jira:NYJ-1656 desc: modified cache dir and url of m3u8/mp4

lizhengwei
Commit b725d68dd12878bd7990eb330c1799c34916d2d6 b725d68d 1 parent cfcc7f7a
Showing 3 changed files with 45 additions and 190 deletions
src/football_replay_match/core/api.py
src/football_replay_match/core/api_test.py
src/football_replay_match/core/utils/football_replay_match_live.py
--- a/src/football_replay_match/core/api.py
View file @b725d68
+++ b/src/football_replay_match/core/api.py
View file @b725d68
@@ -56,11 +56,21 @@ class FootballReplayMatch:
             return None
     def det_goal_replay(self, replay, task_id):
-        return self.videoEventRecognition.video_event(replay, cache_dir=os.path.join(self.cache_dir, "replays"), task_log_dir=os.path.join(self.task_log_dir, task_id))
+        video_url = replay.get('url')
+        if video_url.endswith('.m3u8'):
+            mp4_name = os.path.basename(video_url)
+        else:
+            mp4_name = os.path.basename(os.path.dirname(os.path.dirname(os.path.dirname(video_url))))
+        return self.videoEventRecognition.video_event(replay, cache_dir=os.path.join(self.cache_dir, mp4_name, "replays"), task_log_dir=os.path.join(self.task_log_dir, mp4_name, task_id))
     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, "events"), task_log_dir=os.path.join(self.task_log_dir, task_id))
+        video_url = replay.get('url')
+        if video_url.endswith('.m3u8'):
+            mp4_name = os.path.basename(video_url)
+        else:
+            mp4_name = os.path.basename(os.path.dirname(os.path.dirname(os.path.dirname(video_url))))
+        return self.videoMatchLive.match_batch(replay, events, max_parallel=2, cache_dir=os.path.join(self.cache_dir, mp4_name, "events"), task_log_dir=os.path.join(self.task_log_dir, mp4_name, task_id))
     def replay_match_event(self, data):
         """
--- a/src/football_replay_match/core/api_test.py
View file @b725d68
+++ b/src/football_replay_match/core/api_test.py
View file @b725d68
 """
-api_test.py - 对 api.py 中 FootballReplayMatch.replay_match_event 的单元测试
+api_test.py - 对 api.py 中 FootballReplayMatch.replay_match_event 的测试
 使用伪数据覆盖主要分支：
 1. 时间阈值内直接匹配成功
@@ -20,10 +20,6 @@ try:
 except ImportError:
     from .api import FootballReplayMatch
-    
-    
-    
-
 from pathlib import Path
 def get_video_files_pathlib(folder_path, extensions=None):
     if extensions is None:
@@ -36,17 +32,17 @@ def get_video_files_pathlib(folder_path, extensions=None):
                    if f.is_file() and f.suffix.lower() in extensions]
     return sorted(image_files)
-def _fake_settings(match_by_time_threshold: int = 30, cache_dir_tag: str = "test"):
+def _fake_settings():
     """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,
+        match_by_time_threshold=30,
         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, 
                             api_key='no_key'
                             ),
-        common=SimpleNamespace(cache_dir=f"/root/lzw/tmp_0601_replay_cache_{cache_dir_tag}/videos_cache", 
-                               task_log_dir=f"/root/lzw/tmp_0601_replay_cache_{cache_dir_tag}/tasks_log"
+        common=SimpleNamespace(cache_dir=f"/root/lzw/tmp_0602_replay_cache/videos_cache", 
+                               task_log_dir=f"/root/lzw/tmp_0602_replay_cache/tasks_log"
                                ),
         save_frames_enable=True,
     )
@@ -102,36 +98,21 @@ def _base_data(i, replay_path, event_lis):
     }
-
-
-
-
 def test_match_by_time_success():
-    # """场景1：时间差在阈值内，直接通过 match_by_time 命中。"""
-    
+
     replays_and_events_pair = get_replays_and_events_pair()
-    
     for replays_mp4_dir, events_mp4_dir in replays_and_events_pair:
