jira:NYJ-1540 desc: add api_test.py

+import os
 import logging
 import logging
 from aabd.base.enhance_dict import value_or_default
 from aabd.base.enhance_dict import value_or_default
 
 
@@ -18,25 +19,31 @@ class FootballReplayMatch:
         save_frames_enable = value_or_default(settings.save_frames_enable, False)
         save_frames_enable = value_or_default(settings.save_frames_enable, False)
         
         
         self.videoEventRecognition = FootballReplayVideoEvent(llm_base_url, model_name, temperature, 'no_key', self.cache_dir, save_frames_enable)
         self.videoEventRecognition = FootballReplayVideoEvent(llm_base_url, model_name, temperature, 'no_key', self.cache_dir, save_frames_enable)
-        self.videoMatchLive = FootballReplayMatchLive(llm_base_url, model_name, temperature, 'no_key', self.cache_dir)
+        self.videoMatchLive = FootballReplayMatchLive(llm_base_url, model_name, temperature, 'no_key', self.cache_dir, save_frames_enable)
         
         
     def match_by_time(self, replay, events):
     def match_by_time(self, replay, events):
         start_utc = replay.get('start_utc')
         start_utc = replay.get('start_utc')
 
 
         events = [(start_utc - e.get('event_utc'), e) for e in events]
         events = [(start_utc - e.get('event_utc'), e) for e in events]
-        events.sort(key=lambda x: abs(x[0]))
-        time_diff, event = events[0]
+        events.sort(key=lambda x: x[0])
+        
+        target_index = next((i for i, e in enumerate(events) if e[0] > 0), -1) # 找到第一个 event_utc 早于 start_utc 的事件
+        
+        if target_index == -1:
+            return None
+        
+        time_diff, event = events[target_index]
         if time_diff < self.match_by_time_threshold:
         if time_diff < self.match_by_time_threshold:
             return event
             return event
         else:
         else:
             return None
             return None
 
 
-    def det_goal_replay(self, replay):
-        return self.videoEventRecognition.video_event(replay, cache_dir=self.cache_dir)
+    def det_goal_replay(self, replay, task_id):
+        return self.videoEventRecognition.video_event(replay, cache_dir=os.path.join(self.cache_dir, task_id, "replay"))
             
             
 
 
-    def match_by_llm(self, replay, events):
-        return self.videoMatchLive.match_batch(replay, events, max_parallel=2, cache_dir=self.cache_dir)
+    def match_by_llm(self, replay, events, task_id):
+        return self.videoMatchLive.match_batch(replay, events, max_parallel=2, cache_dir=os.path.join(self.cache_dir, task_id, "envents"))
 
 
     def replay_match_event(self, data):
     def replay_match_event(self, data):
         """
         """
@@ -72,22 +79,25 @@ class FootballReplayMatch:
         goal_live_events = [e for e in live_events if str(e.get('type', '')) == '1']
         goal_live_events = [e for e in live_events if str(e.get('type', '')) == '1']
 
 
         matched_event = self.match_by_time(replay_info, goal_live_events)
         matched_event = self.match_by_time(replay_info, goal_live_events)
-
         if matched_event is None:
         if matched_event is None:
-            replay_det_info = self.det_goal_replay(replay_info)
+            replay_det_info = self.det_goal_replay(replay_info, task_id)
             replay_event_name = replay_det_info.get('event_name', '')
             replay_event_name = replay_det_info.get('event_name', '')
             if replay_event_name == '进球':
             if replay_event_name == '进球':
-                matched_event = self.match_by_llm(replay_info, goal_live_events)
+                matched_event = self.match_by_llm(replay_info, goal_live_events, task_id)
                 if matched_event is None:
                 if matched_event is None:
                     logger.info(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 是进球但是未能找到匹配的事件")
                     logger.info(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 是进球但是未能找到匹配的事件")
+                    # print(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 是进球但是未能找到匹配的事件")
                 else:
                 else:
                     matched_event_id = matched_event.get('video_id')
                     matched_event_id = matched_event.get('video_id')
                     logger.info(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
                     logger.info(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
+                    # print(f"{task_id}_LLM认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
             else:
             else:
                 logger.info(f"{task_id}_LLM判断{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 不是进球，无需匹配")
                 logger.info(f"{task_id}_LLM判断{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 不是进球，无需匹配")
+                # print(f"{task_id}_LLM判断{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 不是进球，无需匹配")
         else:
         else:
             matched_event_id = matched_event.get('id')
             matched_event_id = matched_event.get('id')
             logger.info(f"{task_id}_通过时间找到匹配认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
             logger.info(f"{task_id}_通过时间找到匹配认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
+            # print(f"{task_id}_通过时间找到匹配认为视频{replay_info['url']}_{replay_info['start_utc']}_{replay_info['end_utc']} 对应的进球的事件视频是{matched_event_id}")
         return {
         return {
                 "id": task_id,
                 "id": task_id,
                 "match_id": match_id,
                 "match_id": match_id,

