fix: preserve existing blob hashes when 'X-Goog-Hash header' is not present (#267)

Co-authored-by: Tres Seaver <[email protected]>

Author: MartinKilonzo

google/cloud/storage/blob.py

@@ -811,15 +811,16 @@ def _extract_headers_from_download(self, response):
         #  'X-Goog-Hash': 'crc32c=4gcgLQ==,md5=CS9tHYTtyFntzj7B9nkkJQ==',
         x_goog_hash = response.headers.get("X-Goog-Hash", "")
 
-        digests = {}
-        for encoded_digest in x_goog_hash.split(","):
-            match = re.match(r"(crc32c|md5)=([\w\d/\+/]+={0,3})", encoded_digest)
-            if match:
-                method, digest = match.groups()
-                digests[method] = digest
-
-        self.crc32c = digests.get("crc32c", None)
-        self.md5_hash = digests.get("md5", None)
+        if x_goog_hash:
+            digests = {}
+            for encoded_digest in x_goog_hash.split(","):
+                match = re.match(r"(crc32c|md5)=([\w\d/\+/]+={0,3})", encoded_digest)
+                if match:
+                    method, digest = match.groups()
+                    digests[method] = digest
+
+            self.crc32c = digests.get("crc32c", None)
+            self.md5_hash = digests.get("md5", None)
 
     def _do_download(
         self,

tests/unit/test_blob.py

@@ -1542,10 +1542,16 @@ def test_download_as_string_w_response_headers(self):
 
     def test_download_as_string_w_hash_response_header_none(self):
         blob_name = "blob-name"
+        md5_hash = "CS9tHYTtyFntzj7B9nkkJQ=="
+        crc32c = "4gcgLQ=="
         client = mock.Mock(spec=["_http"])
         bucket = _Bucket(client)
         media_link = "http://example.com/media/"
-        properties = {"mediaLink": media_link}
+        properties = {
+            "mediaLink": media_link,
+            "md5Hash": md5_hash,
+            "crc32c": crc32c,
+        }
         blob = self._make_one(blob_name, bucket=bucket, properties=properties)
 
         response = self._mock_requests_response(
@@ -1556,8 +1562,8 @@ def test_download_as_string_w_hash_response_header_none(self):
         )
         blob._extract_headers_from_download(response)
 
-        self.assertIsNone(blob.md5_hash)
-        self.assertIsNone(blob.crc32c)
+        self.assertEqual(blob.md5_hash, md5_hash)
+        self.assertEqual(blob.crc32c, crc32c)
 
     def test_download_as_bytes_w_generation_match(self):
         GENERATION_NUMBER = 6