Refactor/memory statistics (#99)

* [refactor]Reconstructing forgotten, emotional, situational, and explicit memory statistics * [refactor]Reconstructing forgotten, emotional, situational, and explicit memory statistics * [changes]Improve the code based on AI review
2026-01-13 20:27:27 +08:00
parent 70cbda27eb
commit b71f67f7df
7 changed files with 371 additions and 26 deletions
--- a/api/app/services/memory_base_service.py
+++ b/api/app/services/memory_base_service.py
@@ -9,6 +9,7 @@ from typing import Optional

 from app.core.logging_config import get_logger
 from app.repositories.neo4j.neo4j_connector import Neo4jConnector
+from app.services.emotion_analytics_service import EmotionAnalyticsService

 logger = get_logger(__name__)

@@ -109,3 +110,188 @@ class MemoryBaseService:
        except Exception as e:
            logger.error(f"提取情景记忆情绪时出错: {str(e)}", exc_info=True)
            return None
+    
+    async def get_episodic_memory_count(
+        self,
+        end_user_id: Optional[str] = None
+    ) -> int:
+        """
+        获取情景记忆数量
+        
+        查询 MemorySummary 节点的数量。
+        
+        Args:
+            end_user_id: 可选的终端用户ID，用于过滤特定用户的节点
+            
+        Returns:
+            情景记忆的数量
+        """
+        try:
+            if end_user_id:
+                query = """
+                MATCH (n:MemorySummary)
+                WHERE n.group_id = $group_id
+                RETURN count(n) as count
+                """
+                result = await self.neo4j_connector.execute_query(query, group_id=end_user_id)
+            else:
+                query = """
+                MATCH (n:MemorySummary)
+                RETURN count(n) as count
+                """
+                result = await self.neo4j_connector.execute_query(query)
+            
+            count = result[0]["count"] if result and len(result) > 0 else 0
+            logger.debug(f"情景记忆数量: {count} (end_user_id={end_user_id})")
+            return count
+            
+        except Exception as e:
+            logger.error(f"获取情景记忆数量时出错: {str(e)}", exc_info=True)
+            return 0
+    
+    async def get_explicit_memory_count(
+        self,
+        end_user_id: Optional[str] = None
+    ) -> int:
+        """
+        获取显性记忆数量
+        
+        显性记忆 = 情景记忆（MemorySummary）+ 语义记忆（ExtractedEntity with is_explicit_memory=true）
+        
+        Args:
+            end_user_id: 可选的终端用户ID，用于过滤特定用户的节点
+            
+        Returns:
+            显性记忆的数量
+        """
+        try:
+            # 1. 获取情景记忆数量
+            episodic_count = await self.get_episodic_memory_count(end_user_id)
+            
+            # 2. 获取语义记忆数量（ExtractedEntity 且 is_explicit_memory = true）
+            if end_user_id:
+                semantic_query = """
+                MATCH (e:ExtractedEntity)
+                WHERE e.group_id = $group_id AND e.is_explicit_memory = true
+                RETURN count(e) as count
+                """
+                semantic_result = await self.neo4j_connector.execute_query(
+                    semantic_query, 
+                    group_id=end_user_id
+                )
+            else:
+                semantic_query = """
+                MATCH (e:ExtractedEntity)
+                WHERE e.is_explicit_memory = true
+                RETURN count(e) as count
+                """
+                semantic_result = await self.neo4j_connector.execute_query(semantic_query)
+            
+            semantic_count = semantic_result[0]["count"] if semantic_result and len(semantic_result) > 0 else 0
+            
+            # 3. 计算总数
+            explicit_count = episodic_count + semantic_count
+            logger.debug(
+                f"显性记忆数量: {explicit_count} "
+                f"(情景={episodic_count}, 语义={semantic_count}, end_user_id={end_user_id})"
+            )
+            return explicit_count
+            
+        except Exception as e:
+            logger.error(f"获取显性记忆数量时出错: {str(e)}", exc_info=True)
+            return 0
+    
+    async def get_emotional_memory_count(
+        self,
+        end_user_id: Optional[str] = None,
+        statement_count_fallback: int = 0
+    ) -> int:
+        """
+        获取情绪记忆数量
+        
+        通过 EmotionAnalyticsService 获取情绪标签统计总数。
+        如果获取失败或没有指定 end_user_id，使用 statement_count_fallback 作为后备。
+        
+        Args:
+            end_user_id: 可选的终端用户ID
+            statement_count_fallback: 后备方案的数量（通常是 statement 节点数量）
+            
+        Returns:
+            情绪记忆的数量
+        """
+        try:
+            if end_user_id:
+                emotion_service = EmotionAnalyticsService()
+                
+                emotion_data = await emotion_service.get_emotion_tags(
+                    end_user_id=end_user_id,
+                    emotion_type=None,
+                    start_date=None,
+                    end_date=None,
+                    limit=10
+                )
+                emotion_count = emotion_data.get("total_count", 0)
+                logger.debug(f"情绪记忆数量: {emotion_count} (end_user_id={end_user_id})")
+                return emotion_count
+            else:
+                # 如果没有指定 end_user_id，使用后备方案
+                logger.debug(f"情绪记忆数量: {statement_count_fallback} (使用后备方案)")
+                return statement_count_fallback
+                
+        except Exception as e:
+            logger.warning(f"获取情绪记忆数量失败，使用后备方案: {str(e)}")
+            return statement_count_fallback
+    
+    async def get_forget_memory_count(
+        self,
+        end_user_id: Optional[str] = None,
+        forgetting_threshold: float = 0.3
+    ) -> int:
+        """
+        获取遗忘记忆数量
+        
+        统计激活值低于遗忘阈值的节点数量（low_activation_nodes）。
+        查询范围包括：Statement、ExtractedEntity、MemorySummary、Chunk 节点。
+        
+        Args:
+            end_user_id: 可选的终端用户ID，用于过滤特定用户的节点
+            forgetting_threshold: 遗忘阈值，默认 0.3
+            
+        Returns:
+            遗忘记忆的数量（激活值低于阈值的节点数）
+        """
+        try:
+            # 构建查询语句
+            query = """
+            MATCH (n)
+            WHERE (n:Statement OR n:ExtractedEntity OR n:MemorySummary OR n:Chunk)
+            """
+            
+            if end_user_id:
+                query += " AND n.group_id = $group_id"
+            
+            query += """
+            RETURN sum(CASE WHEN n.activation_value IS NOT NULL AND n.activation_value < $threshold THEN 1 ELSE 0 END) as low_activation_nodes
+            """
+            
+            # 设置查询参数
+            params = {'threshold': forgetting_threshold}
+            if end_user_id:
+                params['group_id'] = end_user_id
+            
+            # 执行查询
+            result = await self.neo4j_connector.execute_query(query, **params)
+            
+            # 提取结果
+            forget_count = result[0]['low_activation_nodes'] if result and len(result) > 0 else 0
+            forget_count = forget_count or 0  # 处理 None 值
+            
+            logger.debug(
+                f"遗忘记忆数量: {forget_count} "
+                f"(threshold={forgetting_threshold}, end_user_id={end_user_id})"
+            )
+            return forget_count
+            
+        except Exception as e:
+            logger.error(f"获取遗忘记忆数量时出错: {str(e)}", exc_info=True)
+            return 0