fix: extract hashes correctly during download (#238)

Co-authored-by: Tres Seaver <tseaver@palladion.com>

Author: HemangChothani

google/cloud/storage/blob.py

@@ -804,7 +804,7 @@ def _extract_headers_from_download(self, response):
 
  digests = {}
  for encoded_digest in x_goog_hash.split(","):
- match = re.match(r"(crc32c|md5)=([\w\d]+)==", encoded_digest)
+ match = re.match(r"(crc32c|md5)=([\w\d/]+={0,3})", encoded_digest)
  if match:
  method, digest = match.groups()
  digests[method] = digest

tests/system/test_system.py

@@ -910,6 +910,18 @@ def test_upload_blob_owner(self):
  owner = same_blob.owner
  self.assertIn(user_email, owner["entity"])
 
+ def test_blob_crc32_md5_hash(self):
+ blob = self.bucket.blob("MyBuffer")
+ file_contents = b"Hello World"
+ blob.upload_from_string(file_contents)
+ self.case_blobs_to_delete.append(blob)
+
+ download_blob = self.bucket.blob("MyBuffer")
+
+ self.assertEqual(download_blob.download_as_string(), file_contents)
+ self.assertEqual(download_blob.crc32c, blob.crc32c)
+ self.assertEqual(download_blob.md5_hash, blob.md5_hash)
+
 
 class TestUnicode(unittest.TestCase):
  @vpcsc_config.skip_if_inside_vpcsc

tests/unit/test_blob.py

@@ -1476,8 +1476,8 @@ def test_download_as_string_w_response_headers(self):
  self.assertEqual(blob.content_encoding, "gzip")
  self.assertEqual(blob.cache_control, "max-age=1337;public")
  self.assertEqual(blob.storage_class, "STANDARD")
- self.assertEqual(blob.md5_hash, "CS9tHYTtyFntzj7B9nkkJQ")
- self.assertEqual(blob.crc32c, "4gcgLQ")
+ self.assertEqual(blob.md5_hash, "CS9tHYTtyFntzj7B9nkkJQ==")
+ self.assertEqual(blob.crc32c, "4gcgLQ==")
 
  def test_download_as_string_w_hash_response_header_none(self):
  blob_name = "blob-name"