如何使用代码解释器工具

将代码解释器工具与代理配合使用

可以使用本文顶部列出的代码示例或 Azure AI Foundry 门户以编程方式将代码解释器工具添加到代理。 如果要使用门户：

1. 在您的代理屏幕中，向下滚动右侧的设置窗格至操作。 然后选择“添加”。

   

2. 选择“代码解释器”，然后按照提示添加该工具。

   

3. 可以选择上传文件，以供代理从数据集读取和解释信息、生成代码，以及使用数据创建图形和图表。

   

初始化

代码首先设置必要的导入并初始化 AI 项目客户端：

import os
from azure.ai.projects import AIProjectClient
from azure.identity import DefaultAzureCredential
from azure.ai.agents.models import CodeInterpreterTool


# Create an Azure AI Client from an endpoint, copied from your Azure AI Foundry project.
# You need to login to Azure subscription via Azure CLI and set the environment variables
project_endpoint = os.environ["PROJECT_ENDPOINT"]  # Ensure the PROJECT_ENDPOINT environment variable is set

# Create an AIProjectClient instance
project_client = AIProjectClient(
    endpoint=project_endpoint,
    credential=DefaultAzureCredential(),  # Use Azure Default Credential for authentication
)

文件上传

此示例上传数据文件进行分析：

file = project_client.agents.upload_file_and_poll(
    file_path="nifty_500_quarterly_results.csv", 
    purpose=FilePurpose.AGENTS
)

代码解释器设置

代码解释器工具使用上传的文件进行初始化：

code_interpreter = CodeInterpreterTool(file_ids=[file.id])

创建代理

使用代码解释器功能创建代理：

agent = project_client.agents.create_agent(
    model=os.environ["MODEL_DEPLOYMENT_NAME"],
    name="my-agent",
    instructions="You are helpful agent",
    tools=code_interpreter.definitions,
    tool_resources=code_interpreter.resources,
)

线程管理

该代码创建会话线程和初始消息：

thread = project_client.agents.threads.create()
message = project_client.agents.messages.create(
    thread_id=thread.id,
    role=MessageRole.USER,
    content="Could you please create bar chart in TRANSPORTATION sector for the operating profit from the uploaded csv file and provide file to me?",
)

消息处理

创建运行以处理消息并执行代码：

run = project_client.agents.runs.create_and_process(thread_id=thread.id, agent_id=agent.id)

文件处理

代码处理输出文件和注释：

messages = project_client.agents.messages.list(thread_id=thread.id)

# Save generated image files
for image_content in messages.image_contents:
    file_id = image_content.image_file.file_id
    file_name = f"{file_id}_image_file.png"
    project_client.agents.save_file(file_id=file_id, file_name=file_name)

# Process file path annotations
for file_path_annotation in messages.file_path_annotations:
    print(f"File Paths:")
    print(f"Type: {file_path_annotation.type}")
    print(f"Text: {file_path_annotation.text}")
    print(f"File ID: {file_path_annotation.file_path.file_id}")

清理行动

完成交互后，代码会正确清理资源：

project_client.agents.delete_file(file.id)
project_client.agents.delete_agent(agent.id)

这可确保适当的资源管理并防止不必要的资源消耗。

创建客户端和代理

首先，使用 appsettings.json 设置配置，创建一个 PersistentAgentsClient，然后使用启用代码解释器工具创建 PersistentAgent。

using Azure;
using Azure.AI.Agents.Persistent;
using Azure.Identity;
using Microsoft.Extensions.Configuration;
using System.Diagnostics;

IConfigurationRoot configuration = new ConfigurationBuilder()
    .SetBasePath(AppContext.BaseDirectory)
    .AddJsonFile("appsettings.json", optional: false, reloadOnChange: true)
    .Build();

var projectEndpoint = configuration["ProjectEndpoint"];
var modelDeploymentName = configuration["ModelDeploymentName"];

PersistentAgentsClient client = new(projectEndpoint, new DefaultAzureCredential());

PersistentAgent agent = client.Administration.CreateAgent(
    model: modelDeploymentName,
    name: "My Friendly Test Agent",
    instructions: "You politely help with math questions. Use the code interpreter tool when asked to visualize numbers.",
    tools: [new CodeInterpreterToolDefinition()]
);

