improve: budget mapping

modules/info_extractor.py

@@ -756,90 +756,85 @@ class InfoExtractor:
     def _extract_budget_from_tokens(self, tokens: list) -> dict:
         
         result = {}
-    
-        # 1. 查找金额和货币单位（不再有默认值）
-        for i, token in enumerate(tokens):
-            amount = None
-            currency = None  # <--- 修改点：不再预设 "RMB"，初始为 None
-        
-            # 处理包含货币的token "2000欧", "5000元"
-            currency_patterns = [
-                (r'(\d+(?:\.\d+)?)欧(?:元)?', 'EUR'),
-                (r'(\d+(?:\.\d+)?)元', 'RMB'),
-                (r'(\d+(?:\.\d+)?)块(?:钱)?', 'RMB'),
-                (r'(\d+(?:\.\d+)?)人民币', 'RMB'),
-                (r'(\d+(?:\.\d+)?)美元', 'USD'),
-                (r'(\d+(?:\.\d+)?)英镑', 'GBP'),
-                (r'(\d+(?:\.\d+)?)瑞(?:士)?法郎', 'CHF'),
-                (r'(\d+(?:\.\d+)?)日元', 'JPY'),
-                (r'(\d+(?:\.\d+)?)韩元', 'KRW'),
-                (r'¥(\d+(?:\.\d+)?)', 'RMB'),
-                (r'€(\d+(?:\.\d+)?)', 'EUR'),
-                (r'\$(\d+(?:\.\d+)?)', 'USD'),
-                (r'£(\d+(?:\.\d+)?)', 'GBP'),
-                (r'(\d+(?:\.\d+)?)rmb', 'RMB'),
-                (r'(\d+(?:\.\d+)?)usd', 'USD'),
-                (r'(\d+(?:\.\d+)?)eur', 'EUR'),
-                (r'(\d+(?:\.\d+)?)gbp', 'GBP'),
-                (r'(\d+(?:\.\d+)?)chf', 'CHF'),
-            ]
-        
-            for pattern, curr in currency_patterns:
-                match = re.search(pattern, token.lower())
-                if match:
-                    amount = float(match.group(1))
-                    currency = curr
-                    break
-        
-            # 处理纯数字token（需要查看上下文）
-            if not amount and re.match(r'^\\d+(?:\\.\\d+)?$', token):
-                number = float(token)
-            
-                # 检查前面的token是否有预算相关词汇
-                budget_indicators = ['预算', '花费', '费用', '成本', '开销', '支出', '总共', '一共', 'budget', 'cost', 'spend']
-                has_budget_context = False
-                if i > 0 and tokens[i-1] in budget_indicators:
-                    has_budget_context = True
-                elif i > 1 and tokens[i-2] in budget_indicators:
-                    has_budget_context = True
-            
-                # 检查后面是否有货币单位
-                if i < len(tokens) - 1:
-                    next_token = tokens[i + 1].lower()
-                    currency_units = {
-                        '元': 'RMB', '块': 'RMB', '钱': 'RMB', '人民币': 'RMB',
-                        '欧': 'EUR', '欧元': 'EUR', '美元': 'USD', '英镑': 'GBP',
-                        '瑞郎': 'CHF', '日元': 'JPY', '韩元': 'KRW',
-                        'rmb': 'RMB', 'usd': 'USD', 'eur': 'EUR', 'gbp': 'GBP', 'chf': 'CHF'
-                    }
-                    if next_token in currency_units:
-                        amount = number
-                        currency = currency_units[next_token]
-            
-                if has_budget_context and not currency:
-                    amount = number
+        text = "".join(tokens).lower()
 
