feat(api-converter): 添加进销存采购订单API配置并实现本地缓存机制

新增api_config_9p04kww1pu.json配置文件，包含进销存采购订单相关的四个核心
API接口（查询列表、新建、详情、编辑），完善了load_api_configs函数，
增加本地文件缓存机制，支持从本地文件加载配置并在配置变更时同步保存，
优化refresh_api_configs函数以同步清理本地文件缓存。

BREAKING CHANGE: API配置方式调整，引入本地缓存机制可能影响原有部署流程

lzwcai_workflow_to_mcp/lzwcai_workflow_to_mcp/main.py

@@ -19,12 +19,12 @@ from mcp.server.models import InitializationOptions
 from mcp.server.stdio import stdio_server
 
 try:
-    from .schema_converter import convert_sql_params_to_input_schema
+    from .schema_converter import convert_sql_params_to_input_schema, sanitize_json_schema
     from .utils.api_client import execute_workflow, get_workflow_by_id, extract_final_output
     from .utils.env_config import get_workflow_id
     from .utils.logger_config import setup_system_logging, get_logger
 except ImportError:
-    from schema_converter import convert_sql_params_to_input_schema
+    from schema_converter import convert_sql_params_to_input_schema, sanitize_json_schema
     from utils.api_client import execute_workflow, get_workflow_by_id, extract_final_output
     from utils.env_config import get_workflow_id
     from utils.logger_config import setup_system_logging, get_logger
@@ -174,13 +174,30 @@ async def handle_list_tools() -> list[types.Tool]:
     for query in _tools_config:
         name = query.get("name", "")
         description = query.get("description") or query.get("toolPrompt") or query.get("uniqueName", "")
-        sql_params = query.get("sqlParams", "[]")
         
         logger.debug(f"处理工具配置: name={name}, description={description[:50] if description else 'None'}...")
         
-        # 转换参数为 inputSchema
-        input_schema = convert_sql_params_to_input_schema(sql_params)
-        logger.debug(f"工具 {name} 的 inputSchema: {json.dumps(input_schema, ensure_ascii=False)}")
+        # 优先使用 inputJsonSchema，如果不存在则从 sqlParams 转换
+        input_json_schema = query.get("inputJsonSchema")
+        if input_json_schema:
+            try:
+                # inputJsonSchema 是 JSON 字符串，需要解析
+                if isinstance(input_json_schema, str):
+                    input_schema = json.loads(input_json_schema)
+                else:
+                    input_schema = input_json_schema
+                # 清理 schema，确保 enum 只包含字符串
+                input_schema = sanitize_json_schema(input_schema)
+                logger.debug(f"使用 inputJsonSchema (已清理): {json.dumps(input_schema, ensure_ascii=False)}")
+            except (json.JSONDecodeError, TypeError) as e:
+                logger.warning(f"解析 inputJsonSchema 失败: {e}，回退到 sqlParams")
+                sql_params = query.get("sqlParams", "[]")
+                input_schema = convert_sql_params_to_input_schema(sql_params)
+        else:
+            # 从 sqlParams 转换
+            sql_params = query.get("sqlParams", "[]")
+            input_schema = convert_sql_params_to_input_schema(sql_params)
+            logger.debug(f"从 sqlParams 转换的 inputSchema: {json.dumps(input_schema, ensure_ascii=False)}")
         
         tools.append(
             types.Tool(

lzwcai_workflow_to_mcp/lzwcai_workflow_to_mcp/schema_converter.py

@@ -12,6 +12,41 @@ import json
 from typing import Any
 
 
+def sanitize_json_schema(schema: dict) -> dict:
+    """
+    清理 JSON Schema，确保 enum 字段只包含字符串值
+    
+    Args:
+        schema: 原始 JSON Schema
+    
+    Returns:
+        dict: 清理后的 JSON Schema
+    """
+    if not isinstance(schema, dict):
+        return schema
+    
+    # 递归处理 properties
+    if "properties" in schema and isinstance(schema["properties"], dict):
+        for prop_name, prop_schema in schema["properties"].items():
+            if isinstance(prop_schema, dict) and "enum" in prop_schema:
+                enum_values = prop_schema["enum"]
+                if isinstance(enum_values, list) and len(enum_values) > 0:
+                    # 检查是否是对象数组
+                    if isinstance(enum_values[0], dict):
+                        # 提取 value 字段
+                        cleaned_enum = [
+                            item.get("value", item.get("label", ""))
+                            for item in enum_values
+                            if isinstance(item, dict)
+                        ]
+                        # 过滤空值
+                        prop_schema["enum"] = [v for v in cleaned_enum if v]
+                        # 保留原始数据
+                        prop_schema["x-options"] = enum_values
+    
+    return schema
+
+
 def convert_param_to_schema_property(param: dict) -> tuple[str, dict, bool]:
     """
     将单个参数转换为 JSON Schema property
@@ -57,7 +92,22 @@ def convert_param_to_schema_property(param: dict) -> tuple[str, dict, bool]:
     elif param_type == "select":
         property_schema["type"] = "string"
         if options:
-            property_schema["enum"] = options
+            # 处理 options 可能是对象数组 [{label, value}] 或字符串数组的情况
+            if isinstance(options, list) and len(options) > 0:
+                if isinstance(options[0], dict):
+                    # 对象数组，提取 value 字段作为 enum，过滤空值
+                    enum_values = [
+                        opt.get("value", opt.get("label", "")) 
+                        for opt in options 
+                        if isinstance(opt, dict)
+                    ]
+                    # 过滤掉空字符串
+                    property_schema["enum"] = [v for v in enum_values if v]
+                    # 保留原始 options 供前端使用（如果前端支持）
+                    property_schema["x-options"] = options
+                else:
+                    # 字符串数组，直接使用，过滤空值
+                    property_schema["enum"] = [v for v in options if v]
             
     elif param_type == "number":
         property_schema["type"] = "number"

lzwcai_workflow_to_mcp/main.py

@@ -7,8 +7,8 @@ import os
 
 if __name__ == "__main__":
     # 设置环境变量
-    os.environ["workflowId"] = "2005892514011795457"
-    os.environ["workflowExecuteKey"] = "wf_ce270212b2ee45ab9c81714a7c243c56"
+    os.environ["workflowId"] = "2019984063311556609"
+    os.environ["workflowExecuteKey"] = "wf_45bd630402664485a2ef5cd3ede5aa53"
     os.environ["backendBaseUrl"] = "http://192.168.2.236:8088"
     
     # Import and run the actual MCP server