fix: skip merging when both usage entry values are None

dspy/utils/usage_tracker.py

@@ -45,7 +45,7 @@ def _merge_usage_entries(
             current_v = result.get(k)
             if isinstance(v, dict) or isinstance(current_v, dict):
                 result[k] = self._merge_usage_entries(current_v, v)
-            else:
+            elif current_v is not None or v is not None:
                 result[k] = (current_v or 0) + (v or 0)
         return result
 

tests/utils/test_usage_tracker.py

@@ -382,3 +382,25 @@ def task2():
 
     # Parent tracker should remain unchanged (workers have independent copies)
     assert len(parent_tracker.usage_data) == 0
+
+
+def test_merge_usage_entries_with_mixed_types():
+    """Test merging usage entries with mixed types (dict, int, None).
+
+    This test verifies the fix for the bug where _merge_usage_entries would fail
+    with TypeError when trying to call len() on a non-dict value during recursive merging.
+    """
+    tracker = UsageTracker()
+
+    # Test None values in details
+    tracker.add_usage("gpt-4o-mini", {"completion_tokens": 129, "prompt_tokens": 975, "completion_tokens_details": None, "prompt_tokens_details": None})
+    tracker.add_usage("gpt-4o-mini", {"completion_tokens": 144, "prompt_tokens": 1418, "completion_tokens_details": None, "prompt_tokens_details": {"cached_tokens": 0}})
+    tracker.add_usage("gpt-4o-mini", {"completion_tokens": 121, "prompt_tokens": 1167, "completion_tokens_details": {"reasoning_tokens": 0}, "prompt_tokens_details": None})
+
+    total_usage = tracker.get_total_tokens()
+    assert total_usage["gpt-4o-mini"]["completion_tokens"] == 394
+    assert total_usage["gpt-4o-mini"]["prompt_tokens"] == 3560
+    assert isinstance(total_usage["gpt-4o-mini"]["prompt_tokens_details"], dict)
+    assert isinstance(total_usage["gpt-4o-mini"]["completion_tokens_details"], dict)
+    assert total_usage["gpt-4o-mini"]["prompt_tokens_details"] == {"cached_tokens": 0}
+    assert total_usage["gpt-4o-mini"]["completion_tokens_details"] == {"reasoning_tokens": 0}