add tool call agent

ks6088ts · ks6088ts · commit ceded0201fd0 · 2025-08-02T11:27:18.000+09:00
diff --git a/docs/index.md b/docs/index.md
@@ -27,12 +27,18 @@ uv run python -m template_langgraph.tasks.run_kabuto_helpdesk_agent "KABUTOの
 
 # BasicWorkflowAgent
 uv run python -m template_langgraph.tasks.draw_basic_workflow_agent_mermaid_png "data/basic_workflow_agent.png"
-uv run python -m template_langgraph.tasks.run_basic_workflow_agent "KABUTOの起動時に、画面全体が紫色に点滅し、システムがフリーズします。"
-uv run python -m template_langgraph.tasks.run_basic_workflow_agent "私の名前はフグ田 サザエ。東京都世田谷区桜新町あさひが丘3丁目に住んでいる 24 歳の主婦です。夫のノリスケと子供のタラちゃんがいます。"
+uv run python -m template_langgraph.tasks.run_basic_workflow_agent
+# 私の名前はフグ田 サザエ。東京都世田谷区桜新町あさひが丘3丁目に住んでいる 24 歳の主婦です。夫のノリスケと子供のタラちゃんがいます
+# KABUTOの起動時に、画面全体が紫色に点滅し、システムがフリーズします。KABUTO のマニュアルから、関連する情報を取得したり過去のシステムのトラブルシュート事例が蓄積されたデータベースから、関連する情報を取得して質問に答えてください
+# 天狗のいたずら という現象について KABUTO のマニュアルから、関連する情報を取得したり過去のシステムのトラブルシュート事例が蓄積されたデータベースから、関連する情報を取得して質問に答えてください
 ```
 
 ## References
 
+### LangGraph
+
+- [Build a custom workflow](https://langchain-ai.github.io/langgraph/concepts/why-langgraph/)
+
 ### Sample Codes
 
 - [「現場で活用するためのAIエージェント実践入門」リポジトリ](https://github.com/masamasa59/genai-agent-advanced-book)
diff --git a/template_langgraph/agents/basic_workflow_agent/agent.py b/template_langgraph/agents/basic_workflow_agent/agent.py
@@ -1,13 +1,41 @@
-from langchain_core.messages import AIMessage
+import json
+
+from langchain_core.messages import AIMessage, ToolMessage
 from langgraph.graph import END, START, StateGraph
 
 from template_langgraph.agents.basic_workflow_agent.models import AgentInput, AgentOutput, AgentState, Profile
 from template_langgraph.llms.azure_openais import AzureOpenAiWrapper
 from template_langgraph.loggers import get_logger
+from template_langgraph.tools.elasticsearch_tool import search_elasticsearch
+from template_langgraph.tools.qdrants import search_qdrant
 
 logger = get_logger(__name__)
 
 
+class BasicToolNode:
+    """A node that runs the tools requested in the last AIMessage."""
+
+    def __init__(self, tools: list) -> None:
+        self.tools_by_name = {tool.name: tool for tool in tools}
+
+    def __call__(self, inputs: dict):
+        if messages := inputs.get("messages", []):
+            message = messages[-1]
+        else:
+            raise ValueError("No message found in input")
+        outputs = []
+        for tool_call in message.tool_calls:
+            tool_result = self.tools_by_name[tool_call["name"]].invoke(tool_call["args"])
+            outputs.append(
+                ToolMessage(
+                    content=json.dumps(tool_result.__str__(), ensure_ascii=False),
+                    name=tool_call["name"],
+                    tool_call_id=tool_call["id"],
+                )
+            )
+        return {"messages": outputs}
+
+
 class BasicWorkflowAgent:
     def __init__(self):
         self.llm = AzureOpenAiWrapper().chat_model
@@ -21,13 +49,34 @@ def create_graph(self):
         workflow.add_node("initialize", self.initialize)
         workflow.add_node("do_something", self.do_something)
         workflow.add_node("extract_profile", self.extract_profile)
+        workflow.add_node("chat_with_tools", self.chat_with_tools)
+        workflow.add_node(
+            "tools",
+            BasicToolNode(
+                tools=[
+                    search_qdrant,
+                    search_elasticsearch,
+                ]
+            ),
+        )
         workflow.add_node("finalize", self.finalize)
 
         # Create edges
         workflow.add_edge(START, "initialize")
         workflow.add_edge("initialize", "do_something")
         workflow.add_edge("do_something", "extract_profile")
-        workflow.add_edge("extract_profile", "finalize")
+        workflow.add_edge("extract_profile", "chat_with_tools")
+        workflow.add_conditional_edges(
+            "chat_with_tools",
+            self.route_tools,
+            # The following dictionary lets you tell the graph to interpret the condition's outputs as a specific node
+            # It defaults to the identity function, but if you
+            # want to use a node named something else apart from "tools",
+            # You can update the value of the dictionary to something else
+            # e.g., "tools": "my_tools"
+            {"tools": "tools", END: "finalize"},
+        )
+        workflow.add_edge("tools", "chat_with_tools")
         workflow.add_edge("finalize", END)
 
         # Compile the graph
@@ -66,6 +115,39 @@ def extract_profile(self, state: AgentState) -> AgentState:
         state["profile"] = profile
         return state
 
+    def chat_with_tools(self, state: AgentState) -> AgentState:
+        """Chat with tools using the state."""