创建线程并添加消息

接下来，为对话创建一个 PersistentAgentThread 并添加初始用户消息。

PersistentAgentThread thread = client.Threads.CreateThread();

client.Messages.CreateMessage(
    thread.Id,
    MessageRole.User,
    "Hi, Agent! Draw a graph for a line with a slope of 4 and y-intercept of 9.");

创建和监视运行

然后，为线程和代理创建一个 ThreadRun 。 轮询运行的状态，直到它完成或者它需要你采取措施。

ThreadRun run = client.Runs.CreateRun(
    thread.Id,
    agent.Id,
    additionalInstructions: "Please address the user as Jane Doe. The user has a premium account.");

do
{
    Thread.Sleep(TimeSpan.FromMilliseconds(500));
    run = client.Runs.GetRun(thread.Id, run.Id);
}
while (run.Status == RunStatus.Queued
    || run.Status == RunStatus.InProgress
    || run.Status == RunStatus.RequiresAction);

处理结果并处理文件

运行完成后，从线程检索所有消息。 循环访问消息以显示文本内容，并通过在本地保存和打开它们来处理任何生成的图像文件。

Pageable<PersistentThreadMessage> messages = client.Messages.GetMessages(
    threadId: thread.Id,
    order: ListSortOrder.Ascending);

foreach (PersistentThreadMessage threadMessage in messages)
{
    foreach (MessageContent content in threadMessage.ContentItems)
    {
        switch (content)
        {
            case MessageTextContent textItem:
                Console.WriteLine($"[{threadMessage.Role}]: {textItem.Text}");
                break;
            case MessageImageFileContent imageFileContent:
                Console.WriteLine($"[{threadMessage.Role}]: Image content file ID = {imageFileContent.FileId}");
                BinaryData imageContent = client.Files.GetFileContent(imageFileContent.FileId);
                string tempFilePath = Path.Combine(AppContext.BaseDirectory, $"{Guid.NewGuid()}.png");
                File.WriteAllBytes(tempFilePath, imageContent.ToArray());
                client.Files.DeleteFile(imageFileContent.FileId);

                ProcessStartInfo psi = new()
                {
                    FileName = tempFilePath,
                    UseShellExecute = true
                };
                Process.Start(psi);
                break;
        }
    }
}

清理资源

最后，删除线程和代理以清理此示例中创建的资源。

    client.Threads.DeleteThread(threadId: thread.Id);
    client.Administration.DeleteAgent(agentId: agent.Id);

创建项目客户端

若要使用代码解释器，首先需要创建一个项目客户端，其中包含 AI 项目的终结点，并将用于对 API 调用进行身份验证。

const { AgentsClient, isOutputOfType, ToolUtility } = require("@azure/ai-agents");
const { delay } = require("@azure/core-util");
const { DefaultAzureCredential } = require("@azure/identity");
const fs = require("fs");
const path = require("node:path");
require("dotenv/config");

const projectEndpoint = process.env["PROJECT_ENDPOINT"];

// Create an Azure AI Client
const client = new AgentsClient(projectEndpoint, new DefaultAzureCredential());

上传文件

可以上传文件，然后由代理或消息引用。 上传后，可将其添加到工具实用工具进行引用。

// Upload file and wait for it to be processed
const filePath = "./data/nifty500QuarterlyResults.csv";
const localFileStream = fs.createReadStream(filePath);
const localFile = await client.files.upload(localFileStream, "assistants", {
  fileName: "localFile",
});

console.log(`Uploaded local file, file ID : ${localFile.id}`);

使用代码解释器工具创建代理

// Create code interpreter tool
const codeInterpreterTool = ToolUtility.createCodeInterpreterTool([localFile.id]);

// Notice that CodeInterpreter must be enabled in the agent creation, otherwise the agent will not be able to see the file attachment
const agent = await client.createAgent("gpt-4o", {
  name: "my-agent",
  instructions: "You are a helpful agent",
  tools: [codeInterpreterTool.definition],
  toolResources: codeInterpreterTool.resources,
});
console.log(`Created agent, agent ID: ${agent.id}`);

创建线程、消息并获取代理响应

// Create a thread
const thread = await client.threads.create();
console.log(`Created thread, thread ID: ${thread.id}`);

