From f6b20daceadcbcf204a4d469d0cee05731948934 Mon Sep 17 00:00:00 2001
From: BSAIPF <96092130+BSAIPF@users.noreply.github.com>
Date: Wed, 21 Oct 2026 15:57:48 +0800
Subject: [PATCH] fix: honor partial pairwise scorers in Python ListContains

---
 py/autoevals/list.py        |  7 ++-----
 py/autoevals/test_values.py | 29 +++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/py/autoevals/list.py b/py/autoevals/list.py
index 6f5d2aa1..d993a725 100644
--- a/py/autoevals/list.py
+++ b/py/autoevals/list.py
@@ -26,10 +26,7 @@ async def _run_eval_async(self, output, expected=None, **kwargs):
             raise ValueError("ListContains requires an expected value")
 
         similarities_futures = [
-            [
-                self.pairwise_scorer._run_eval_async(output=output_item, expected=expected_item)
-                for expected_item in expected
-            ]
+            [self.pairwise_scorer.eval_async(output=output_item, expected=expected_item) for expected_item in expected]
             for output_item in output
         ]
 
@@ -45,7 +42,7 @@ def _run_eval_sync(self, output, expected=None, **kwargs):
             raise ValueError("ListContains requires an expected value")
 
         similarities = [
-            [self.pairwise_scorer._run_eval_sync(output_item, expected_item).score for expected_item in expected]
+            [self.pairwise_scorer.eval(output_item, expected_item).score for expected_item in expected]
             for output_item in output
         ]
 
diff --git a/py/autoevals/test_values.py b/py/autoevals/test_values.py
index 25bd6f24..026d2a29 100644
--- a/py/autoevals/test_values.py
+++ b/py/autoevals/test_values.py
@@ -1,7 +1,10 @@
+import pytest
 from pytest import approx
 
 from autoevals.list import ListContains
 from autoevals.number import NumericDiff
+from autoevals.partial import ScorerWithPartial
+from autoevals.score import Score
 from autoevals.string import LevenshteinScorer
 from autoevals.value import ExactMatch
 
@@ -80,6 +83,32 @@ def test_list_contains():
     )
 
 
+@pytest.mark.parametrize("run_async", [False, True])
+@pytest.mark.parametrize("as_class", [False, True])
+@pytest.mark.asyncio
+async def test_list_contains_partial_pairwise_scorer(run_async, as_class):
+    class CaseMatch(ScorerWithPartial):
+        def _run_eval_sync(self, output, expected=None, ignore_case=False, **kwargs):
+            if ignore_case:
+                output, expected = output.casefold(), expected.casefold()
+            return Score(name=self._name(), score=float(output == expected))
+
+    pairwise_scorer = CaseMatch.partial(ignore_case=True)
+    scorer = ListContains(pairwise_scorer=pairwise_scorer if as_class else pairwise_scorer())
+    output = ["PARIS", "Rome", "extra"]
+    expected = ["rome", "paris"]
+
+    if run_async:
+        assert (await pairwise_scorer().eval_async("PARIS", "paris")).score == 1
+        result = await scorer.eval_async(output=output, expected=expected)
+    else:
+        assert pairwise_scorer().eval("PARIS", "paris").score == 1
+        result = scorer.eval(output=output, expected=expected)
+
+    assert result.score == approx(2 / 3)
+    assert result.metadata["pairs"] == [("PARIS", "paris", 1), ("Rome", "rome", 1)]
+
+
 def test_exact_match():
     cases = [
         ["hello", "hello", 1],