Puppeteer を使用して PHP で Web スクレイピングを行う際の reCAPTCHA の解決方法

<?php

require_once 'vendor/autoload.php';
use Nesk\Puphpeteer\Puppeteer;
use Nesk\Rialto\Data\JsFunction;
use GuzzleHttp\Client;

$puppeteer = new Puppeteer;
$browser = $puppeteer->launch();

// TODO: 構成を設定する
$api_key = "YOUR_API_KEY"; // CapSolverのAPIキー
$site_key = "6Le-wvkSAAAAAPBMRTvw0Q4Muexq9bi0DJwx_mJ-"; // ターゲットサイトのサイトキー
$site_url = "https://www.google.com/recaptcha/api2/demo"; // ターゲットサイトのページURL

function capsolver()
{
    global $api_key, $site_key, $site_url;
    $client = new Client();
    $payload = [
        'clientKey' => $api_key,
        'task' => [
            'type' => 'ReCaptchaV2TaskProxyLess',
            'websiteKey' => $site_key,
            'websiteURL' => $site_url,
        ]
    ];

    try {
        $response = $client->post("https://api.capsolver.com/createTask", [
            'json' => $payload
        ]);
        $data = json_decode($response->getBody(), true);
        $task_id = $data['taskId'] ?? null;
        if (!$task_id) {
            echo "タスクの作成に失敗しました: " . json_encode($data) . PHP_EOL;
            return null;
        }
        echo "taskIdを取得しました: " . $task_id . PHP_EOL;

        while (true) {
            sleep(1);

            $getResultPayload = [
                'clientKey' => $api_key,
                'taskId' => $task_id
            ];
            $resp = $client->post("https://api.capsolver.com/getTaskResult", [


```php
<?php

function capsolver() {
    global $api_key, $site_key;

    try {
        $client = new \GuzzleHttp\Client();
        $resp = $client->post('https://api.capsolver.com/createTask', [
            'json' => [
                'clientKey' => $api_key,
                'task' => [
                    'type' => 'recaptchaV2',
                    'websiteURL' => 'https://www.google.com/recaptcha/api2/demo',
                    'websiteKey' => $site_key,
                    'userAgent' => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/117.0.0.0 Safari/537.36',
                    'proxy' => 'http://127.0.0.1:8118', // Replace with your own proxy
                    'proxyType' => 'socks5' // Replace with your proxy type
                ]
            ]
        ]);
        $data = json_decode($resp->getBody(), true);
        $taskId = $data['taskId'];

        while (true) {
            $resp = $client->post('https://api.capsolver.com/getTaskResult', [
                'json' => [
                    'clientKey' => $api_key,
                    'taskId' => $taskId
                ]
            ]);
            $data = json_decode($resp->getBody(), true);
            $status = $data['status'] ?? null;

            if ($status === "ready") {
                return $data['solution']['gRecaptchaResponse'];
            }
            if ($status === "failed" || isset($data['errorId'])) {
                echo "Solve failed! response: " . json_encode($data) . PHP_EOL;
                return null;
            }
        }
    } catch (\Exception $e) {
        echo "Error: " . $e->getMessage() . PHP_EOL;
        return null;
    }
}

function reqSite() {
    global $site_url, $browser;
    $token = capsolver();
    if ($token === null) {
        return;
    }
    echo $token . PHP_EOL;

    $page = $browser->newPage();
    $page->goto($site_url);
    $evaluate_script = <<<EOD
        document.getElementById("g-recaptcha-response").value="$token";
        onSuccess("$token");
    EOD;
    $page->evaluate(JsFunction::createWithBody($evaluate_script));
    $product_element = $page->querySelector('#recaptcha-demo-submit');
    if ($product_element instanceof ElementHandle) {
        $product_element->click();
    } else {
        echo 'Element not found.' . PHP_EOL;
    }
}

reqSite();