chore: Update agent tool descriptions to support chart tool

src/agent/tool_chart.py

+from typing import Type
+from pydantic import BaseModel, Field
+from langchain_core.tools import BaseTool
+
+class ChartArgs(BaseModel):
+    name: str = Field(..., description="图表名称，用于显示在图表上方")
+    chart_type: str = Field(..., description="图表类型，如 line, bar, scatter, pie 等")
+    x: list = Field(..., description="x 轴数据，列表形式")
+    y: list = Field(..., description="y 轴数据，列表形式")
+    x_label: str = Field(..., description="x 轴标签")
+    y_label: str = Field(..., description="y 轴标签")
+
+class Chart(BaseTool):
+    name = "chart"
+    description = "组装生成图表的中间数据"
+    args_schema: Type[BaseModel] = ChartArgs
+
+    def _run(
+        self, name: str, x: list, y: list, x_label: str, y_label: str
+    ) -> str:
+        """Use the tool."""
+        result = {
+            "name": name,
+            "x": x,
+            "y": y,
+            "x_label": x_label,
+            "y_label": y_label
+        }
+        return result
+
+
+# 生成图表
+def chart_image(chart_data):
+    """
+    生成图表
+
+    Args:
+        chart_data: dict 图表数据 
+            {
+                "name": str, 图表名称
+                "chart_type": str, 图表类型，如 line, bar, scatter, pie 等
+                "x": list, x 轴数据，列表形式
+                "y": list, y 轴数据，列表形式
+                "x_label": str, x 轴标签
+                "y_label": str, y 轴标签
+            }
+
+    Returns:
+        PIL Image 图表图片
+    """
+    import matplotlib.pyplot as plt
+
+    plt.figure(figsize=(10, 6))
+    match chart_data["chart_type"]:
+        case "line":
+            plt.plot(chart_data["x"], chart_data["y"])
+        case "bar":
+            plt.bar(chart_data["x"], chart_data["y"])
+        case "scatter":
+            plt.scatter(chart_data["x"], chart_data["y"])
+        case "pie":
+            plt.pie(chart_data["y"], labels=chart_data["x"], autopct="%1.1f%%")
+        case _:
+            raise ValueError("Invalid chart type")
+        
+    plt.xlabel(chart_data["x_label"])
+    plt.ylabel(chart_data["y_label"])
+    plt.title(chart_data["name"])
+    # plt.show()
+
+    from io import BytesIO
+    
+    buf = BytesIO()
+    plt.savefig(buf, format="png")
+
+    from PIL import Image
+    
+    image = Image.open(buf)
+    # image.show()
+    return image
\ No newline at end of file

src/agent/tool_divisions.py

@@ -132,7 +132,7 @@ class AdministrativeDivisionArgs(BaseModel):
 
 class AdministrativeDivision(BaseTool):
     name = "administrative_division"
-    description = "根据输入补全行政区划信息，明确具体的省、市、县信息。比如输入县，补全所属省市，输入市则补全省级以及下辖所有县区"
+    description = "根据用户提问中涉及到的地区信息补全其行政区划信息，明确具体的省、市、县信息。比如输入县，补全所属省市，输入市则补全省级以及下辖所有县区"
     args_schema: Type[BaseModel] = AdministrativeDivisionArgs
 
     def _run(self, input_text: str) -> str:

src/config/prompts.py

@@ -37,7 +37,49 @@ Action:
     }}
     ```
 
-开始！始终以有效的单个操作的 JSON 对象回复。如有必要，请使用工具。如果合适，请直接回复。格式为 Action:```$JSON_BLOB```然后 Observation
+开始！始终以有效的单个操作的 JSON 对象回复。如有必要，请使用工具。如果你知道答案，请直接回复。格式为 Action:```$JSON_BLOB```然后 Observation
+"""
+
+PROMPT_AGENT_CAHRT_SYS = """请尽量帮助人类并准确回答问题。您可以使用以下工具：
+
+{tools}
+
+使用 JSON 对象指定工具，提供一个 action 键（工具名称）和一个 action_input 键（工具输入）, 以及 action_cache 键（必要时存储工具中间结果） 。
+
+有效的 "action" 值： "Final Answer" 或 {tool_names}
+
+每个 $JSON_BLOB 只提供一个操作，如下所示：
+    ``` 
+    {{
+    "action": $TOOL_NAME,
+    "action_input": $INPUT,
+    "action_cache": $CACHE
+    }}
+    ```
+
+按照以下格式：
+
+Question: 输入要回答的问题
+Thought: 考虑前后步骤
+Action:
+    ```
+    $JSON_BLOB
+    ```
+Observation: 操作结果
+...（重复 Thought/Action/Observation N 次）
+Thought: 我知道如何回复
+Action:
+    ```
+    {{
+    "action": "Final Answer",
+    "action_input": "最终回复给人类",
+    "action_cache": "中间结果缓存"
+    }}
+    ```
+开始！始终以有效的单个操作的 JSON 对象回复。如有必要，请使用工具。如果你知道答案，请直接回复。如果有生成图表的需求，请使用图表生成工具 {chart_tool}，并将结果存储到 $CACHE 中。
+格式为 Action:```$JSON_BLOB```然后 Observation
+
+
 """
 
