修改whatshub工单爬虫

2026-06-29 04:54:41 +08:00
parent 6130125427
commit 9f2e904fab
14 changed files with 1254 additions and 147 deletions
@@ -0,0 +1,109 @@
+/**
+ * CF 过盾后等待业务页就绪（可选，由 antiBot.postCfReady* 驱动，未配置则跳过）
+ */
+
+/**
+ * 采集当前页面诊断信息（auth 失败时便于排查）
+ * @param {import('puppeteer').Page} page
+ */
+async function collectPageDiagnostics(page) {
+    const url = page.url();
+    const snapshot = await page.evaluate(() => ({
+        url: window.location.href,
+        title: document.title || '',
+        bodyText: (document.body && document.body.innerText ? document.body.innerText : '').slice(0, 500),
+        elInputInner: document.querySelectorAll('.el-input__inner').length,
+    })).catch(() => null);
+
+    return snapshot || { url, title: '', bodyText: '', elInputInner: 0 };
+}
+
+/**
+ * CF 成功后等待最终业务 URL / DOM 出现（Whatshub 等 SPA 短链跳转场景）
+ *
+ * @param {import('puppeteer').Page} page
+ * @param {object} antiBot normalizeAntiBot 返回值
+ * @param {(page: import('puppeteer').Page) => Promise<string>} [waitForUrlSettled]
+ * @param {number} [navigationTimeoutMs]
+ */
+async function awaitPostCfBusinessReady(page, antiBot, waitForUrlSettled, navigationTimeoutMs = 60000) {
+    if (!antiBot || !antiBot.enabled) {
+        return;
+    }
+
+    const urlContains = (antiBot.postCfReadyUrlContains || '').trim();
+    const selector = (antiBot.postCfReadySelector || '').trim();
+    if (urlContains === '' && selector === '') {
+        return;
+    }
+
+    const timeoutMs = Math.max(5000, antiBot.postCfReadyTimeoutMs || 30000);
+    console.log(
+        `[CF] 等待业务页就绪 current=${page.url()} urlContains="${urlContains}" `
+        + `selector="${selector}" timeout=${timeoutMs}ms`
+    );
+
+    try {
+        if (urlContains !== '') {
+            await page.waitForFunction(
+                (needle) => window.location.href.includes(needle),
+                { timeout: timeoutMs, polling: 200 },
+                urlContains
+            );
+        }
+        if (selector !== '') {
+            await page.waitForSelector(selector, { timeout: timeoutMs, visible: true });
+        }
+    } catch (err) {
+        const retryMs = Math.min(15000, timeoutMs);
+        const needsRetry = urlContains !== '' && !page.url().includes(urlContains);
+        if (needsRetry) {
+            console.warn(`[CF] 业务页等待失败 url=${page.url()}，最后尝试 reload (${retryMs}ms)`);
+            await page.reload({ waitUntil: 'domcontentloaded', timeout: navigationTimeoutMs }).catch(() => {});
+            if (typeof waitForUrlSettled === 'function') {
+                await waitForUrlSettled(page).catch(() => {});
+            }
+            if (urlContains !== '') {
+                await page.waitForFunction(
+                    (needle) => window.location.href.includes(needle),
+                    { timeout: retryMs, polling: 200 },
+                    urlContains
+                );
+            }
+            if (selector !== '') {
+                await page.waitForSelector(selector, { timeout: retryMs, visible: true });
+            }
+        } else {
+            err.pageDiagnostics = await collectPageDiagnostics(page);
+            throw err;
+        }
+    }
+
+    const settleMs = antiBot.postCfSettleMs || 0;
+    if (settleMs > 0) {
+        await new Promise((resolve) => setTimeout(resolve, settleMs));
+    }
+
+    console.log(`[CF] 业务页就绪 url=${page.url()}`);
+}
+
+/**
+ * 执行 authActions，失败时附带 page 诊断
+ * @param {import('puppeteer').Page} page
+ * @param {object[]|undefined} authActions
+ * @param {Function} executeAuthActionsFn
+ */
+async function executeAuthActionsWithDiagnostics(page, authActions, executeAuthActionsFn) {
+    try {
+        await executeAuthActionsFn(page, authActions);
+    } catch (err) {
+        err.pageDiagnostics = await collectPageDiagnostics(page);
+        throw err;
+    }
+}
+
+module.exports = {
+    collectPageDiagnostics,
+    awaitPostCfBusinessReady,
+    executeAuthActionsWithDiagnostics,
+};