fix(score): raise ValueError when create_score receives None name or value

langfuse/_client/client.py

@@ -39,6 +39,7 @@
     _agnosticcontextmanager,
 )
 from packaging.version import Version
+from pydantic import ValidationError
 from typing_extensions import deprecated
 
 from langfuse._client.attributes import (
@@ -1952,6 +1953,8 @@ def create_score(
                     force_sample=force_sample,
                 )
 
+        except ValidationError as e:
+            raise ValueError(f"Invalid score parameters: {e}") from e
         except Exception as e:
             langfuse_logger.exception(
                 f"Error creating score: Failed to process score event for trace_id={trace_id}, name={name}. Error: {e}"

tests/unit/test_create_score_validation.py

@@ -0,0 +1,36 @@
+"""Unit tests for create_score input validation.
+
+Ensures that programmer errors (None name or value) raise ValueError
+at the call site rather than being silently swallowed.
+"""
+
+import pytest
+
+from langfuse import Langfuse
+from langfuse._client.resource_manager import LangfuseResourceManager
+
+
+@pytest.fixture(autouse=True)
+def _clear_singleton():
+    yield
+    with LangfuseResourceManager._lock:
+        LangfuseResourceManager._instances.clear()
+
+
+@pytest.fixture()
+def lf():
+    return Langfuse(
+        public_key="pk-lf-test",
+        secret_key="sk-lf-test",
+        host="http://localhost:19999",
+    )
+
+
+def test_create_score_raises_on_none_value(lf):
+    with pytest.raises(ValueError, match="Invalid score parameters"):
+        lf.create_score(name="accuracy", value=None, trace_id="fake-trace-id")
+
+
+def test_create_score_raises_on_none_name(lf):
+    with pytest.raises(ValueError, match="Invalid score parameters"):
+        lf.create_score(name=None, value=0.9, trace_id="fake-trace-id")