fix: model usage computation

python/src/cairo_coder/core/types.py

@@ -210,7 +210,9 @@ def combine_usage(usage1: LMUsage, usage2: LMUsage) -> LMUsage:
             # Merge metrics
             for key, value in metrics.items():
                 if isinstance(value, int | float):
-                    result[model][key] = result[model].get(key, 0) + value
+                    # Use (x or 0) to handle None values - dict.get() returns None
+                    # if the key exists with value None, not the default
+                    result[model][key] = (result[model].get(key) or 0) + value
                 elif isinstance(value, dict):
                     if key not in result[model] or result[model][key] is None:
                         result[model][key] = value.copy()
@@ -219,7 +221,7 @@ def combine_usage(usage1: LMUsage, usage2: LMUsage) -> LMUsage:
                         for detail_key, detail_value in value.items():
                             if isinstance(detail_value, int | float):
                                 result[model][key][detail_key] = (
-                                    result[model][key].get(detail_key, 0) + detail_value
+                                    (result[model][key].get(detail_key) or 0) + detail_value
                                 )
     return result
 

python/tests/unit/test_rag_pipeline.py

@@ -14,7 +14,14 @@
     RagPipeline,
     RagPipelineFactory,
 )
-from cairo_coder.core.types import Document, DocumentSource, Message, Role, StreamEventType
+from cairo_coder.core.types import (
+    Document,
+    DocumentSource,
+    Message,
+    Role,
+    StreamEventType,
+    combine_usage,
+)
 from cairo_coder.dspy.retrieval_judge import RetrievalJudge
 
 
@@ -853,3 +860,88 @@ def test_create_pipeline_with_custom_components(self, mock_vector_store_config):
         assert pipeline.config.max_source_count == 20
         assert pipeline.config.similarity_threshold == 0.6
         assert pipeline.config.sources == [DocumentSource.CAIRO_BOOK]
+
+
+class TestCombineUsage:
+    """Tests for the combine_usage function."""
+
+    @pytest.mark.parametrize(
+        "usage1,usage2,expected",
+        [
+            # Test handling None values in usage dicts
+            pytest.param(
+                {
+                    "gpt-4": {
+                        "prompt_tokens": None,  # Key exists but value is None
+                        "completion_tokens": 100,
+                    }
+                },
+                {
+                    "gpt-4": {
+                        "prompt_tokens": 50,
+                        "completion_tokens": 50,
+                    }
+                },
+                {
+                    "gpt-4": {
+                        "prompt_tokens": 50,
+                        "completion_tokens": 150,
+                    }
+                },
+                id="none_values_in_usage_dict",
+            ),
+            # Test handling None values in nested dicts
+            pytest.param(
+                {
+                    "gpt-4": {
+                        "details": {
+                            "audio_tokens": None,  # Key exists but value is None
+                            "cached_tokens": 100,
+                        }
+                    }
+                },
+                {
+                    "gpt-4": {
+                        "details": {
+                            "audio_tokens": 25,
+                            "cached_tokens": 50,
+                        }
+                    }
+                },
+                {
+                    "gpt-4": {
+                        "details": {
+                            "audio_tokens": 25,
+                            "cached_tokens": 150,
+                        }
+                    }
+                },
+                id="none_values_in_nested_dict",
+            ),
+            # Test basic combining of usage dicts
+            pytest.param(
+                {"gpt-4": {"prompt_tokens": 100, "completion_tokens": 50}},
+                {"gpt-4": {"prompt_tokens": 200, "completion_tokens": 100}},
+                {"gpt-4": {"prompt_tokens": 300, "completion_tokens": 150}},
+                id="basic_combining",
+            ),
+            # Test combining with empty dicts
+            pytest.param(
+                {},
+                {},
+                {},
+                id="both_empty",
+            ),
+            # Test combining with one empty dict
+            pytest.param(
+                {"gpt-4": {"tokens": 100}},
+                {},
+                {"gpt-4": {"tokens": 100}},
+                id="second_empty",
+            ),
+        ],
+    )
+    def test_combine_usage(self, usage1, usage2, expected):
+        """Test combine_usage with various input scenarios."""
+        result = combine_usage(usage1, usage2)
+        assert result == expected