fix(blob): base64 includes additional characters (googleapis#258)

william-silversmith · remmykilonzo · commit 818dc5af6e2c · 2020-09-15T22:37:44.000Z
Hashes were not being parsed correctly. I forgot that base64 includes the "+" and "/" characters. https://en.wikipedia.org/wiki/Base64 <img width="656" alt="image" src="https://user-images.githubusercontent.com/2517065/91491542-3329ee80-e882-11ea-9665-4eaa564b406b.png">
diff --git a/google/cloud/storage/blob.py b/google/cloud/storage/blob.py
@@ -813,7 +813,7 @@ def _extract_headers_from_download(self, response):
 
         digests = {}
         for encoded_digest in x_goog_hash.split(","):
-            match = re.match(r"(crc32c|md5)=([\w\d/]+={0,3})", encoded_digest)
+            match = re.match(r"(crc32c|md5)=([\w\d/\+/]+={0,3})", encoded_digest)
             if match:
                 method, digest = match.groups()
                 digests[method] = digest
diff --git a/tests/unit/test_blob.py b/tests/unit/test_blob.py
@@ -1522,6 +1522,24 @@ def test_download_as_string_w_response_headers(self):
         self.assertEqual(blob.md5_hash, "CS9tHYTtyFntzj7B9nkkJQ==")
         self.assertEqual(blob.crc32c, "4gcgLQ==")
 
+        response = self._mock_requests_response(
+            http_client.OK,
+            headers={
+                "Content-Type": "application/octet-stream",
+                "Content-Language": "en-US",
+                "Cache-Control": "max-age=1337;public",
+                "Content-Encoding": "gzip",
+                "X-Goog-Storage-Class": "STANDARD",
+                "X-Goog-Hash": "crc32c=4/c+LQ==,md5=CS9tHYTt/+ntzj7B9nkkJQ==",
+            },
+            content=b"",
+        )
+        blob._extract_headers_from_download(response)
+        self.assertEqual(blob.content_type, "application/octet-stream")
+        self.assertEqual(blob.content_language, "en-US")
+        self.assertEqual(blob.md5_hash, "CS9tHYTt/+ntzj7B9nkkJQ==")
+        self.assertEqual(blob.crc32c, "4/c+LQ==")
+
     def test_download_as_string_w_hash_response_header_none(self):
         blob_name = "blob-name"
         md5_hash = "CS9tHYTtyFntzj7B9nkkJQ=="
