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

修改whatshub工单爬虫

Browse Source
This commit is contained in:
root
2026-06-29 04:54:41 +08:00
parent 6130125427
commit 9f2e904fab
14 changed files with 1254 additions and 147 deletions
Download Patch File Download Diff File Expand all files Collapse all files
patches/puppeteer-api/lib/cf-handler.js
+194 -22
View File
@@ -1,5 +1,8 @@
/**
* Cloudflare 挑战统一处理：检测 → 会话快速通道 → 内置等待 → Captcha API 兜底
*
* 当 antiBot.postCfReadyUrlContains 已配置时，session_reuse 仅在业务 URL 已到达时才走快速通道；
* 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');
@@ -7,8 +10,135 @@ const { solveTurnstile, isCaptchaConfigured } = require('./captcha-solver');
const { isSessionLikelyValid } = require('./session-store');
/**
* 是否配置了 CF 后必须到达的业务 URL
* @param {object} antiBot
*/
function needsPostCfBusinessUrl(antiBot) {
return !!(antiBot?.postCfReadyUrlContains || '').trim();
}
/**
* 当前页面 URL 是否已包含业务路径片段
* @param {import('puppeteer').Page} page
* @param {{ turnstile?: boolean, solverFallback?: boolean, challengeTimeoutMs?: number }} antiBot
* @param {object} antiBot
*/
function isPostCfBusinessUrlReady(page, antiBot) {
const needle = (antiBot?.postCfReadyUrlContains || '').trim();
if (!needle) {
return true;
}
return page.url().includes(needle);
}
/**
* 清除 cf_clearance，避免「有 cookie 但 SPA 未触发跳转」的假过关
* @param {import('puppeteer').Page} page
*/
async function clearCfClearanceCookies(page) {
const cookies = await page.cookies();
for (const c of cookies) {
if (c.name !== 'cf_clearance') {
continue;
}
const host = (c.domain || '').replace(/^\./, '');
if (host) {
await page.deleteCookie({ name: c.name, url: `https://${host}/` }).catch(() => {});
}
await page.deleteCookie({
name: c.name,
domain: c.domain,
path: c.path || '/',
}).catch(() => {});
}
}
/**
* reload 后检查业务 URL 是否就绪
* @param {import('puppeteer').Page} page
* @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
* @param {number} timeoutMs
* @param {object} antiBot
*/
async function tryPostCfRecoverViaReload(page, waitForUrlSettled, timeoutMs, antiBot) {
if (!needsPostCfBusinessUrl(antiBot) || isPostCfBusinessUrlReady(page, antiBot)) {
return isPostCfBusinessUrlReady(page, antiBot);
}
console.log(`[CF] 业务 URL 未就绪 current=${page.url()}，尝试 reload`);
try {
await page.reload({ waitUntil: 'domcontentloaded', timeout: Math.min(timeoutMs, 45000) });
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
return isPostCfBusinessUrlReady(page, antiBot);
} catch (reloadErr) {
console.warn('[CF] reload 失败:', reloadErr.message);
return false;
}
}
/**
* 清除 cf_clearance 并 reload，使 Turnstile 重新出现
* @param {import('puppeteer').Page} page
* @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
* @param {number} timeoutMs
*/
async function invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs) {
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);
}
/**
* session_reuse 快速返回前校验：有 postCf 要求时必须已到业务页
* @param {import('puppeteer').Page} page
* @param {object} antiBot
* @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
* @param {number} timeoutMs
* @param {number} started
* @param {boolean} cfDetected
*/
async function trySessionReuseFastPath(page, antiBot, waitForUrlSettled, timeoutMs, started, cfDetected) {
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
if (!needsPostCfBusinessUrl(antiBot) || isPostCfBusinessUrlReady(page, antiBot)) {
return {
success: true,
code: null,
challengeType: null,
stage: 'session_reuse',
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected,
};
}
console.log(`[CF] cf_clearance/无挑战但业务页未就绪 url=${page.url()}，降级重试`);
if (await tryPostCfRecoverViaReload(page, waitForUrlSettled, timeoutMs, antiBot)) {
return {
success: true,
code: null,
challengeType: null,
stage: 'session_reuse_reload',
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected,
};
}
return null;
}
/**
* @param {import('puppeteer').Page} page
* @param {{ turnstile?: boolean, solverFallback?: boolean, challengeTimeoutMs?: number, postCfReadyUrlContains?: string }} antiBot
* @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
* @param {import('./session-store').StoredSession|null} [storedSession]
*/
@@ -19,36 +149,71 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
let cfState = await detectCloudflare(page);
if (!cfState.blocked && cfState.hasCfClearance) {
return {
success: true,
code: null,
challengeType: null,
stage: 'session_reuse',
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected: false,
};
const fast = await trySessionReuseFastPath(
page, antiBot, waitForUrlSettled, timeoutMs, started, false
);
if (fast) {
return fast;
}
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs);
}
if (!cfState.blocked) {
return {
success: true,
code: null,
challengeType: null,
stage: storedSession && isSessionLikelyValid(storedSession) ? 'session_reuse' : null,
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected: false,
};
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
if (!needsPostCfBusinessUrl(antiBot) || isPostCfBusinessUrlReady(page, antiBot)) {
return {
success: true,
code: null,
challengeType: null,
stage: storedSession && isSessionLikelyValid(storedSession) ? 'session_reuse' : null,
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected: false,
};
}
if (await tryPostCfRecoverViaReload(page, waitForUrlSettled, timeoutMs, antiBot)) {
return {
success: true,
code: null,
challengeType: null,
stage: 'post_nav_reload',
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected: false,
};
}
if (cfState.hasCfClearance) {
cfState = await invalidateStaleCfAndReload(page, waitForUrlSettled, timeoutMs);
} else {
return {
success: true,
code: null,
challengeType: null,
stage: 'pending_business_nav',
solverUsed: false,
elapsedMs: Date.now() - started,
cfDetected: false,
};
}
}
if (storedSession && isSessionLikelyValid(storedSession)) {
console.log('[CF] 会话有效但仍被拦截，尝试 reload 一次');
try {
await page.reload({ waitUntil: 'domcontentloaded', timeout: Math.min(timeoutMs, 45000) });
if (waitForUrlSettled) await waitForUrlSettled(page).catch(() => {});
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
if (!cfState.blocked) {
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
return {
success: true,
code: null,
@@ -69,7 +234,9 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
if (antiBot.turnstile !== false) {
const builtIn = await waitForTurnstile(page, { timeoutMs, useBuiltInClick: true });
if (builtIn.success) {
if (waitForUrlSettled) await waitForUrlSettled(page).catch(() => {});
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
if (!cfState.blocked) {
return {
@@ -96,7 +263,9 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
const apiWaitMs = Math.min(timeoutMs, 30000);
await waitForTurnstile(page, { timeoutMs: apiWaitMs });
if (waitForUrlSettled) await waitForUrlSettled(page).catch(() => {});
if (waitForUrlSettled) {
await waitForUrlSettled(page).catch(() => {});
}
cfState = await detectCloudflare(page);
if (!cfState.blocked) {
@@ -138,4 +307,7 @@ async function handleCloudflareChallenge(page, antiBot, waitForUrlSettled, store
module.exports = {
handleCloudflareChallenge,
isCaptchaConfigured,
needsPostCfBusinessUrl,
isPostCfBusinessUrlReady,
clearCfClearanceCookies,
};