From e68e4b7d57738776868e96ac343ee63bf150d179 Mon Sep 17 00:00:00 2001
From: iomgaa <iomgaaycz@gmail.com>
Date: Tue, 14 Jul 2026 13:51:05 -0400
Subject: [PATCH] fix: restore sub_pattern when loading benchmark JSON

---
 app/question_gen/loader.py                    |  1 +
 .../test_generated_question_sub_pattern.py    | 30 +++++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/app/question_gen/loader.py b/app/question_gen/loader.py
index bad3950..878bfef 100644
--- a/app/question_gen/loader.py
+++ b/app/question_gen/loader.py
@@ -51,6 +51,7 @@ def load_benchmark(questions_dir: Path) -> list[GeneratedQuestion]:
                     family=qa.get("family"),
                     skill_target=qa.get("skill_target"),
                     difficulty_steps=qa.get("difficulty_steps"),
+                    sub_pattern=qa.get("sub_pattern"),
                 )
             )
     return results
diff --git a/tests/unit/test_generated_question_sub_pattern.py b/tests/unit/test_generated_question_sub_pattern.py
index 44f4659..c40feca 100644
--- a/tests/unit/test_generated_question_sub_pattern.py
+++ b/tests/unit/test_generated_question_sub_pattern.py
@@ -90,3 +90,33 @@ def test_append_to_json_writes_sub_pattern(tmp_path):
 
     data = json.loads((tmp_path / "v1.json").read_text(encoding="utf-8"))
     assert data[0]["sub_pattern"] == "temporal_reasoning_failure"
+
+
+def test_load_benchmark_restores_sub_pattern(tmp_path):
+    from app.question_gen.loader import load_benchmark
+    import json
+    entry = {
+        "question_id": "v1_Action Recognition_0001", "video_id": "v1",
+        "task_type": "Action Recognition", "question": "?",
+        "options": ["A. 蒸", "B. 炒", "C. 煮", "D. 炸"], "answer": "A",
+        "source_nodes": ["n1"], "difficulty": "hard",
+        "family": "ACTION_RECOGNITION", "skill_target": "M1_AR",
+        "sub_pattern": "temporal_reasoning_failure",
+    }
+    (tmp_path / "v1.json").write_text(json.dumps([entry]), encoding="utf-8")
+    loaded = load_benchmark(tmp_path)
+    assert loaded[0].sub_pattern == "temporal_reasoning_failure"
+
+
+def test_load_benchmark_sub_pattern_defaults_none_when_absent(tmp_path):
+    from app.question_gen.loader import load_benchmark
+    import json
+    entry = {
+        "question_id": "v1_Object Recognition_0001", "video_id": "v1",
+        "task_type": "Object Recognition", "question": "?",
+        "options": ["A. a", "B. b", "C. c", "D. d"], "answer": "A",
+        "source_nodes": ["n1"], "difficulty": "medium",
+    }
+    (tmp_path / "v1.json").write_text(json.dumps([entry]), encoding="utf-8")
+    loaded = load_benchmark(tmp_path)
+    assert loaded[0].sub_pattern is None