+"""
+api_test.py - 对 api.py 中 FootballReplayMatch.replay_match_event 的单元测试
+
+使用伪数据覆盖主要分支：
+1. 时间阈值内直接匹配成功
+2. 时间匹配失败 + LLM 判定不是进球
+3. 时间匹配失败 + LLM 判定是进球 + LLM 匹配成功
+4. 时间匹配失败 + LLM 判定是进球 + LLM 匹配失败
+5. 无进球事件 / events 为空
+6. 边界：缺少 replay 键时抛出异常
+"""
+
+import unittest
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+try:
+    from api import FootballReplayMatch
+except ImportError:
+    from .api import FootballReplayMatch
+
+
+def _fake_settings(match_by_time_threshold: int = 30):
+    """构造一个 settings 伪对象，兼容 api.py 中的 value_or_default 读取逻辑。"""
+    return SimpleNamespace(
+        match_by_time_threshold=match_by_time_threshold,
+        llm=SimpleNamespace(base_url="http://192.168.1.59:11434/v1", model_name="Qwen3.6-35B-A3B-UD-Q8_K_XL.gguf", temperature=0.7),
+        common=SimpleNamespace(cache_dir="/root/lzw/tmp_0518_replay_cache0521_01"),
+        save_frames_enable=True,
+    )
+
+
+def _base_data():
+    """返回一份基础伪数据，replay.start_utc = 1000（毫秒）。"""
+    
+    url = "http://video.mam.miguvideo.com/mnt6/fastclip3/wsc/2026/04/29/df760b8d15394cbc9ed0d0a4a9c4b786_1080PS/29133605/vodtmp/3b3e99cf4ca84c3782503d8817242de2.m3u8"
+    return {
+        "id": "task_001",
+        "match_id": "match_001",
+        "replay": {
+            "id": "replay_001",
+            "url": url,
+            "start_utc": 1777444502543,
+            "end_utc": 1777444531063
+        },
+        "events": [
+            {"id": "event_001", "type": "1", "url": url, "event_utc": 1777444502543-40000}, 
+            {"id": "event_001", "type": "1", "url": url, "event_utc": 1777444502543-20000},  
+            {"id": "event_001", "type": "1", "url": url, "event_utc": 1777444531063+20000}
+        ],
+    }
+
+
+class TestReplayMatchEvent(unittest.TestCase):
+
+    @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    def test_match_by_time_success(self, mock_video_cls, mock_live_cls):
+        """场景1：时间差在阈值内，直接通过 match_by_time 命中。"""
+        settings = _fake_settings(match_by_time_threshold=10)  # threshold = 30 * 1000 = 30000 ms
+        frm = FootballReplayMatch(settings)
+
+        data = _base_data()
+        result = frm.replay_match_event(data)
+        
+        print(result)
+        # exit()
+
+    #     self.assertEqual(result["id"], "task_001")
+    #     self.assertEqual(result["match_id"], "match_001")
+    #     self.assertEqual(result["replay_id"], "replay_001")
+    #     # event_001 与 replay.start_utc 差值 20 ms < 30000 ms，应命中 event_001 的 id
+    #     self.assertEqual(result["event_id"], "event_001")
+    #     # 未调用 LLM 相关方法
+    #     mock_video_cls.return_value.video_event.assert_not_called()
+    #     mock_live_cls.return_value.match_batch.assert_not_called()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_match_by_time_fail_then_not_goal(self, mock_video_cls, mock_live_cls):
+    #     """场景2：时间匹配失败，det_goal_replay 判定不是进球。"""
+    #     mock_video_instance = MagicMock()
+    #     mock_video_instance.video_event.return_value = {"event_name": "无进球", "description": "宣传片"}
+    #     mock_video_cls.return_value = mock_video_instance
+
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     result = frm.replay_match_event(data)
+
+    #     self.assertEqual(result["event_id"], None)
+    #     mock_video_instance.video_event.assert_called_once()
+    #     mock_live_cls.return_value.match_batch.assert_not_called()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_match_by_time_fail_then_goal_then_llm_match(self, mock_video_cls, mock_live_cls):
+    #     """场景3：时间匹配失败 -> 判定是进球 -> LLM 匹配成功。"""
+    #     mock_video_instance = MagicMock()
