feat: add optional structured market sentiment tool for QueryAgent

.env.example

@@ -64,6 +64,8 @@ KEYWORD_OPTIMIZER_MODEL_NAME=
 # ================== 网络工具配置 ====================
 # Tavily API密钥，用于Tavily网络搜索，申请地址：https://www.tavily.com/
 TAVILY_API_KEY=
+# Adanos Market Sentiment API 密钥（可选，用于结构化股票与市场情绪研究）
+ADANOS_API_KEY=
 
 # 网络搜索工具类型，支持BochaAPI或AnspireAPI两种，默认为AnspireAPI
 SEARCH_TOOL_TYPE=AnspireAPI
@@ -74,4 +76,4 @@ ANSPIRE_API_KEY=
 
 # Bocha AI Search API（用于Bocha多模态搜索，这里密钥名称虽然是Web Search，但其实是要AI Search的，申请地址：https://open.bochaai.com/）
 BOCHA_BASE_URL=https://api.bocha.cn/v1/ai-search
-BOCHA_WEB_SEARCH_API_KEY=
+BOCHA_WEB_SEARCH_API_KEY=

QueryEngine/agent.py

@@ -19,7 +19,7 @@
     ReportFormattingNode
 )
 from .state import State
-from .tools import TavilyNewsAgency, TavilyResponse
+from .tools import AdanosSentimentAgency, TavilyNewsAgency, TavilyResponse
 from .utils import Settings, format_search_results_for_prompt
 from loguru import logger
 
@@ -42,6 +42,11 @@ def __init__(self, config: Optional[Settings] = None):
 
         # 初始化搜索工具集
         self.search_agency = TavilyNewsAgency(api_key=self.config.TAVILY_API_KEY)
+        self.market_sentiment_agency = (
+            AdanosSentimentAgency(api_key=self.config.ADANOS_API_KEY)
+            if self.config.ADANOS_API_KEY
+            else None
+        )
 
         # 初始化节点
         self._initialize_nodes()
@@ -54,7 +59,10 @@ def __init__(self, config: Optional[Settings] = None):
 
         logger.info(f"Query Agent已初始化")
         logger.info(f"使用LLM: {self.llm_client.get_model_info()}")
-        logger.info(f"搜索工具集: TavilyNewsAgency (支持6种搜索工具)")
+        search_tools = "TavilyNewsAgency (支持6种新闻搜索工具)"
+        if self.market_sentiment_agency:
+            search_tools += " + AdanosSentimentAgency (结构化市场情绪工具)"
+        logger.info(f"搜索工具集: {search_tools}")
 
     def _initialize_llm(self) -> LLMClient:
         """初始化LLM客户端"""
@@ -66,8 +74,15 @@ def _initialize_llm(self) -> LLMClient:
 
     def _initialize_nodes(self):
         """初始化处理节点"""
-        self.first_search_node = FirstSearchNode(self.llm_client)
-        self.reflection_node = ReflectionNode(self.llm_client)
+        enable_market_sentiment = bool(self.config.ADANOS_API_KEY)
+        self.first_search_node = FirstSearchNode(
+            self.llm_client,
+            enable_market_sentiment=enable_market_sentiment,
+        )
+        self.reflection_node = ReflectionNode(
+            self.llm_client,
+            enable_market_sentiment=enable_market_sentiment,
+        )
         self.first_summary_node = FirstSummaryNode(self.llm_client)
         self.reflection_summary_node = ReflectionSummaryNode(self.llm_client)
         self.report_formatting_node = ReportFormattingNode(self.llm_client)
@@ -96,6 +111,30 @@ def _validate_date_format(self, date_str: str) -> bool:
             return True
         except ValueError:
             return False
+
+    def _build_search_kwargs(self, tool_name: str, tool_output: Dict[str, Any], log_prefix: str) -> tuple[str, Dict[str, Any]]:
+        """Normalize optional tool arguments and downgrade invalid date searches safely."""
+        search_kwargs: Dict[str, Any] = {}
+
+        if tool_name != "search_news_by_date":
+            return tool_name, search_kwargs
+
+        start_date = tool_output.get("start_date")
+        end_date = tool_output.get("end_date")
+
+        if start_date and end_date:
+            if self._validate_date_format(start_date) and self._validate_date_format(end_date):
+                search_kwargs["start_date"] = start_date
+                search_kwargs["end_date"] = end_date
+                logger.info(f"{log_prefix}时间范围: {start_date} 到 {end_date}")
+                return tool_name, search_kwargs
+
+            logger.info(f"{log_prefix}⚠️  日期格式错误（应为YYYY-MM-DD），改用基础搜索")
+            logger.info(f"{log_prefix}    提供的日期: start_date={start_date}, end_date={end_date}")
+            return "basic_search_news", search_kwargs
+
+        logger.info(f"{log_prefix}⚠️  search_news_by_date工具缺少时间参数，改用基础搜索")
+        return "basic_search_news", search_kwargs
 
