Relax the digest specification to handle more formats

digest/digest_tools.py

@@ -3,8 +3,9 @@ import os.path
 import hashlib
 
 
-DIGEST_PATTERN = r'(tarsum\.(v[\w]+)\+)?([\w]+):([0-9a-f]+)'
+DIGEST_PATTERN = r'([A-Za-z0-9_+.-]+):([A-Fa-f0-9]+)'
-
+REPLACE_WITH_PATH = re.compile(r'[+.]')
+REPLACE_DOUBLE_SLASHES = re.compile(r'/+')
 
 class InvalidDigestException(RuntimeError):
   pass
@@ -13,15 +14,11 @@ class InvalidDigestException(RuntimeError):
 class Digest(object):
   DIGEST_REGEX = re.compile(DIGEST_PATTERN)
 
-  def __init__(self, hash_alg, hash_bytes, is_tarsum=False, tarsum_version=None):
+  def __init__(self, hash_alg, hash_bytes):
     self._hash_alg = hash_alg
     self._hash_bytes = hash_bytes
-    self._is_tarsum = is_tarsum
-    self._tarsum_version = tarsum_version
 
   def __str__(self):
-    if self._is_tarsum:
-      return 'tarsum.{0}+{1}:{2}'.format(self._tarsum_version, self._hash_alg, self._hash_bytes)
     return '{0}:{1}'.format(self._hash_alg, self._hash_bytes)
 
   def __eq__(self, rhs):
@@ -34,16 +31,7 @@ class Digest(object):
     if match is None or match.end() != len(digest):
       raise InvalidDigestException('Not a valid digest: %s', digest)
 
-    is_tarsum = match.group(1) is not None
+    return Digest(match.group(1), match.group(2))
-    return Digest(match.group(3), match.group(4), is_tarsum, match.group(2))
-
-  @property
-  def is_tarsum(self):
-    return self._is_tarsum
-
-  @property
-  def tarsum_version(self):
-    return self._tarsum_version
 
   @property
   def hash_alg(self):
@@ -59,14 +47,12 @@ def content_path(digest):
   parsed = Digest.parse_digest(digest)
   components = []
 
-  if parsed.is_tarsum:
-    components.extend(['tarsum', parsed.tarsum_version])
-
   # Generate a prefix which is always two characters, and which will be filled with leading zeros
   # if the input does not contain at least two characters. e.g. ABC -> AB, A -> 0A
   prefix = parsed.hash_bytes[0:2].zfill(2)
-  components.extend([parsed.hash_alg, prefix, parsed.hash_bytes])
+  pathish = REPLACE_WITH_PATH.sub('/', parsed.hash_alg)
-
+  normalized = REPLACE_DOUBLE_SLASHES.sub('/', pathish).lstrip('/')
+  components.extend([normalized, prefix, parsed.hash_bytes])
   return os.path.join(*components)
 
 

test/test_digest_tools.py

@@ -5,12 +5,13 @@ from digest.digest_tools import Digest, content_path, InvalidDigestException
 class TestParseDigest(unittest.TestCase):
   def test_parse_good(self):
     examples = [
-      ('tarsum.v123123+sha1:123deadbeef', ('sha1', '123deadbeef', True, 'v123123')),
+      ('tarsum.v123123+sha1:123deadbeef', ('tarsum.v123123+sha1', '123deadbeef')),
-      ('tarsum.v1+sha256:123123', ('sha256', '123123', True, 'v1')),
+      ('tarsum.v1+sha256:123123', ('tarsum.v1+sha256', '123123')),
-      ('tarsum.v0+md5:abc', ('md5', 'abc', True, 'v0')),
+      ('tarsum.v0+md5:abc', ('tarsum.v0+md5', 'abc')),
-      ('sha1:123deadbeef', ('sha1', '123deadbeef', False, None)),
+      ('tarsum+sha1:abc', ('tarsum+sha1', 'abc')),
-      ('sha256:123123', ('sha256', '123123', False, None)),
+      ('sha1:123deadbeef', ('sha1', '123deadbeef')),
-      ('md5:abc', ('md5', 'abc', False, None)),
+      ('sha256:123123', ('sha256', '123123')),
+      ('md5:abc', ('md5', 'abc')),
     ]
 
     for digest, output_args in examples:
@@ -21,9 +22,7 @@ class TestParseDigest(unittest.TestCase):
 
   def test_parse_fail(self):
     examples = [
-      'tarsum.v++sha1:123deadbeef',
+      'tarsum.v+md5:abc:',
-      '.v1+sha256:123123',
-      'tarsum.v+md5:abc',
       'sha1:123deadbeefzxczxv',
       'sha256123123',
       'tarsum.v1+',
@@ -45,6 +44,8 @@ class TestDigestPath(unittest.TestCase):
       ('sha256:123123', 'sha256/12/123123'),
       ('md5:abc', 'md5/ab/abc'),
       ('md5:1', 'md5/01/1'),
+      ('md5.....+++:1', 'md5/01/1'),
+      ('.md5.:1', 'md5/01/1'),
     ]
 
     for digest, path in examples: