From 620a6f1781b7ef9a6a5402ca7dc0838dd5498154 Mon Sep 17 00:00:00 2001
From: lonpatovaadelina <lonpatovaadelina@mail.ru>
Date: Wed, 27 May 2026 14:33:53 +0000
Subject: [PATCH] =?UTF-8?q?Stream-=D1=80=D0=B5=D0=B6=D0=B8=D0=BC=20AI-?=
 =?UTF-8?q?=D0=B0=D0=B3=D0=B5=D0=BD=D1=82=D0=B0:=20agent.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../agent.py                                    | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/solutions/699cc158d6d3a5544a3ed35b_Stream-режим_AI-агента/agent.py b/solutions/699cc158d6d3a5544a3ed35b_Stream-режим_AI-агента/agent.py
index d8d2d33..a5a0596 100644
--- a/solutions/699cc158d6d3a5544a3ed35b_Stream-режим_AI-агента/agent.py
+++ b/solutions/699cc158d6d3a5544a3ed35b_Stream-режим_AI-агента/agent.py
@@ -5,7 +5,7 @@ from langchain_ollama import ChatOllama
 from nomic_embed_text import NomicEmbedText
 from langchain.tools import BaseTool
 from langchain.schema import HumanMessage, AIMessage, SystemMessage
-from langchain.agents import ToolExecutor, AgentExecutor, create_openai_tools_agent
+from langchain.agents import AgentExecutor, initialize_agent, load_tools
 from langchain.prompts import ChatPromptTemplate
 
 # ---------- Настройки LLM и эмбеддингов ----------
@@ -29,7 +29,6 @@ class DummyTool(BaseTool):
 
 tool = DummyTool()
 tools = [tool]
-tools_dict = {t.name: t for t in tools}
 
 # ---------- Создание агента ----------
 prompt_template = ChatPromptTemplate.from_messages(
@@ -39,8 +38,13 @@ prompt_template = ChatPromptTemplate.from_messages(
     ]
 )
 
-agent = create_openai_tools_agent(llm, tools, prompt=prompt_template)
-executor = AgentExecutor(agent=agent, tools=tools, verbose=False)
+agent_executor = initialize_agent(
+    tools,
+    llm,
+    agent="openai-tools",
+    verbose=False,
+    prompt=prompt_template,
+)
 
 # ---------- Функции форматирования ----------
 def format_message(message) -> str:
@@ -66,8 +70,9 @@ def run_agent(user_input: str):
     global step
     step = 1
     # Инициализируем состояние с пользовательским сообщением
-    init_state = {"messages": [HumanMessage(content=user_input)]}
-    stream = executor.stream(init_state, stream_mode=["messages", "updates"])
+    init_state = {"input": user_input}
+    # В LangChain 0.2 AgentExecutor имеет метод stream()
+    stream = agent_executor.stream(init_state, stream_mode=["messages", "updates"])
 
     for chunk_type, chunk_data in stream:
         if chunk_type == "messages":