 PROMPT_AGENT_HUMAN = """{input}\n\n{agent_scratchpad}\n (请注意，无论如何都要以 JSON 对象回复)"""
@@ -128,6 +170,8 @@ A: Laf 是一个云函数开发平台。
 检索词: """
 
 
+
+
 # 结合历史问答对话，生成新的提问，引导用户继续对话
 PROMPT_QA_EXTEND_QUESTION = """
 作为一个问答助手，你的任务是结合历史记录，生成三个新的问题，引导用户继续对话。生成的问题要求与对话内容相关且指向对象清晰明确，并与“原问题语言相同”。例如：

src/server/agent.py

-from typing import Any, List
+from typing import Any, List, Sequence, Union
 
 from langchain_core.prompts import PromptTemplate
 from langchain.agents import AgentExecutor, create_tool_calling_agent,create_structured_chat_agent
 from langchain.tools import BaseTool
 from langgraph.prebuilt import create_react_agent
 
+
+from langchain_core.language_models import BaseLanguageModel
+from langchain_core.prompts.chat import ChatPromptTemplate
+from langchain_core.runnables import Runnable, RunnablePassthrough
+from langchain_core.output_parsers.json import parse_json_markdown
+from langchain_core.agents import AgentAction, AgentFinish
+from langchain_core.exceptions import OutputParserException
+from langchain.tools.render import ToolsRenderer, render_text_description_and_args
+from langchain.agents.format_scratchpad import format_log_to_str
+from langchain.agents.agent import AgentOutputParser
+
+class ChartAgentOutputParser(AgentOutputParser):
+    """Parses tool invocations and final answers in JSON format.
+
+    Expects output to be in one of two formats.
+
+    If the output signals that an action should be taken,
+    should be in the below format. This will result in an AgentAction
+    being returned.
+
+    ```
+    {
+      "action": "search",
+      "action_input": "2+2",
+      "action_cache": ""
+    }
+    ```
+
+    If the output signals that a final answer should be given,
+    should be in the below format. This will result in an AgentFinish
+    being returned.
+
+    ```
+    {
+      "action": "Final Answer",
+      "action_input": "4",
+      "action_cache": ""
+    }
+    ```
+    """
+
+    def parse(self, text: str) -> Union[AgentAction, AgentFinish]:
+        try:
+            response = parse_json_markdown(text)
+            if isinstance(response, list):
+                # gpt turbo frequently ignores the directive to emit a single action
+                # logger.warning("Got multiple action responses: %s", response)
+                response = response[0]
+            if response["action"] == "Final Answer":
+                if "action_cache" in response:
+                    return AgentFinish({"output": response["action_input"],"cache":response["action_cache"]}, text)
+                else:
+                    return AgentFinish({"output": response["action_input"]}, text)
+            else:
+                return AgentAction(
+                    response["action"], response.get("action_input", {}), text
+                )
+        except Exception as e:
+            raise OutputParserException(f"Could not parse LLM output: {text}") from e
+
+    @property
+    def _type(self) -> str:
+        return "chart-agent"
+
+
+def create_chart_agent(
+    llm: BaseLanguageModel,
+    tools: Sequence[BaseTool],
+    prompt: ChatPromptTemplate,
+    chart_tool: str,
+    tools_renderer: ToolsRenderer = render_text_description_and_args,
+    *,
+    stop_sequence: Union[bool, List[str]] = True,
+) -> Runnable:
+    """Create an agent aimed at supporting chart tools with multiple inputs.
+    """
+    missing_vars = {"tools", "tool_names", "agent_scratchpad","chart_tool"}.difference(
+        prompt.input_variables + list(prompt.partial_variables)
+    )
+    if missing_vars:
+        raise ValueError(f"Prompt missing required variables: {missing_vars}")
+
+    prompt = prompt.partial(
+        tools=tools_renderer(list(tools)),
+        tool_names=", ".join([t.name for t in tools]),
+        chart_tool=chart_tool,
+    )
+    if stop_sequence:
+        stop = ["\nObservation"] if stop_sequence is True else stop_sequence
+        llm_with_stop = llm.bind(stop=stop)
+    else:
+        llm_with_stop = llm
+
+    agent = (
+        RunnablePassthrough.assign(
+            agent_scratchpad=lambda x: format_log_to_str(x["intermediate_steps"]),
+        )
+        | prompt
+        | llm_with_stop
+        | ChartAgentOutputParser()
+    )
+    return agent
+
 class Agent:
     def __init__(self, llm, tools: List[BaseTool],prompt: PromptTemplate = None, verbose: bool = False):
         self.llm = llm
@@ -15,7 +118,7 @@ class Agent:
             agent = create_react_agent(llm, tools,debug=verbose)
             self.agent_executor = agent
         else:
-            agent = create_structured_chat_agent(llm, tools, prompt)
+            agent = create_chart_agent(llm, tools, prompt)
             self.agent_executor = AgentExecutor(agent=agent,tools=tools,verbose=verbose)        
 