         # replays_mp4_dir = "/root/lzw/finished/69e5102bfd87d43d09bba90d-auto-10154993/replays/videos"
         # events_mp4_dir = r"/root/lzw/finished/69e5102bfd87d43d09bba90d-auto-10154993/live/videos"
         print(100*"=")
         print(f"replays_mp4_dir: {replays_mp4_dir}, events_mp4_dir: {events_mp4_dir}")
-        tag = os.path.basename(os.path.dirname(os.path.dirname(replays_mp4_dir))) 
-        settings = _fake_settings(match_by_time_threshold=30, cache_dir_tag=tag)  # threshold = 30 * 1000 = 30000 ms
-        
-        
+        settings = _fake_settings()
         frm = FootballReplayMatch(settings)
-        
         relays_mp4_names = get_video_files_pathlib(replays_mp4_dir)
-        
-        
-        
         event_list = get_event_list(events_mp4_dir)
-        
-        
         all_time = 0.0
         max_time = 0.0
         for i, replay_name in enumerate(relays_mp4_names):
@@ -157,145 +138,5 @@ def test_match_by_time_success():
         print(f"all_time: {all_time}")
         # exit()
-
-
-
 if __name__ == "__main__":
     test_match_by_time_success()
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-    #     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)
--- a/src/football_replay_match/core/utils/football_replay_match_live.py
View file @b725d68
+++ b/src/football_replay_match/core/utils/football_replay_match_live.py
View file @b725d68
@@ -218,29 +218,33 @@ class FootballReplayMatchLive:
         if result.strip() == "null":
             result_json = {"video_id": None}
         else:
-            try:
-                # 1. 深度清洗：去掉 Markdown 标记
-                cleaned_result = result.replace("```json", "").replace("```", "")
-                
-                # 2. 核心修复：使用 re.DOTALL 确保正则能跨行匹配，精准提取最外层完整的 JSON 对象
-                # 这样可以防止因为模型输出换行导致提取不完整
-                json_match = re.search(r'(\{.*\})', cleaned_result, re.DOTALL)
-                
-                if json_match:
-                    json_str = json_match.group(1)
-                    # 截断第一个 } 之后的所有内容，彻底干掉 <time_location> 等尾巴
-                    json_str = json_str[:json_str.rfind('}') + 1]
-                else:
-                    # 如果连花括号都找不到，直接抛出原始内容方便排查
-                    print("无法提取JSON，原始内容:", result)
-                    raise ValueError("响应中未找到有效的 JSON 对象")
-
-                # 3. 使用 strict=False 容忍字符串中的非法控制字符（如直接换行）
-                result_json = json.loads(json_str, strict=False)
+
+            # 1. 深度清洗：去掉 Markdown 标记
+            cleaned_result = result.replace("```json", "").replace("```", "")
+            
+            # 2. 核心修复：使用 re.DOTALL 确保正则能跨行匹配，精准提取最外层完整的 JSON 对象
+            # 这样可以防止因为模型输出换行导致提取不完整
+            json_match = re.search(r'(\{.*\})', cleaned_result, re.DOTALL)
+            
+            if json_match:
+                json_str = json_match.group(1)
+                # 截断第一个 } 之后的所有内容，彻底干掉 <time_location> 等尾巴
+                json_str = json_str[:json_str.rfind('}') + 1]
+            else:
+                # 如果连花括号都找不到，直接抛出原始内容方便排查
+                json_str = {
+                    "replay_summary": {
+                        "visual_cues": ""
+                    },
+                    "video_id": "null"
+                }
+                logger.info("Unable_to_extract_JSON,original_content:", result)
+ 
+
+            # 3. 使用 strict=False 容忍字符串中的非法控制字符（如直接换行）
+            result_json = json.loads(json_str, strict=False)
-            except Exception as e:
-                print("JSON解析依然失败，清洗后的内容:", json_str if 'json_str' in locals() else result)
-                raise e
+
         video_id = result_json.get("video_id", None)
         result_live = live_map.get(video_id, None)