修复A2C工单蜘蛛

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

@@ -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,
 };