+        logger.info(f"Chatting with tools using state: {state}")
+        llm_with_tools = self.llm.bind_tools(
+            tools=[
+                search_qdrant,
+                search_elasticsearch,
+            ],
+        )
+        return {
+            "messages": [
+                llm_with_tools.invoke(state["messages"]),
+            ]
+        }
+
+    def route_tools(
+        self,
+        state: AgentState,
+    ):
+        """
+        Use in the conditional_edge to route to the ToolNode if the last message
+        has tool calls. Otherwise, route to the end.
+        """
+        if isinstance(state, list):
+            ai_message = state[-1]
+        elif messages := state.get("messages", []):
+            ai_message = messages[-1]
+        else:
+            raise ValueError(f"No messages found in input state to tool_edge: {state}")
+        if hasattr(ai_message, "tool_calls") and len(ai_message.tool_calls) > 0:
+            return "tools"
+        return END
+
     def finalize(self, state: AgentState) -> AgentState:
         """Finalize the agent's work and prepare the output."""
         logger.info(f"Finalizing BasicWorkflowAgent with state: {state}")
diff --git a/template_langgraph/tasks/run_basic_workflow_agent.py b/template_langgraph/tasks/run_basic_workflow_agent.py
@@ -1,27 +1,33 @@
 import logging
-import sys
 
-from template_langgraph.agents.basic_workflow_agent.agent import AgentInput, BasicWorkflowAgent
+from template_langgraph.agents.basic_workflow_agent.agent import AgentState
+from template_langgraph.agents.basic_workflow_agent.agent import graph as basic_workflow_agent_graph
 from template_langgraph.loggers import get_logger
 
 logger = get_logger(__name__)
 logger.setLevel(logging.INFO)
 
-if __name__ == "__main__":
-    question = "「鬼灯」を実行すると、KABUTOが急に停止します。原因と対策を教えてください。"
-    if len(sys.argv) > 1:
-        # sys.argv[1] が最初の引数
-        question = sys.argv[1]
 
-    # Agentのインスタンス化
-    agent = BasicWorkflowAgent()
+def stream_graph_updates(
+    state: AgentState,
+) -> dict:
+    for event in basic_workflow_agent_graph.stream(input=state):
+        logger.info("-" * 20)
+        logger.info(f"Event: {event}")
+    return event
 
-    # AgentInputの作成
-    agent_input = AgentInput(
-        request=question,
-    )
 
-    # エージェントの実行
-    logger.info(f"Running BasicWorkflowAgent with input: {agent_input.model_dump_json(indent=2)}")
-    agent_output = agent.run_agent(input=agent_input)
-    logger.info(f"Agent output: {agent_output.model_dump_json(indent=2)}")
+if __name__ == "__main__":
+    user_input = input("User: ")
+    state = AgentState(
+        messages=[
+            {
+                "role": "user",
+                "content": user_input,
+            }
+        ],
+        profile=None,
+    )
+    last_event = stream_graph_updates(state)
+    for value in last_event.values():
+        logger.info(f"Final state: {value['messages'][-1].content}")  # noqa: E501
