Wait for hydrated ChatGPT share messages

2026-07-08 02:25:33 +00:00
parent 11229e911f
commit 202c7a4466
2 changed files with 162 additions and 2 deletions
@@ -101,8 +101,21 @@ async function extractConversation(browser: Browser, sourceUrl: string): Promise
  const page = await context.newPage();
  page.setDefaultTimeout(config.captureTimeoutMs);
  await page.goto(sourceUrl, { waitUntil: "domcontentloaded", timeout: config.captureTimeoutMs });
-  await page.waitForLoadState("networkidle", { timeout: Math.min(config.captureTimeoutMs, 60_000) }).catch(() => undefined);
-  await page.waitForTimeout(2500);
+  await page
+    .waitForFunction(
+      () => {
+        const bodyText = document.body?.innerText || document.body?.textContent || "";
+        if (/verify you are human|captcha|unusual activity/i.test(bodyText)) return true;
+        if (document.querySelectorAll("[data-message-author-role]").length > 0) return true;
+        const markdown = document.querySelector("main .markdown, main [class*=markdown]");
+        if ((markdown?.textContent || "").trim().length > 0) return true;
+        return /(?:^|\s)(You said:|ChatGPT said:)/.test(bodyText);
+      },
+      { timeout: config.captureTimeoutMs }
+    )
+    .catch(() => undefined);
+  await page.waitForLoadState("networkidle", { timeout: 5_000 }).catch(() => undefined);
+  await page.waitForTimeout(1000);

  const extracted = await page.evaluate(() => {
    function cleanHtml(root: Element): string {