value %s"
% (file_size, attr_size))
# workaround for earkweb.log in AIP metadata/ folder on IP root level
if file_path[-22:] == './metadata/earkweb.log':
err.pop()
log.append(
'Forced validation result \'True\' for file: %s' %
(file_path))
# validate checksum
checksum_validation = ChecksumValidation()
checksum_result = checksum_validation.validate_checksum(
file_path, attr_checksum, attr_checksumtype)
# workaround for earkweb.log in AIP metadata/ folder on IP root level
if file_path[-22:] == './metadata/earkweb.log':
checksum_result = True
if not checksum_result == True:
err.append('Checksum validation failed for: %s' % file_path)
for error in err:
print 'File validation error: ' + error
self.validation_errors.append(error)
def fsize(file_path, wd=None):
fp = remove_protocol(file_path)
path = fp if wd is None else os.path.join(wd, fp)
return int(os.path.getsize(path))
def checksum(file_path, wd=None, alg=ChecksumAlgorithm.SHA256):
fp = remove_protocol(file_path)
path = fp if wd is None else os.path.join(wd, fp)
return ChecksumFile(path).get(alg)