Handle non-string redirect URLs

Author: guyua9

guidance/_utils.py

@@ -50,7 +50,11 @@ def bytes_from(
             with urllib.request.urlopen(src, timeout=timeout) as response:
                 response = cast(http.client.HTTPResponse, response)
                 final_url = response.geturl()
-                if isinstance(final_url, str) and final_url != src:
+                if not isinstance(final_url, str):
+                    raise TypeError(
+                        f"Expected urlopen response URL to be a string, got {type(final_url).__name__}"
+                    )
+                if final_url != src:
                     validate_uri(
                         final_url,
                         allowed_schemes=allowed_schemes,

tests/unit/test_bytes_from.py

@@ -17,6 +17,7 @@ def test_allows_http_when_configured(self):
         with patch("urllib.request.urlopen") as mock_urlopen:
             mock_urlopen.return_value.__enter__ = lambda s: s
             mock_urlopen.return_value.__exit__ = lambda s, *a: None
+            mock_urlopen.return_value.geturl.return_value = "http://example.com/file.bin"
             mock_urlopen.return_value.read.return_value = b"data"
             with patch("guidance._uri_validation._validate_hostname_not_private"):
                 bytes_from("http://example.com/file.bin", allow_local=True, allowed_schemes=("http", "https"))
@@ -31,6 +32,7 @@ def test_allows_private_when_configured(self):
             with patch("urllib.request.urlopen") as mock_urlopen:
                 mock_urlopen.return_value.__enter__ = lambda s: s
                 mock_urlopen.return_value.__exit__ = lambda s, *a: None
+                mock_urlopen.return_value.geturl.return_value = "https://internal.example.com/file.bin"
                 mock_urlopen.return_value.read.return_value = b"data"
                 result = bytes_from("https://internal.example.com/file.bin", allow_local=True, allow_private=True)
                 assert result == b"data"
@@ -70,6 +72,18 @@ def fake_getaddrinfo(hostname, *args, **kwargs):
 
                 response.read.assert_not_called()
 
+    def test_rejects_non_string_redirect_target(self):
+        with patch("socket.getaddrinfo", return_value=[(None, None, None, None, ("93.184.216.34", 443))]):
+            with patch("urllib.request.urlopen") as mock_urlopen:
+                response = mock_urlopen.return_value.__enter__.return_value
+                response.geturl.return_value = b"https://example.com/file.bin"
+                response.read.return_value = b"data"
+
+                with pytest.raises(TypeError, match="response URL to be a string"):
+                    bytes_from("https://example.com/file.bin", allow_local=True)
+
+                response.read.assert_not_called()
+
     def test_local_path_still_works(self, tmp_path):
         test_file = tmp_path / "test.bin"
         test_file.write_bytes(b"local data")
@@ -109,6 +123,7 @@ def test_timeout_passed_to_urlopen(self):
             with patch("urllib.request.urlopen") as mock_urlopen:
                 mock_urlopen.return_value.__enter__ = lambda s: s
                 mock_urlopen.return_value.__exit__ = lambda s, *a: None
+                mock_urlopen.return_value.geturl.return_value = "https://example.com/f"
                 mock_urlopen.return_value.read.return_value = b"data"
                 bytes_from("https://example.com/f", allow_local=True, timeout=42.0)
                 mock_urlopen.assert_called_once_with("https://example.com/f", timeout=42.0)