     def execute_search_tool(self, tool_name: str, query: str, **kwargs) -> TavilyResponse:
         """
@@ -109,6 +148,7 @@ def execute_search_tool(self, tool_name: str, query: str, **kwargs) -> TavilyRes
                 - "search_news_last_week": 本周新闻
                 - "search_images_for_news": 新闻图片搜索
                 - "search_news_by_date": 按日期范围搜索新闻
+                - "search_market_sentiment": 结构化股票与市场情绪研究
             query: 搜索查询
             **kwargs: 额外参数（如start_date, end_date, max_results）
 
@@ -134,6 +174,16 @@ def execute_search_tool(self, tool_name: str, query: str, **kwargs) -> TavilyRes
             if not start_date or not end_date:
                 raise ValueError("search_news_by_date工具需要start_date和end_date参数")
             return self.search_agency.search_news_by_date(query, start_date, end_date)
+        elif tool_name == "search_market_sentiment":
+            if not self.market_sentiment_agency:
+                logger.info("  → Adanos市场情绪工具未配置，跳过结构化情绪研究")
+                return TavilyResponse(
+                    query=query,
+                    answer="Structured market sentiment tool is not configured.",
+                    results=[],
+                )
+            days = kwargs.get("days", 7)
+            return self.market_sentiment_agency.search_market_sentiment(query, days=days)
         else:
             logger.warning(f"  ⚠️  未知的搜索工具: {tool_name}，使用默认基础搜索")
             return self.search_agency.basic_search_news(query)
@@ -238,26 +288,7 @@ def _initial_search_and_summary(self, paragraph_index: int):
         # 执行搜索
         logger.info("  - 执行网络搜索...")
 
-        # 处理search_news_by_date的特殊参数
-        search_kwargs = {}
-        if search_tool == "search_news_by_date":
-            start_date = search_output.get("start_date")
-            end_date = search_output.get("end_date")
-
-            if start_date and end_date:
-                # 验证日期格式
-                if self._validate_date_format(start_date) and self._validate_date_format(end_date):
-                    search_kwargs["start_date"] = start_date
-                    search_kwargs["end_date"] = end_date
-                    logger.info(f"  - 时间范围: {start_date} 到 {end_date}")
-                else:
-                    logger.info(f"  ⚠️  日期格式错误（应为YYYY-MM-DD），改用基础搜索")
-                    logger.info(f"      提供的日期: start_date={start_date}, end_date={end_date}")
-                    search_tool = "basic_search_news"
-            else:
-                logger.info(f"  ⚠️  search_news_by_date工具缺少时间参数，改用基础搜索")
-                search_tool = "basic_search_news"
-
+        search_tool, search_kwargs = self._build_search_kwargs(search_tool, search_output, "  - ")
         search_response = self.execute_search_tool(search_tool, search_query, **search_kwargs)
 
         # 转换为兼容格式
@@ -328,27 +359,7 @@ def _reflection_loop(self, paragraph_index: int):
             logger.info(f"    选择的工具: {search_tool}")
             logger.info(f"    反思推理: {reasoning}")
 