-            # 处理万、千等单位
-            if amount:
-                # (这部分逻辑保持不变)
-                if i > 0:
-                    prev_token = tokens[i-1]
-                    if '万' in prev_token or 'w' in prev_token.lower():
-                        amount *= 10000
-                    elif '千' in prev_token or 'k' in prev_token.lower():
-                        amount *= 1000
-                elif i < len(tokens) - 1:
-                    next_token = tokens[i+1]
-                    if '万' in next_token or 'w' in next_token.lower():
-                        amount *= 10000
-                    elif '千' in next_token or 'k' in next_token.lower():
-                        amount *= 1000
-            
-                if amount > 0:
-                    result["amount"] = int(amount)
-                    if currency: # 只有当识别到货币时才赋值
-                        result["currency"] = currency
-                    break
+        # --- 1. 统一提取金额和货币 ---
+        # 按优先级排列正则表达式，越精确的模式越靠前
+        # 模式捕获: (金额数字, 乘数单位[千/万/k/w], 货币单位[元/欧/usd...])
+        patterns = [
+            {'regex': r'¥\s*(\d+\.?\d*)', 'groups': {'amount': 1}, 'currency': 'RMB'},
+            {'regex': r'€\s*(\d+\.?\d*)', 'groups': {'amount': 1}, 'currency': 'EUR'},
+            {'regex': r'\$\s*(\d+\.?\d*)', 'groups': {'amount': 1}, 'currency': 'USD'},
+            {'regex': r'£\s*(\d+\.?\d*)', 'groups': {'amount': 1}, 'currency': 'GBP'},
+            {'regex': r'(usd|rmb|eur|gbp|chf|jpy)\s*(\d+\.?\d*)\s*(百|hundred|千|k|thousand|万|w)?', 
+            'groups': {'currency': 1, 'amount': 2, 'multiplier': 3}},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(人民币|元|块|块钱|rmb)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'RMB'},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(欧元|欧|euros?|eur)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'EUR'},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(美元|dollars?|dollar|usd)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'USD'},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(英镑|pounds?|pound|gbp)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'GBP'},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(日元|yen|jpy)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'JPY'},
+            {'regex': r'(\d+\.?\d*)\s*(十|百|hundred|千|k|thousand|万|w)?\s*(瑞郎|瑞士法郎|chf)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'currency': 'CHF'},
+
+            {'regex': r'(\d+\.?\d+)\s*(十|百|hundred|千|k|thousand|万|w)', 
+            'groups': {'amount': 1, 'multiplier': 2}, 'context_needed': True},
+            {'regex': r'(\d+\.?\d+)', 
+            'groups': {'amount': 1}, 'context_needed': True},
+        ]
+
+        amount_found = False
+        for pattern, default_multiplier, default_currency in patterns:
+            match = re.search(pattern, text)
+            if match:
+                # 检查是否是纯数字模式，是的话需要上下文
+                if pattern in [r'(\d+\.?\d+)\s*(千|k|万|w)', r'(\d+\.?\d+)']:
+                    budget_indicators = ['预算', '花费', '费用', '成本', '开销', '支出', 'budget', 'cost']
+                    if not any(indicator in text for indicator in budget_indicators):
+                        continue # 如果没有上下文，则跳过纯数字匹配
+
+                groups = match.groups()
+            
+                # 提取金额
+                amount = float(groups[0])
+            
+                # 确定乘数
+                multiplier = 1
+                multiplier_token = ''
+                if 'multiplier' in p['groups'] and p['groups']['multiplier'] <= len(groups) and groups[p['groups']['multiplier']-1]:
+                    multiplier_token = groups[p['groups']['multiplier']-1]
+            
+                if '十' in multiplier_token:
+                    multiplier = 10
+                elif '百' in multiplier_token or 'hundred' in multiplier_token:
+                    multiplier = 100
+                elif '千' in multiplier_token or 'k' in multiplier_token or 'thousand' in multiplier_token:
+                    multiplier = 1000
+                elif '万' in multiplier_token or 'w' in multiplier_token:
+                    multiplier = 10000
+                
+                final_amount = amount * multiplier
+                result['amount'] = int(final_amount)
+
+                # 确定货币
+                currency_token = ''
+                if default_currency:
+                    result['currency'] = default_currency
+                elif len(groups) > 2 and groups[2]:
+                    currency_token = groups[2]
+            
+                currency_map = {
+                    'rmb': 'RMB', 'usd': 'USD', 'eur': 'EUR', 'gbp': 'GBP', 'chf': 'CHF', 'jpy': 'JPY'
+                }
+                if currency_token in currency_map:
+                    result['currency'] = currency_map[currency_token]
+            
+                amount_found = True
+                break
     
         # 2. 查找预算类型（此部分逻辑与金额完全无关）
         budget_type_keywords = {