def test_hexdigest(filepath, extra_hash, algorithm, expected_hash):
"""Test that hexdigest returns correct sha1 and MD5 hashes."""
if algorithm is None:
assert hexdigest(filepath, extra_hash=extra_hash) == expected_hash
else:
assert hexdigest(filepath, algorithm=algorithm,
extra_hash=extra_hash) == expected_hash
def test_hexdigest(filepath, extra_hash, algorithm, expected_hash):
"""
Test that hexdigest returns correct sha1 and MD5 hashes.
:filepath: Test file name
:extra_hash: Hash of depending file(s), or None if no such dependency
:algorithm: Hash algorithm
:expected_hash: Expected hash result
"""
if algorithm is None:
assert hexdigest(filepath, extra_hash=extra_hash) == expected_hash
else:
assert hexdigest(filepath, algorithm=algorithm,
extra_hash=extra_hash) == expected_hash
def calculate_md5(file_path):
"""
This function calculates md5sum for a file.
:file_path: path of file from which the md5sum is calculated.
:returns: a string with md5-hexdigest.
"""
return hexdigest(file_path, algorithm='md5')
def checksum(self, algorithm="MD5"):
"""
Return the checksum of the file with given algorithm.
:algorithm: MD5 or SHA variant
:returns: Calculated checksum
"""
return hexdigest(self.filename, algorithm)
def check_checksums(mets_path):
"""Check checksums for all digital objects in METS
:mets_path: Path to mets
:returns: Iterable containing all error messages
"""
checked_files = {}
if os.path.isdir(mets_path):
mets_path = os.path.join(mets_path, 'mets.xml')
sip_path = os.path.dirname(mets_path)
def _message(metadata_info, message):
"""Format error message"""
return ensure_text("%s: %s" % (
message, os.path.relpath(metadata_info["filename"], sip_path)))
mets_tree = u.readfile(mets_path)
for metadata_info in iter_metadata_info(mets_tree, mets_path):
checked_files[metadata_info["filename"]] = None
if metadata_info['algorithm'] is None:
yield _message(metadata_info, "Could not find checksum algorithm")
else:
try:
hex_digest = hexdigest(metadata_info['filename'],
metadata_info['algorithm'])
except IOError as exception:
if exception.errno == errno.ENOENT:
yield _message(metadata_info, "File does not exist")
continue
if hex_digest == metadata_info["digest"]:
print(_message(metadata_info, "Checksum OK"))
else:
yield _message(metadata_info, "Invalid Checksum")
for path in iter_files(sip_path):
if path.endswith("ignore_validation_errors"):
continue
if path not in checked_files:
yield _message({'filename': path}, "Nonlisted file")
def _generate_xslt_filename(self):
"""
Generate XSLT filename from schematron file.
:returns: XSLT filename
"""
try:
os.makedirs(self._cachepath)
except OSError:
if not os.path.isdir(self._cachepath):
raise
extra = ''
if self._verbose:
extra = 'verbose'
if self._extra_hash is not None:
extra = "%s%s" % (extra, self._extra_hash)
schema_digest = hexdigest(self._schematron_file, extra_hash=extra)
schema_basename = os.path.basename(self._schematron_file)
return os.path.join(self._cachepath, '%s.%s.validator.xsl' % (
schema_basename, schema_digest))