-            # 执行反思搜索
-            # 处理search_news_by_date的特殊参数
-            search_kwargs = {}
-            if search_tool == "search_news_by_date":
-                start_date = reflection_output.get("start_date")
-                end_date = reflection_output.get("end_date")
-
-                if start_date and end_date:
-                    # 验证日期格式
-                    if self._validate_date_format(start_date) and self._validate_date_format(end_date):
-                        search_kwargs["start_date"] = start_date
-                        search_kwargs["end_date"] = end_date
-                        logger.info(f"    时间范围: {start_date} 到 {end_date}")
-                    else:
-                        logger.info(f"    ⚠️  日期格式错误（应为YYYY-MM-DD），改用基础搜索")
-                        logger.info(f"        提供的日期: start_date={start_date}, end_date={end_date}")
-                        search_tool = "basic_search_news"
-                else:
-                    logger.info(f"    ⚠️  search_news_by_date工具缺少时间参数，改用基础搜索")
-                    search_tool = "basic_search_news"
-
+            search_tool, search_kwargs = self._build_search_kwargs(search_tool, reflection_output, "    ")
             search_response = self.execute_search_tool(search_tool, search_query, **search_kwargs)
 
             # 转换为兼容格式

QueryEngine/nodes/search_node.py

@@ -4,12 +4,12 @@
 """
 
 import json
-from typing import Dict, Any
+from typing import Any, Dict
 from json.decoder import JSONDecodeError
 from loguru import logger
 
 from .base_node import BaseNode
-from ..prompts import SYSTEM_PROMPT_FIRST_SEARCH, SYSTEM_PROMPT_REFLECTION
+from ..prompts import build_first_search_prompt, build_reflection_prompt
 from ..utils.text_processing import (
     remove_reasoning_from_output,
     clean_json_tags,
@@ -21,14 +21,15 @@
 class FirstSearchNode(BaseNode):
     """为段落生成首次搜索查询的节点"""
 
-    def __init__(self, llm_client):
+    def __init__(self, llm_client, enable_market_sentiment: bool = False):
         """
         初始化首次搜索节点
 
         Args:
             llm_client: LLM客户端
         """
         super().__init__(llm_client, "FirstSearchNode")
+        self.system_prompt = build_first_search_prompt(enable_market_sentiment)
 
     def validate_input(self, input_data: Any) -> bool:
         """验证输入数据"""
@@ -42,7 +43,7 @@ def validate_input(self, input_data: Any) -> bool:
             return "title" in input_data and "content" in input_data
         return False
 
-    def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
+    def run(self, input_data: Any, **kwargs) -> Dict[str, Any]:
         """
         调用LLM生成搜索查询和理由
 
@@ -66,7 +67,7 @@ def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
             logger.info("正在生成首次搜索查询")
 
             # 调用LLM
-            response = self.llm_client.stream_invoke_to_string(SYSTEM_PROMPT_FIRST_SEARCH, message)
+            response = self.llm_client.stream_invoke_to_string(self.system_prompt, message)
 
             # 处理响应
             processed_response = self.process_output(response)
@@ -78,7 +79,7 @@ def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
             logger.exception(f"生成首次搜索查询失败: {str(e)}")
             raise e
 
-    def process_output(self, output: str) -> Dict[str, str]:
+    def process_output(self, output: str) -> Dict[str, Any]:
         """
         处理LLM输出，提取搜索查询和推理
 
@@ -123,22 +124,28 @@ def process_output(self, output: str) -> Dict[str, str]:
             # 验证和清理结果
             search_query = result.get("search_query", "")
             reasoning = result.get("reasoning", "")
+            search_tool = result.get("search_tool") or "basic_search_news"
+            start_date = result.get("start_date")
+            end_date = result.get("end_date")
 
             if not search_query:
                 logger.warning("未找到搜索查询，使用默认查询")
                 return self._get_default_search_query()
 
             return {
                 "search_query": search_query,
-                "reasoning": reasoning
+                "search_tool": search_tool,
+                "reasoning": reasoning,
+                "start_date": start_date,
+                "end_date": end_date,
             }
 
         except Exception as e:
             self.log_error(f"处理输出失败: {str(e)}")
             # 返回默认查询
             return self._get_default_search_query()
 
-    def _get_default_search_query(self) -> Dict[str, str]:
+    def _get_default_search_query(self) -> Dict[str, Any]:
         """
         获取默认搜索查询
 