+    #     mock_video_instance.video_event.return_value = {"event_name": "进球", "description": "世界波"}
+    #     mock_video_cls.return_value = mock_video_instance
+
+    #     mock_live_instance = MagicMock()
+    #     mock_live_instance.match_batch.return_value = {"video_id": "event_003", "video_path": "http://example.com/event3.mp4", "asr_text": ""}
+    #     mock_live_cls.return_value = mock_live_instance
+
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     result = frm.replay_match_event(data)
+
+    #     self.assertEqual(result["event_id"], "event_003")
+    #     mock_video_instance.video_event.assert_called_once()
+    #     mock_live_instance.match_batch.assert_called_once()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_match_by_time_fail_then_goal_then_llm_no_match(self, mock_video_cls, mock_live_cls):
+    #     """场景4：时间匹配失败 -> 判定是进球 -> LLM 未匹配到任何事件。"""
+    #     mock_video_instance = MagicMock()
+    #     mock_video_instance.video_event.return_value = {"event_name": "进球", "description": "头球破门"}
+    #     mock_video_cls.return_value = mock_video_instance
+
+    #     mock_live_instance = MagicMock()
+    #     mock_live_instance.match_batch.return_value = None
+    #     mock_live_cls.return_value = mock_live_instance
+
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     result = frm.replay_match_event(data)
+
+    #     self.assertEqual(result["event_id"], None)
+    #     mock_video_instance.video_event.assert_called_once()
+    #     mock_live_instance.match_batch.assert_called_once()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_no_goal_events(self, mock_video_cls, mock_live_cls):
+    #     """场景5：events 中没有任何 type='1' 的进球事件。"""
+    #     mock_video_instance = MagicMock()
+    #     mock_video_instance.video_event.return_value = {"event_name": "无进球", "description": "无候选事件"}
+    #     mock_video_cls.return_value = mock_video_instance
+
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     data["events"] = [
+    #         {"id": "event_005", "type": "2", "url": "http://example.com/event5.mp4", "event_utc": 1020},
+    #     ]
+    #     result = frm.replay_match_event(data)
+
+    #     self.assertEqual(result["event_id"], None)
+    #     mock_video_instance.video_event.assert_called_once()
+    #     mock_live_cls.return_value.match_batch.assert_not_called()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_empty_events(self, mock_video_cls, mock_live_cls):
+    #     """场景6：events 为空列表。"""
+    #     mock_video_instance = MagicMock()
+    #     mock_video_instance.video_event.return_value = {"event_name": "无进球", "description": "空列表"}
+    #     mock_video_cls.return_value = mock_video_instance
+
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     data["events"] = []
+    #     result = frm.replay_match_event(data)
+
+    #     self.assertEqual(result["event_id"], None)
+    #     mock_video_instance.video_event.assert_called_once()
+    #     mock_live_cls.return_value.match_batch.assert_not_called()
+
+    # @patch("utils.football_replay_match_live.FootballReplayMatchLive")
+    # @patch("utils.football_replay_video_event_by_llm.FootballReplayVideoEvent")
+    # def test_missing_replay_key(self, mock_video_cls, mock_live_cls):
+    #     """场景7：data 缺少 replay 键，应抛出 KeyError。"""
+    #     settings = _fake_settings(match_by_time_threshold=30)
+    #     frm = FootballReplayMatch(settings)
+
+    #     data = _base_data()
+    #     del data["replay"]
+
+    #     with self.assertRaises(KeyError):
+    #         frm.replay_match_event(data)
+
+
+if __name__ == "__main__":
+    unittest.main()

