Targeting an API assistant in PHP

API
,
1

Hello! I’m developing a chatbot based on an API assistant previously created via the OpenAI platform. I only use PHP, Javascript, html and css. No Python. I can communicate with an instance of chatGPT, but not with “my” API Assistant. Note that my assistant uses gpt-4-turbo. How do I target it?
Here’s my PHP code:

<?php
require 'config.php';

header('Content-Type: application/json');

function callAPI($model, $messages) {
    $url = 'https://api.openai.com/v1/chat/completions';
    $data = [
        'model' => $model,
        'messages' => $messages
    ];
    $headers = [
        "Content-Type: application/json",
        "Authorization: Bearer " . API_KEY
    ];

    $ch = curl_init($url);
    curl_setopt($ch, CURLOPT_CUSTOMREQUEST, "POST");
    curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($data));
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);

    $response = curl_exec($ch);
    $httpcode = curl_getinfo($ch, CURLINFO_HTTP_CODE);

    if (!$response || $httpcode != 200) {
        http_response_code($httpcode ?: 500);
        echo json_encode(['error' => 'API Error: ' . curl_error($ch)]);
        curl_close($ch);
        exit;
    }
    curl_close($ch);
    return $response;
}

$data = json_decode(file_get_contents('php://input'), true);
if (!isset($data['prompt'])) {
    http_response_code(400);
    echo json_encode(['error' => 'Missing prompt']);
    exit;
}

$prompt = htmlspecialchars($data['prompt'], ENT_QUOTES, 'UTF-8');

$response = callAPI('gpt-4-turbo', [
    ["role" => "system", "content" => "Vous êtes un professeur d'histoire."],
    ["role" => "user", "content" => $prompt]
]);
echo $response;
?>
2

Here’s an overview of Assistants…

https://platform.openai.com/docs/assistants/overview

You won’t be able to directly connect with a ChatGPT Custom GPT via the Assistants API, though. You need to recreate it via Assistants.

3

Potential Issues

  1. Calling the chat API here:

  • 'https://api.openai.com/v1/chat/completions' :x:

  • Instead point this to assistant API https://api.openai.com/v1/assistants :white_check_mark:

  • Start Here Assistant Creation or Skip to Thread : Link

  1. Remember to include the assistant ID, not present in your current call :
  • GET https://api.openai.com/v1/assistants/{assistant_id}
  1. Reference Only Code by GPT:
<?php
require 'config.php';

header('Content-Type: application/json');

function callAPI($endpoint, $data) {
    $url = 'https://api.openai.com/v1/' . $endpoint;
    $headers = [
        "Content-Type: application/json",
        "Authorization: Bearer " . API_KEY
    ];

    $ch = curl_init($url);
    curl_setopt($ch, CURLOPT_CUSTOMREQUEST, "POST");
    curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($data));
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);

    $response = curl_exec($ch);
    $httpcode = curl_getinfo($ch, CURLINFO_HTTP_CODE);

    if (!$response || $httpcode != 200) {
        http_response_code($httpcode ?: 500);
        echo json_encode(['error' => 'API Error: ' . curl_error($ch)]);
        curl_close($ch);
        exit;
    }
    curl_close($ch);
    return json_decode($response, true);
}

$data = json_decode(file_get_contents('php://input'), true);
if (!isset($data['prompt'])) {
    http_response_code(400);
    echo json_encode(['error' => 'Missing prompt']);
    exit;
}

$prompt = htmlspecialchars($data['prompt'], ENT_QUOTES, 'UTF-8');

// Create a thread
$threadData = ['model' => 'gpt-4-turbo', 'assistant_id' => 'your-assistant-id'];
$threadResponse = callAPI('threads', $threadData);
$threadId = $threadResponse['data']['id'];

// Add message to the thread
$messageData = [
    'thread_id' => $threadId,
    'messages' => [
        ["role" => "system", "content" => "Vous êtes un professeur d'histoire."],
        ["role" => "user", "content" => $prompt]
    ]
];
$messageResponse = callAPI('threads/' . $threadId . '/messages', $messageData);

// Run the thread
$runData = ['thread_id' => $threadId];
$runResponse = callAPI('threads/' . $threadId . '/run', $runData);

// Retrieve messages from the thread
$retrieveData = ['thread_id' => $threadId];
$retrieveResponse = callAPI('threads/' . $threadId . '/messages', $retrieveData);

echo json_encode($retrieveResponse);
?>
1 Like
4

I knew this. Maybe I did not express myself correctly…
Thank you anyway !

5

Mmm, a step forward. But I still have an error in my console :

Failed to load resource: the server responded with a status of 400 ()

This is certainly because I don’t know where should I include the assistant ID ? I mean where to add the following line of code

GET https://api.openai.com/v1/assistants/asst_b2bZI9WQu1l6c5sDCdvye7z0

I put the assistant ID in this bloc of code too :

// Create a thread
$threadData = ['model' => 'gpt-4-turbo', 'assistant_id' => 'asst_b2bZI9WQu1l6c5sDCdvye7z0'];
$threadResponse = callAPI('threads', $threadData);
$threadId = $threadResponse['data']['id'];

Is this correct ?

I really appreciate a lot you help !
Thank you so much. :pray:

6

The documentation is very clear on how to’s !

Steps to undertake and validate in code:

  • create thread,
  • add messages,
  • run thread
  • retrieve run status,
  • on completion retrieve message.

Note: Streaming is enabled too recently in assistant API, handle it as per official documentation.

Best of luck…

7

Thank you ‘Engagepy’ !
I’ll try to do my best, but as I’m not developper, I need some luck :wink:

1 Like
8

Your articulation is good, just put GPT to the task. Best of luck !