[PR #2582/59c8b0f8 backport][2.27] Parse unquoted values in bearer auth challenges

CHANGES/+unquoted-auth-challenge.bugfix

@@ -0,0 +1 @@
+Fixed pull-through signature lookups failing with a 500 when the upstream registry (e.g. gcr.io) sends unquoted values in its bearer auth challenge.

pulp_container/app/downloaders.py

@@ -115,14 +115,7 @@ async def update_token(self, response_auth_header, used_token, repo_name):
             ):
                 return
             log.info("Updating bearer token")
-            bearer_info_string = response_auth_header[len("Bearer ") :]
-            bearer_info_list = re.split(",(?=[^=,]+=)", bearer_info_string)
-
-            # The remaining string consists of comma seperated key=value pairs
-            auth_query_dict = {}
-            for key, value in (item.split("=") for item in bearer_info_list):
-                # The value is a string within a string, ex: '"value"'
-                auth_query_dict[key] = json.loads(value)
+            auth_query_dict = self.parse_auth_challenge(response_auth_header)
             try:
                 token_base_url = auth_query_dict.pop("realm")
             except KeyError:
@@ -156,6 +149,24 @@ async def update_token(self, response_auth_header, used_token, repo_name):
             token = token_js.get("token") or token_js.get("access_token")
             self.factory.registry_auth["bearer"] = token
 
+    @staticmethod
+    def parse_auth_challenge(response_auth_header):
+        """
+        Parse the parameters of a Bearer auth challenge into a dictionary.
+
+        Values may be quoted strings or plain tokens (RFC 7235), e.g. gcr.io responds with
+        'Bearer realm="https://gcr.io/v2/token",service=gcr.io'.
+        """
+        bearer_info_string = response_auth_header[len("Bearer ") :]
+        auth_query_dict = {}
+        for match in re.finditer(r'([^=,\s]+)=(?:"((?:[^"\\]|\\.)*)"|([^,]*))', bearer_info_string):
+            key, quoted, plain = match.groups()
+            if quoted is not None:
+                auth_query_dict[key] = re.sub(r"\\(.)", r"\1", quoted)
+            else:
+                auth_query_dict[key] = plain.strip()
+        return auth_query_dict
+
     @staticmethod
     def auth_header(token, basic_auth):
         """

pulp_container/tests/unit/test_downloaders.py

@@ -0,0 +1,33 @@
+import pytest
+
+from pulp_container.app.downloaders import RegistryAuthHttpDownloader
+
+
+@pytest.mark.parametrize(
+    "header,expected",
+    [
+        (
+            'Bearer realm="https://auth.docker.io/token",service="registry.docker.io"',
+            {"realm": "https://auth.docker.io/token", "service": "registry.docker.io"},
+        ),
+        (
+            'Bearer realm="https://gcr.io/v2/token",service=gcr.io',
+            {"realm": "https://gcr.io/v2/token", "service": "gcr.io"},
+        ),
+        (
+            'Bearer realm="https://quay.io/v2/auth",service="quay.io",'
+            'scope="repository:foo/bar:pull,push"',
+            {
+                "realm": "https://quay.io/v2/auth",
+                "service": "quay.io",
+                "scope": "repository:foo/bar:pull,push",
+            },
+        ),
+        (
+            'Bearer realm="https://example.com/token", service=example.com, error="a \\"b\\""',
+            {"realm": "https://example.com/token", "service": "example.com", "error": 'a "b"'},
+        ),
+    ],
+)
+def test_parse_auth_challenge(header, expected):
+    assert RegistryAuthHttpDownloader.parse_auth_challenge(header) == expected