@@ -82,7 +82,6 @@ class FootballReplayMatchLive:
         #                        api_key='no_key')
         #                        api_key='no_key')
         self.model = ChatOpenAI(base_url=base_url, model=model, temperature=temperature, api_key=api_key,
         self.model = ChatOpenAI(base_url=base_url, model=model, temperature=temperature, api_key=api_key,
                                 extra_body={"chat_template_kwargs": {"enable_thinking": False}})
                                 extra_body={"chat_template_kwargs": {"enable_thinking": False}})
-        
         self.cache_dir = cache_dir
         self.cache_dir = cache_dir
         self.video2frame = Video2Frame(cache_dir=cache_dir, save_frames_enable=save_frames_enable)
         self.video2frame = Video2Frame(cache_dir=cache_dir, save_frames_enable=save_frames_enable)
 
 
@@ -117,8 +116,10 @@ class FootballReplayMatchLive:
             asr_text = live_video.get("asr_text", '')
             asr_text = live_video.get("asr_text", '')
             live_video_contents = live_video.get("llm_contents", None)
             live_video_contents = live_video.get("llm_contents", None)
             if live_video_contents is None:
             if live_video_contents is None:
-                event_start = event_utc - timedelta(seconds=10) if event_utc is not None else None
-                event_end = event_utc + timedelta(seconds=5) if event_utc is not None else None
+                # event_start = event_utc - timedelta(seconds=10) if event_utc is not None else None
+                event_start = event_utc - 10000 if event_utc is not None else None
+                # event_end = event_utc + timedelta(seconds=5) if event_utc is not None else None
+                event_end = event_utc + 5000 if event_utc is not None else None
                 live_video_contents = self.video2frame.to_llm_contents( live_video_path, 
                 live_video_contents = self.video2frame.to_llm_contents( live_video_path, 
                                                                         cache=self.cache_dir,
                                                                         cache=self.cache_dir,
                                                                         fps=2, 
                                                                         fps=2, 
@@ -132,6 +133,7 @@ class FootballReplayMatchLive:
             live_map[live_video_id] = {
             live_map[live_video_id] = {
                 "video_id": live_video_id,
                 "video_id": live_video_id,
                 "video_path": live_video_path,
                 "video_path": live_video_path,
+                "event_utc": event_utc,
                 "asr_text": asr_text,
                 "asr_text": asr_text,
                 "contents": live_video_contents,
                 "contents": live_video_contents,
 
 
@@ -139,6 +141,7 @@ class FootballReplayMatchLive:
             live_records[live_video_id] = {
             live_records[live_video_id] = {
                 "video_id": live_video_id,
                 "video_id": live_video_id,
                 "video_path": live_video_path,
                 "video_path": live_video_path,
+                "event_utc": event_utc,
                 "asr_text": asr_text,
                 "asr_text": asr_text,
             }
             }
             live_videos_contents.extend(live_video_contents)
             live_videos_contents.extend(live_video_contents)
@@ -207,6 +210,7 @@ class FootballReplayMatchLive:
 
 
     def match_batch(self, replay_video: dict, live_videos: list[dict], max_parallel: int = 3, cache_dir=None):
     def match_batch(self, replay_video: dict, live_videos: list[dict], max_parallel: int = 3, cache_dir=None):
         
