links/patches/puppeteer-api/lib/turnstile-handler.js

/**
 * Turnstile 内置求解：等待 widget 自动完成 + token 轮询 + 多 frame sitekey 提取
 */
const { detectCloudflare, isTurnstileSolved, urlIndicatesCfChallenge } = require('./cf-detector');

/**
 * @typedef {Object} TurnstileResult
 * @property {boolean} success
 * @property {string} stage built_in|timeout|already_clear
 * @property {number} elapsedMs
 */

/**
 * 在单个 frame 内提取 sitekey
 * @param {import('puppeteer').Frame} frame
 * @returns {Promise<string|null>}
 */
async function extractSitekeyFromFrame(frame) {
    return frame.evaluate(() => {
        const widget = document.querySelector('.cf-turnstile[data-sitekey], [data-sitekey]');
        if (widget) {
            const key = widget.getAttribute('data-sitekey');
            if (key) {
                return key;
            }
        }

        const iframes = document.querySelectorAll('iframe[src*="challenges.cloudflare.com"], iframe[src*="challenge-platform"]');
        for (const iframe of iframes) {
            const src = iframe.getAttribute('src') || '';
            const match = src.match(/[?&]sitekey=([^&]+)/i);
            if (match) {
                return decodeURIComponent(match[1]);
            }
        }

        const scripts = document.querySelectorAll('script');
        for (const script of scripts) {
            const text = script.textContent || '';
            const match = text.match(/sitekey['"\s:=]+['"]([0-9x_a-zA-Z-]{10,})['"]/i)
                || text.match(/data-sitekey=['"]([0-9x_a-zA-Z-]{10,})['"]/i);
            if (match) {
                return match[1];
            }
        }

        return null;
    }).catch(() => null);
}

/**
 * 从主文档及所有子 frame 提取 Turnstile sitekey
 * @param {import('puppeteer').Page} page
 * @returns {Promise<string|null>}
 */
async function extractTurnstileSitekey(page) {
    const mainKey = await extractSitekeyFromFrame(page.mainFrame());
    if (mainKey) {
        return mainKey;
    }

    for (const frame of page.frames()) {
        if (frame === page.mainFrame()) {
            continue;
        }
        const frameKey = await extractSitekeyFromFrame(frame);
        if (frameKey) {
            return frameKey;
        }
    }

    return null;
}

/**
 * Captcha API 调用前：等待 Turnstile 挑战面出现或 URL 进入 CF 挑战态
 *
 * @param {import('puppeteer').Page} page
 * @param {number} [timeoutMs]
 * @returns {Promise<string|null>} 若已提取到 sitekey 则返回，否则 null
 */
async function waitForTurnstileSurface(page, timeoutMs = 20000) {
    const started = Date.now();
    const pollMs = 400;

    while (Date.now() - started < timeoutMs) {
        const sitekey = await extractTurnstileSitekey(page);
        if (sitekey) {
            console.log(`[CF] 已检测到 Turnstile sitekey (${sitekey.slice(0, 8)}...)`);
            return sitekey;
        }

        const url = page.url();
        if (urlIndicatesCfChallenge(url)) {
            await new Promise((r) => setTimeout(r, pollMs));
            continue;
        }

        const state = await detectCloudflare(page);
        if (!state.blocked && state.hasCfClearance) {
            return null;
        }

        const hasSurface = await page.evaluate(() => {
            return !!document.querySelector(
                '.cf-turnstile, [data-sitekey], [name="cf-turnstile-response"], '
                + 'iframe[src*="challenges.cloudflare.com"], iframe[src*="challenge-platform"]'
            );
        }).catch(() => false);

        if (hasSurface) {
            await new Promise((r) => setTimeout(r, pollMs));
            continue;
        }

        await new Promise((r) => setTimeout(r, pollMs));
    }

    return null;
}

/**
 * 等待 Turnstile 通过（内置点击 / 自动跳转）
 * @param {import('puppeteer').Page} page
 * @param {{ timeoutMs?: number, pollMs?: number, useBuiltInClick?: boolean }} [options]
 * @returns {Promise<TurnstileResult>}
 */
async function waitForTurnstile(page, options = {}) {
    const timeoutMs = options.timeoutMs ?? 60000;
    const pollMs = options.pollMs ?? 500;
    const started = Date.now();

    const initial = await detectCloudflare(page);
    const initialUrlChallenge = urlIndicatesCfChallenge(page.url());

    if ((!initial.blocked && !initialUrlChallenge) || initial.hasCfClearance) {
        if (initial.hasCfClearance && !initialUrlChallenge) {
            return { success: true, stage: 'already_clear', elapsedMs: Date.now() - started };
        }
        if (!initial.blocked && !initialUrlChallenge) {
            return { success: true, stage: 'already_clear', elapsedMs: Date.now() - started };
        }
    }

    while (Date.now() - started < timeoutMs) {
        if (await isTurnstileSolved(page) && !urlIndicatesCfChallenge(page.url())) {
            return { success: true, stage: 'built_in', elapsedMs: Date.now() - started };
        }

        const state = await detectCloudflare(page);
        if (!state.blocked && !urlIndicatesCfChallenge(page.url()) && state.hasCfClearance) {
            return { success: true, stage: 'built_in', elapsedMs: Date.now() - started };
        }

        await new Promise((r) => setTimeout(r, pollMs));
    }

    return { success: false, stage: 'timeout', elapsedMs: Date.now() - started };
}

/**
 * 将 Captcha API 返回的 token 注入页面
 * @param {import('puppeteer').Page} page
 * @param {string} token
 */
async function injectTurnstileToken(page, token) {
    await page.evaluate((turnstileToken) => {
        const input = document.querySelector('[name="cf-turnstile-response"]');
        if (input) {
            input.value = turnstileToken;
            input.dispatchEvent(new Event('input', { bubbles: true }));
            input.dispatchEvent(new Event('change', { bubbles: true }));
        }
        const form = document.querySelector('form');
        if (form) {
            form.submit();
        }
    }, token);
}

module.exports = {
    waitForTurnstile,
    waitForTurnstileSurface,
    extractTurnstileSitekey,
    extractSitekeyFromFrame,
    injectTurnstileToken,
};