     def exec(self, prompt_args: dict = {}, stream: bool = False):

test/agent_test.py

@@ -15,8 +15,8 @@ from langchain.agents import AgentExecutor, create_tool_calling_agent,create_str
 
 from pydantic import BaseModel, Field
 
-from src.server.agent import Agent
-from src.config.prompts import PROMPT_AGENT_SYS, PROMPT_AGENT_HUMAN
+from src.server.agent import Agent, create_chart_agent
+from src.config.prompts import PROMPT_AGENT_SYS, PROMPT_AGENT_HUMAN, PROMPT_AGENT_CAHRT_SYS
 
 from src.agent.tool_divisions import AdministrativeDivision, CountryInfo
 
@@ -51,10 +51,10 @@ llm = ChatOpenAI(
 )
 
 # prompt = hub.pull("hwchase17/openai-functions-agent")
-input_variables=['agent_scratchpad', 'input', 'tool_names', 'tools'] 
+input_variables=['agent_scratchpad', 'input', 'tool_names', 'tools', "chart_tool"] 
 input_types={'chat_history': List[Union[langchain_core.messages.ai.AIMessage, langchain_core.messages.human.HumanMessage, langchain_core.messages.chat.ChatMessage, langchain_core.messages.system.SystemMessage, langchain_core.messages.function.FunctionMessage, langchain_core.messages.tool.ToolMessage]]} 
 messages=[
-  SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=['tool_names', 'tools'], template=PROMPT_AGENT_SYS)), 
+  SystemMessagePromptTemplate(prompt=PromptTemplate(input_variables=['tool_names', 'tools', "chart_tool"], template=PROMPT_AGENT_CAHRT_SYS)), 
   MessagesPlaceholder(variable_name='chat_history', optional=True), 
   HumanMessagePromptTemplate(prompt=PromptTemplate(input_variables=['agent_scratchpad', 'input'], template=PROMPT_AGENT_HUMAN))
 ]
@@ -68,6 +68,7 @@ prompt = ChatPromptTemplate(
 def test_add():
     tools = [Calc()]
     agent = Agent(llm=llm, tools=tools, prompt=prompt, verbose=True)
+    agent = create_chart_agent(llm, tools, prompt,chart_tool="chart")
 
     res = agent.exec(prompt_args={"input": "what is 1 + 1?"})
 
@@ -88,5 +89,20 @@ def test_agent_division():
     res = agent.exec(prompt_args={"input": "我想知道陇南市西和县和文县的降雨量谁的多"})
     print(res)
 
+def test_chart_tool():
+    from src.agent.tool_chart import chart_image
+    x = [1,2,3,4,5]
+    y = [1,4,9,16,25]
+    chart_data = {
+        "name": "test",
+        "chart_type": "bar",
+        "x": x,
+        "y": y,
+        "x_label": "x axis",
+        "y_label": "y axis"
+    }
+    chart_image(chart_data)
+
 if __name__ == "__main__":
-    test_agent_division()
\ No newline at end of file
+    # test_agent_division()
+    test_chart_tool()
\ No newline at end of file