         
+        self.cache_dir = cache_dir
         cache_path = os.path.join(cache_dir, 'match_live.json')
         cache_path = os.path.join(cache_dir, 'match_live.json')
         if cache_path is not None and os.path.exists(cache_path):
         if cache_path is not None and os.path.exists(cache_path):
             try:
             try:
@@ -219,7 +223,7 @@ class FootballReplayMatchLive:
         event_start = replay_video.get("start_utc", None)
         event_start = replay_video.get("start_utc", None)
         event_end = replay_video.get("end_utc", None)
         event_end = replay_video.get("end_utc", None)
         replay_video_contents = self.video2frame.to_llm_contents(replay_video_path, 
         replay_video_contents = self.video2frame.to_llm_contents(replay_video_path, 
-                                                    cache=self.cache_dir,
+                                                    cache=os.path.join(os.path.dirname(cache_dir), "replay"),
                                                     fps=2, 
                                                     fps=2, 
                                                     start=event_start, 
                                                     start=event_start, 
                                                     end=event_end, 
                                                     end=event_end, 
@@ -234,6 +238,7 @@ class FootballReplayMatchLive:
             result_no_content = {
             result_no_content = {
                 "video_id": result.get("video_id", None),
                 "video_id": result.get("video_id", None),
                 "video_path": result.get("video_path", None),
                 "video_path": result.get("video_path", None),
+                "event_utc": result.get("event_utc", None),
                 "asr_text": result.get("asr_text", None),
                 "asr_text": result.get("asr_text", None),
             }
             }
         else:
         else:

@@ -78,7 +78,7 @@ class FootballReplayVideoEvent:
         self.model = ChatOpenAI(base_url=base_url, model=model, temperature=temperature, api_key=api_key,
         self.model = ChatOpenAI(base_url=base_url, model=model, temperature=temperature, api_key=api_key,
                                 extra_body={"chat_template_kwargs": {"enable_thinking": False}})
                                 extra_body={"chat_template_kwargs": {"enable_thinking": False}})
         
         
-        self.cache_dir = cache_dir
+
         self.video2frame = Video2Frame(cache_dir=cache_dir, save_frames_enable=save_frames_enable)
         self.video2frame = Video2Frame(cache_dir=cache_dir, save_frames_enable=save_frames_enable)
 
 
     def video_event(self, replay_pack: dict, asr_text: str = '无', cache_dir=None):
     def video_event(self, replay_pack: dict, asr_text: str = '无', cache_dir=None):
@@ -92,7 +92,7 @@ class FootballReplayVideoEvent:
         event_start = replay_pack.get("start_utc", None)
         event_start = replay_pack.get("start_utc", None)
         event_end = replay_pack.get("end_utc", None)
         event_end = replay_pack.get("end_utc", None)
         contents = self.video2frame.to_llm_contents(replay_video_path, 
         contents = self.video2frame.to_llm_contents(replay_video_path, 
-                                                    cache=self.cache_dir,
+                                                    cache=cache_dir,
                                                     fps=2, 
                                                     fps=2, 
                                                     start=event_start, 
                                                     start=event_start, 
                                                     end=event_end, 
                                                     end=event_end, 

@@ -87,7 +87,9 @@ class Video2Frame:
 
 
         if cache_name.exists():
         if cache_name.exists():
             logger.info(f"use_cache: {cache_name}")
             logger.info(f"use_cache: {cache_name}")
-            return json.load(cache_name.open('r'))
+            # return json.load(cache_name.open('r'))
+            with cache_name.open('r', encoding='utf-8') as f:
+                return json.load(f)
         else:
         else:
             cache_dir.mkdir(parents=True, exist_ok=True)
             cache_dir.mkdir(parents=True, exist_ok=True)
             data = {
             data = {
@@ -117,7 +119,9 @@ class Video2Frame:
                     cv2.imwrite(frame_path, frame)
                     cv2.imwrite(frame_path, frame)
 
 
             cache_name.parent.mkdir(parents=True, exist_ok=True)
             cache_name.parent.mkdir(parents=True, exist_ok=True)
-            json.dump(data, cache_name.open('w'), indent=4)
+            # json.dump(data, cache_name.open('w'), indent=4)
+            with cache_name.open('w', encoding='utf-8') as f:
+                json.dump(data, f, indent=4)
             logger.info(f"save_cache: {cache_name}")
             logger.info(f"save_cache: {cache_name}")
             return data
             return data