def checkhash(dir,csvf):
"""
Check hash of file. Only do stuff with it if it has changed
"""
sha512hasher = FileHash('sha512')
process = {}
for f in csvf:
hashfile = f+".sha512"
hashvalue = sha512hasher.hash_file(dir + "/" + f)
if os.path.isfile(dir+"/"+hashfile):
# file with hash value is present, compare hashes
print("Hashfile ("+hashfile+") present")
checksums = dict(sha512hasher.verify_checksums(dir+"/"+hashfile))
for x, y in checksums.items():
if x == dir+"/"+f:
if y:
#print("Hashes match, I have already seen this file")
flag = False
else:
print("Hashes do not match ... we need to process this file; updating hashfile as well")
writehashfile(dir,f,hashfile,hashvalue)
flag = True
else:
print("Hashfile not present, creating it ...")
writehashfile(dir,f,hashfile,hashvalue)
flag = True
process[f] = flag
return process
def test_verify_checksums(self):
"""Test the verify_checksums() method."""
for algo in SUPPORTED_ALGORITHMS:
hasher = FileHash(algo)
results = [
result.hashes_match
for result in hasher.verify_checksums("hashes." + algo)
]
self.assertTrue(all(results))
def test_checksum(self):
tested = tempfile.mkstemp()[1]
fn = 'tests/data/reference_nc.nc'
nc2mmd = Nc_to_mmd(fn, check_only=True)
nc2mmd.to_mmd()
checksum = nc2mmd.metadata['storage_information']['checksum']
with open(tested, 'w') as tt:
tt.write('%s *%s'%(checksum, fn))
md5hasher = FileHash('md5')
self.assertTrue(md5hasher.verify_checksums(tested)[0].hashes_match)