damon/links
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

修复A2C工单蜘蛛

Browse Source
This commit is contained in:
root
2026-07-01 01:26:27 +08:00
parent 9f2e904fab
commit 54c13c54ef
22 changed files with 1614 additions and 272 deletions
Download Patch File Download Diff File Expand all files Collapse all files
patches/puppeteer-api/lib/browser-factory.js
+4
View File
@@ -26,12 +26,16 @@ function normalizeAntiBot(antiBot) {
profile: antiBot.profile === 'real' ? 'real' : 'standard',
turnstile: antiBot.turnstile !== false,
solverFallback: antiBot.solverFallback !== false,
cfClearanceRequired: antiBot.cfClearanceRequired !== false,
sessionKey: antiBot.sessionKey || '',
challengeTimeoutMs: antiBot.challengeTimeoutMs || 60000,
postCfReadyUrlContains: antiBot.postCfReadyUrlContains || '',
postCfReadySelector: antiBot.postCfReadySelector || '',
postCfReadyTimeoutMs: antiBot.postCfReadyTimeoutMs || 0,
postCfSettleMs: antiBot.postCfSettleMs || 0,
postCfSpaWaitMs: antiBot.postCfSpaWaitMs || 0,
businessHostHint: antiBot.businessHostHint || '',
landingUrlHint: antiBot.landingUrlHint || '',
};
}
patches/puppeteer-api/lib/cf-detector.js
+31 -12
View File
@@ -1,5 +1,8 @@
/**
* Cloudflare / Turnstile 挑战页检测
*
* antiBot.cfClearanceRequired === false（A2C 业务域）：仅真实 CF 中间页算 blocked，
* 不因页面引用 turnstile.js 或缺少 cf_clearance cookie 误判。
*/
/**
@@ -17,6 +20,14 @@ function urlIndicatesCfChallenge(url) {
|| url.includes('/cdn-cgi/challenge');
}
/**
* 是否要求 cf_clearance 才视为过盾完成（A2C user.a2c.chat 为 false）
* @param {object|null|undefined} antiBot
*/
function isCfClearanceRequired(antiBot) {
return antiBot?.cfClearanceRequired !== false;
}
/**
* @typedef {Object} CloudflareState
* @property {boolean} blocked 是否处于 CF 挑战中
@@ -29,9 +40,10 @@ function urlIndicatesCfChallenge(url) {
/**
* 检测页面是否被 Cloudflare 拦截
* @param {import('puppeteer').Page} page
* @param {object|null} [antiBot]
* @returns {Promise<CloudflareState>}
*/
async function detectCloudflare(page) {
async function detectCloudflare(page, antiBot = null) {
const url = page.url();
const title = await page.title().catch(() => '');
@@ -61,22 +73,28 @@ async function detectCloudflare(page) {
const urlChallenge = urlIndicatesCfChallenge(url);
const realChallenge = urlChallenge
|| domSignals.titleHasJustAMoment
|| domSignals.bodyHasJustAMoment
|| domSignals.hasChallengeRunning
|| title.includes('Just a moment');
const turnstileSignals = domSignals.hasTurnstileInput || domSignals.hasTurnstileWidget;
let challengeType = null;
if (domSignals.hasTurnstileInput || domSignals.hasTurnstileWidget || url.includes('__cf_chl')) {
if (turnstileSignals || url.includes('__cf_chl')) {
challengeType = 'turnstile';
} else if (urlChallenge || domSignals.titleHasJustAMoment || domSignals.bodyHasJustAMoment) {
} else if (realChallenge) {
challengeType = 'js_challenge';
}
const blocked = !hasCfClearance && (
urlChallenge
|| domSignals.titleHasJustAMoment
|| domSignals.bodyHasJustAMoment
|| domSignals.hasTurnstileInput
|| domSignals.hasTurnstileWidget
|| domSignals.hasChallengeRunning
|| title.includes('Just a moment')
);
const requireClearance = isCfClearanceRequired(antiBot);
let blocked;
if (requireClearance) {
blocked = !hasCfClearance && (realChallenge || turnstileSignals);
} else {
blocked = realChallenge;
}
return {
blocked,
@@ -110,4 +128,5 @@ module.exports = {
detectCloudflare,
isTurnstileSolved,
urlIndicatesCfChallenge,
isCfClearanceRequired,
};
patches/puppeteer-api/lib/cf-handler.js
+17 -10
View File
@@ -5,7 +5,7 @@
* 否则清除可能过期的 cf_clearance 并重新触发 Turnstile（Whatshub 短链 → work-order-sharing 场景）。
*/
const { detectCloudflare } = require('./cf-detector');
const { waitForTurnstile, extractTurnstileSitekey, injectTurnstileToken } = require('./turnstile-handler');
const { waitForTurnstile, extractTurnstileSitekey, injectTurnstileToken, waitForTurnstileSurface } = require('./turnstile-handler');
const { solveTurnstile, isCaptchaConfigured } = require('./captcha-solver');
const { isSessionLikelyValid } = require('./session-store');
