fix : recover file

modules/info_extractor.py

@@ -0,0 +1,119 @@
+import re
+from .config_loader import ConfigLoader
+from utils.logger import log
+
+class InfoExtractor:
+    def __init__(self, config_loader: ConfigLoader):
+        self.configs = config_loader
+
+    def extract(self, user_input: str) -> dict:
+        """从用户输入中提取目的地、天数和旅行风格"""
+        extracted_info = {}
+        user_lower = user_input.lower()
+
+        try:
+            # 提取目的地
+            destination = self._extract_destination(user_lower)
+            if destination:
+                extracted_info["destination"] = destination
+
+            # 提取天数
+            duration = self._extract_duration(user_input)
+            if duration:
+                extracted_info["duration"] = duration
+            
+            # 提取旅行风格 (persona)
+            persona = self._extract_persona(user_input)
+            if persona:
+                extracted_info["persona"] = persona
+            
+            return extracted_info
+            
+        except Exception as e:
+            log.error(f"❌ 信息提取失败: {e}", exc_info=True)
+            return {}
+
+    def _extract_destination(self, user_lower: str) -> dict:
+        """提取目的地信息"""
+        for alias, city_info in self.configs.cities.items():
+            if alias in user_lower:
+                return city_info
+        return None
+
+    def _extract_duration(self, user_input: str) -> dict:
+        """提取天数信息 - 改进版本"""
+        patterns = [
+            (r'(\d+)\s*天', lambda x: int(x)),
+            (r'(\d+)\s*日', lambda x: int(x)),
+            (r'一周|7天|七天', lambda x: 7),
+            (r'两周|14天|十四天', lambda x: 14),
+            (r'周末', lambda x: 2),
+            (r'三天|3天', lambda x: 3),
+            (r'五天|5天', lambda x: 5),
+            (r'十天|10天', lambda x: 10),
+        ]
+        
+        for pattern, converter in patterns:
+            match = re.search(pattern, user_input)
+            if match:
+                try:
+                    if match.groups():
+                        days = converter(match.group(1))
+                    else:
+                        days = converter(None)
+                    
+                    if 1 <= days <= 30:
+                        return {"days": days, "description": f"{days}天旅行"}
+                except (ValueError, TypeError):
+                    continue
+        return None
+
+    def _extract_persona(self, user_input: str) -> dict:
+        """提取用户偏好 - 根据实际personas.json结构"""
+        user_lower = user_input.lower()
+    
+        # 检查personas配置
+        for persona_key, persona_info in self.configs.personas.items():
+            # 检查persona名称中的关键词
+            persona_name = persona_info.get('name', '').lower()
+            # 移除emoji并检查关键词
+            clean_name = ''.join(char for char in persona_name if char.isalpha() or char.isspace())
+        
+            if any(keyword in user_lower for keyword in clean_name.split() if len(keyword) > 1):
+                return {**persona_info, "key": persona_key}  # 添加key方便后续使用
+        
+            # 检查风格关键词
+            style = persona_info.get('style', '').lower()
+            if any(keyword in user_lower for keyword in style.split('、') if len(keyword) > 1):
+                return {**persona_info, "key": persona_key}
+        
+            # 检查特征描述
+            characteristics = persona_info.get('characteristics', [])
+            for char in characteristics:
+                char_keywords = char.lower().split()[:3]  # 取前3个词作为关键词
+                if any(keyword in user_lower for keyword in char_keywords if len(keyword) > 1):
+                    return {**persona_info, "key": persona_key}
+    
+        # 基于关键词的简单映射
+        keyword_persona_map = {
+            '规划': 'planner',
+            '计划': 'planner', 
+            '效率': 'planner',
+            '预算': 'planner',
+            '分享': 'social',
+            '朋友': 'social',
+            '拍照': 'social',
+            '打卡': 'social',
+            '深度': 'experiential',
+            '文化': 'experiential',
+            '地道': 'experiential',
+            '当地': 'experiential'
+        }
+    
+        for keyword, persona_key in keyword_persona_map.items():
+            if keyword in user_input:
+                persona_info = self.configs.personas.get(persona_key, {})
+                if persona_info:
+                    return {**persona_info, "key": persona_key}
+    
+        return None