// Create a message
const message = await client.messages.create(
  thread.id,
  "user",
  "Could you please create a bar chart in the TRANSPORTATION sector for the operating profit from the uploaded CSV file and provide the file to me?",
  {
    attachments: [
      {
        fileId: localFile.id,
        tools: [codeInterpreterTool.definition],
      },
    ],
  },
);

console.log(`Created message, message ID: ${message.id}`);

// Create and execute a run
let run = await client.runs.create(thread.id, agent.id);
while (run.status === "queued" || run.status === "in_progress") {
  await delay(1000);
  run = await client.runs.get(thread.id, run.id);
}
if (run.status === "failed") {
  // Check if you got "Rate limit is exceeded.", then you want to get more quota
  console.log(`Run failed: ${run.lastError}`);
}
console.log(`Run finished with status: ${run.status}`);

// Delete the original file from the agent to free up space (note: this does not delete your version of the file)
await client.files.delete(localFile.id);
console.log(`Deleted file, file ID: ${localFile.id}`);

// Print the messages from the agent
const messagesIterator = client.messages.list(thread.id);
const allMessages = [];
for await (const m of messagesIterator) {
  allMessages.push(m);
}
console.log("Messages:", allMessages);

// Get most recent message from the assistant
const assistantMessage = allMessages.find((msg) => msg.role === "assistant");
if (assistantMessage) {
  const textContent = assistantMessage.content.find((content) => isOutputOfType(content, "text"));
  if (textContent) {
    console.log(`Last message: ${textContent.text.value}`);
  }
}

// Save the newly created file
console.log(`Saving new files...`);
const imageFile = allMessages[0].content[0].imageFile;
console.log(`Image file ID : ${imageFile.fileId}`);
const imageFileName = path.resolve(
  "./data/" + (await client.files.get(imageFile.fileId)).filename + "ImageFile.png",
);

const fileContent = await (await client.files.getContent(imageFile.fileId).asNodeStream()).body;
if (fileContent) {
  const chunks = [];
  for await (const chunk of fileContent) {
    chunks.push(Buffer.isBuffer(chunk) ? chunk : Buffer.from(chunk));
  }
  const buffer = Buffer.concat(chunks);
  fs.writeFileSync(imageFileName, buffer);
} else {
  console.error("Failed to retrieve file content: fileContent is undefined");
}
console.log(`Saved image file to: ${imageFileName}`);

// Iterate through messages and print details for each annotation
console.log(`Message Details:`);
allMessages.forEach((m) => {
  console.log(`File Paths:`);
  console.log(`Type: ${m.content[0].type}`);
  if (isOutputOfType(m.content[0], "text")) {
    const textContent = m.content[0];
    console.log(`Text: ${textContent.text.value}`);
  }
  console.log(`File ID: ${m.id}`);
});

// Delete the agent once done
await client.deleteAgent(agent.id);
console.log(`Deleted agent, agent ID: ${agent.id}`);

请按照 REST API 快速入门 为环境变量 AGENT_TOKEN、AZURE_AI_FOUNDRY_PROJECT_ENDPOINT 和 API_VERSION 设置正确的值。

上传文件

curl --request POST \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/files?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN" \
  -F purpose="assistants" \
  -F file="@c:\\path_to_file\\file.csv"

使用代码解释器工具创建代理

curl --request POST \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/assistants?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "instructions": "You are an AI assistant that can write code to help answer math questions.",
    "tools": [
      { "type": "code_interpreter" }
    ],
    "model": "gpt-4o-mini",
    "tool_resources"{
      "code interpreter": {
          "file_ids": ["assistant-1234"]
      }
    }
  }'

创建线程

curl --request POST \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/threads?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN" \
  -H "Content-Type: application/json" \
  -d ''

将用户问题添加到线程

curl --request POST \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/threads/thread_abc123/messages?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
      "role": "user",
      "content": "I need to solve the equation `3x + 11 = 14`. Can you help me?"
    }'

运行线程

curl --request POST \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/threads/thread_abc123/runs?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "assistant_id": "asst_abc123",
  }'

获取运行状态

curl --request GET \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/threads/thread_abc123/runs/run_abc123?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN"

检索代理响应

curl --request GET \
  --url $AZURE_AI_FOUNDRY_PROJECT_ENDPOINT/threads/thread_abc123/messages?api-version=$API_VERSION \
  -H "Authorization: Bearer $AGENT_TOKEN"