@@ -147,21 +154,25 @@ def _get_default_search_query(self) -> Dict[str, str]:
         """
         return {
             "search_query": "相关主题研究",
-            "reasoning": "由于解析失败，使用默认搜索查询"
+            "search_tool": "basic_search_news",
+            "reasoning": "由于解析失败，使用默认搜索查询",
+            "start_date": None,
+            "end_date": None,
         }
 
 
 class ReflectionNode(BaseNode):
     """反思段落并生成新搜索查询的节点"""
 
-    def __init__(self, llm_client):
+    def __init__(self, llm_client, enable_market_sentiment: bool = False):
         """
         初始化反思节点
 
         Args:
             llm_client: LLM客户端
         """
         super().__init__(llm_client, "ReflectionNode")
+        self.system_prompt = build_reflection_prompt(enable_market_sentiment)
 
     def validate_input(self, input_data: Any) -> bool:
         """验证输入数据"""
@@ -177,7 +188,7 @@ def validate_input(self, input_data: Any) -> bool:
             return all(field in input_data for field in required_fields)
         return False
 
-    def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
+    def run(self, input_data: Any, **kwargs) -> Dict[str, Any]:
         """
         调用LLM反思并生成搜索查询
 
@@ -201,7 +212,7 @@ def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
             logger.info("正在进行反思并生成新搜索查询")
 
             # 调用LLM
-            response = self.llm_client.stream_invoke_to_string(SYSTEM_PROMPT_REFLECTION, message)
+            response = self.llm_client.stream_invoke_to_string(self.system_prompt, message)
 
             # 处理响应
             processed_response = self.process_output(response)
@@ -213,7 +224,7 @@ def run(self, input_data: Any, **kwargs) -> Dict[str, str]:
             logger.exception(f"反思生成搜索查询失败: {str(e)}")
             raise e
 
-    def process_output(self, output: str) -> Dict[str, str]:
+    def process_output(self, output: str) -> Dict[str, Any]:
         """
         处理LLM输出，提取搜索查询和推理
 
@@ -258,22 +269,28 @@ def process_output(self, output: str) -> Dict[str, str]:
             # 验证和清理结果
             search_query = result.get("search_query", "")
             reasoning = result.get("reasoning", "")
+            search_tool = result.get("search_tool") or "basic_search_news"
+            start_date = result.get("start_date")
+            end_date = result.get("end_date")
 
             if not search_query:
                 logger.warning("未找到搜索查询，使用默认查询")
                 return self._get_default_reflection_query()
 
             return {
                 "search_query": search_query,
-                "reasoning": reasoning
+                "search_tool": search_tool,
+                "reasoning": reasoning,
+                "start_date": start_date,
+                "end_date": end_date,
             }
 
         except Exception as e:
             logger.exception(f"处理输出失败: {str(e)}")
             # 返回默认查询
             return self._get_default_reflection_query()
 
-    def _get_default_reflection_query(self) -> Dict[str, str]:
+    def _get_default_reflection_query(self) -> Dict[str, Any]:
         """
         获取默认反思搜索查询
 
@@ -282,5 +299,8 @@ def _get_default_reflection_query(self) -> Dict[str, str]:
         """
         return {
             "search_query": "深度研究补充信息",
-            "reasoning": "由于解析失败，使用默认反思搜索查询"
+            "search_tool": "basic_search_news",
+            "reasoning": "由于解析失败，使用默认反思搜索查询",
+            "start_date": None,
+            "end_date": None,
         }

QueryEngine/prompts/__init__.py

@@ -15,7 +15,9 @@
     output_schema_first_summary,
     output_schema_reflection,
     output_schema_reflection_summary,
-    input_schema_report_formatting
+    input_schema_report_formatting,
+    build_first_search_prompt,
+    build_reflection_prompt,
 )
 
 __all__ = [
@@ -30,5 +32,7 @@
     "output_schema_first_summary", 
     "output_schema_reflection",
     "output_schema_reflection_summary",
-    "input_schema_report_formatting"
+    "input_schema_report_formatting",
+    "build_first_search_prompt",
+    "build_reflection_prompt",
 ]