added token-level deduplication

Author: daniel-sanche

packages/google-api-core/google/api_core/gapic_v1/method.py

@@ -58,6 +58,25 @@ def _apply_decorators(func, decorators):
     return func
 
 
+def _deduplicate_metadata_tokens(*headers: str) -> str:
+    """
+    Given one or more metadata payload strings, create a combined
+    string with deduplicated tokens, while preserving token order.
+
+    Inputs are expected contain a set of metadata tokens separated by spaces
+    Example: `gl-python/3.14.0 grpc/1.76.0 gax/2.29.0 gapic/3.8.0 pb/6.33.4`
+
+    Args:
+      *headers: one or more metadata payload strings
+
+    Returns:
+        a single combined payload string
+    """
+    # Split all non-empty headers into individual tokens
+    token_list = " ".join(filter(None, headers)).split()
+    # Deduplicate while preserving order
+    return " ".join(dict.fromkeys(token_list))
+
 def _extract_metrics_header(metadata) -> Tuple[str, List[Tuple[str, str]]]:
     """Extract x-google-api-client header from metadata list.
 
@@ -74,8 +93,9 @@ def _extract_metrics_header(metadata) -> Tuple[str, List[Tuple[str, str]]]:
 
     key_to_find = client_info.METRICS_METADATA_KEY
 
-    metric_str = " ".join([v for k, v in metadata if k == key_to_find])
-
+    metric_str = _deduplicate_metadata_tokens(
+        " ".join([v for k, v in metadata if k == key_to_find])
+    )
     if not metric_str:
         return "", list(metadata)
 
@@ -147,14 +167,15 @@ def __call__(
         # Apply all applicable decorators.
         wrapped_func = _apply_decorators(self._target, [retry, timeout])
 
-        # Add the user agent metadata to the call.
         if user_metadata := kwargs.get("metadata"):
+            # Add the user agent metadata to the call.
             final_metadata = list(self._static_metadata)
             user_x_goog, remaining = _extract_metrics_header(user_metadata)
-            api_client_tokens = [t for t in [user_x_goog, self._x_goog_api_client] if t]
-            if api_client_tokens:
+
+            merged_header = _deduplicate_metadata_tokens(self._x_goog_api_client, user_x_goog)
+            if merged_header:
                 final_metadata.append(
-                    (client_info.METRICS_METADATA_KEY, " ".join(api_client_tokens))
+                    (client_info.METRICS_METADATA_KEY, merged_header)
                 )
             final_metadata.extend(remaining)
             kwargs["metadata"] = final_metadata

packages/google-api-core/tests/unit/gapic/test_method.py

@@ -119,6 +119,21 @@ def test_invoke_wrapped_method_with_metadata_as_none():
     assert len(metadata) == 1
 
 
+def test_extract_metrics_header_duplicate_tokens():
+    metadata = [
+        ("x-goog-api-client", "token1 token2"),
+        ("x-goog-api-client", "token2 token3 token1"),
+        ("other-header", "value"),
+        ("x-goog-api-client", "token4 token2"),
+    ]
+    
+    metric_str, arbitrary_metadata = google.api_core.gapic_v1.method._extract_metrics_header(metadata)
+    
+    # Should maintain order of first appearance and eliminate duplicates
+    assert metric_str == "token1 token2 token3 token4"
+    assert arbitrary_metadata == [("other-header", "value")]
+
+
 def test_invoke_wrapped_method_with_duplicate_x_goog_api_client_metadata():
     method = mock.Mock(spec=["__call__"])
 
@@ -134,11 +149,13 @@ def test_invoke_wrapped_method_with_duplicate_x_goog_api_client_metadata():
         method, client_info=client_info
     )
 
-    # Invoke the wrapped method with an explicit user-provided custom header
+    # Invoke the wrapped method with an explicit user-provided custom header that contains duplicates
+    # both within its own items and overlapping with the default client_info
     wrapped_method(
         mock.sentinel.request,
         metadata=[
             ("x-goog-api-client", "override-client/2.0"),
+            ("x-goog-api-client", "override-client/2.0 grpc/1.76.0 custom-user-agent/1.0"),
             ("other-header", "value"),
         ],
     )
@@ -156,7 +173,7 @@ def test_invoke_wrapped_method_with_duplicate_x_goog_api_client_metadata():
     # Verify both the user-provided override value and the library system telemetry are merged explicitly
     assert (
         metadata_dict["x-goog-api-client"]
-        == "override-client/2.0 custom-user-agent/1.0 gl-python/3.14.0 grpc/1.76.0 gax/2.29.0"
+        == "custom-user-agent/1.0 gl-python/3.14.0 grpc/1.76.0 gax/2.29.0 override-client/2.0"
     )
 
 
@@ -289,3 +306,24 @@ def test_wrap_method_with_call_not_supported():
     with pytest.raises(ValueError) as exc_info:
         google.api_core.gapic_v1.method.wrap_method(method, with_call=True)
     assert "with_call=True is only supported for unary calls" in str(exc_info.value)
+
+
+@pytest.mark.parametrize(
+    "headers,expected",
+    [
+        ((), ""),
+        (("",), ""),
+        ((None,), ""),
+        (("", None, ""), ""),
+        (("token1",), "token1"),
+        (("token1 token1",), "token1"),
+        (("token1", "token1"), "token1"),
+        (("token1 token2 token1",), "token1 token2"),
+        (("token1", "token2", "token1"), "token1 token2"),
+        (("token1 token2", "token2 token3"), "token1 token2 token3"),
+        (("token1", None, "token2", "", "token1"), "token1 token2"),
+    ],
+)
+def test__deduplicate_metadata_tokens(headers, expected):
+    dedup = google.api_core.gapic_v1.method._deduplicate_metadata_tokens
+    assert dedup(*headers) == expected