@@ -83,14 +83,14 @@ async function tryPostCfRecoverViaReload(page, waitForUrlSettled, timeoutMs, ant
* @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
* @param {number} timeoutMs
*/
async function invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs) {
async function invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs, antiBot = null) {
console.log('[CF] 清除 cf_clearance 并 reload，重新触发 Turnstile');
await clearCfClearanceCookies(page);
await page.reload({ waitUntil: 'domcontentloaded', timeout: Math.min(timeoutMs, 45000) });
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
return detectCloudflare(page);
return detectCloudflare(page, antiBot);
}
/**
@@ -146,7 +146,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
const started = Date.now();
const timeoutMs = antiBot.challengeTimeoutMs || 60000;
let cfState = await detectCloudflare(page);
let cfState = await detectCloudflare(page, antiBot);
if (!cfState.blocked && cfState.hasCfClearance) {
const fast = await trySessionReuseFastPath(
@@ -155,7 +155,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
if (fast) {
return fast;
}
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs);
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs, antiBot);
}
if (!cfState.blocked) {
@@ -188,7 +188,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
}
if (cfState.hasCfClearance) {
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs);
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs, antiBot);
} else {
return {
success: true,
@@ -209,7 +209,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
cfState = await detectCloudflare(page, antiBot);
if (!cfState.blocked) {
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
@@ -237,7 +237,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
cfState = await detectCloudflare(page, antiBot);
if (!cfState.blocked) {
return {
success: true,
@@ -254,8 +254,15 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
if (antiBot.solverFallback !== false && isCaptchaConfigured()) {
try {
const sitekey = await extractTurnstileSitekey(page);
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
let sitekey = await waitForTurnstileSurface(page, Math.min(timeoutMs, 25000));
if (!sitekey) {
sitekey = await extractTurnstileSitekey(page);
}
const pageUrl = page.url();
console.log(`[CF] Captcha API 求解 pageUrl=${pageUrl} sitekey=${sitekey ? 'ok' : 'missing'}`);
const { token, provider } = await solveTurnstile({ pageUrl, sitekey });
console.log(`[CF] Captcha API (${provider}) 返回 token，注入页面`);
await injectTurnstileToken(page, token);
@@ -267,7 +274,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
cfState = await detectCloudflare(page, antiBot);
if (!cfState.blocked) {
return {
success: true,
patches/puppeteer-api/lib/turnstile-handler.js
+112 -21
View File
@@ -1,5 +1,5 @@
/**
* Turnstile 内置求解：等待 widget 自动完成 + token 轮询
* 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');
@@ -10,6 +10,115 @@ const { detectCloudflare, isTurnstileSolved, urlIndicatesCfChallenge } = require
* @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
@@ -49,26 +158,6 @@ async function waitForTurnstile(page, options = {}) {
return { success: false, stage: 'timeout', elapsedMs: Date.now() - started };
}
/**
* 从页面提取 Turnstile sitekey
* @param {import('puppeteer').Page} page
* @returns {Promise<string|null>}
*/
async function extractTurnstileSitekey(page) {
return page.evaluate(() => {
const widget = document.querySelector('.cf-turnstile[data-sitekey], [data-sitekey]');
if (widget) {
return widget.getAttribute('data-sitekey');
}
const iframe = document.querySelector('iframe[src*="challenges.cloudflare.com"]');
if (iframe && iframe.src) {
const match = iframe.src.match(/[?&]sitekey=([^&]+)/);
if (match) return decodeURIComponent(match[1]);
}
return null;
}).catch(() => null);
}
/**
* 将 Captcha API 返回的 token 注入页面
* @param {import('puppeteer').Page} page
@@ -91,6 +180,8 @@ async function injectTurnstileToken(page, token) {
module.exports = {
waitForTurnstile,
waitForTurnstileSurface,
extractTurnstileSitekey,
extractSitekeyFromFrame,
injectTurnstileToken,
};