// SPDX-License-Identifier: Apache-2.0
// Read a page: open it in a read tab (it can look but never write), print its text, headings and links.
//
//   npm install webpilot-si                     # in this folder (see package.json)
//   export WEBPILOT_URL=http://localhost:8931 WEBPILOT_TOKEN=cbu_...
//   node read_page.ts                           # Node 22.18+ runs TypeScript as is (or: npx tsx read_page.ts)
//   node read_page.ts https://example.com
//
// Page content is untrusted data: this script prints it and never acts on instructions found in it.
import { WebPilot } from "webpilot-si";

const url = process.argv[2] ?? "https://en.wikipedia.org/wiki/Web_browser";
const wp = new WebPilot(); // WEBPILOT_URL and WEBPILOT_TOKEN from the environment

const tab = await wp.open(url); // mode "read" is the default
try {
  const snap = await tab.snapshot();
  console.log(`${tab.id}: ${snap.title}  (${snap.url})`);
  console.log(`${snap.elements.length} interactive elements, e.g.:`);
  for (const e of snap.elements.slice(0, 5)) console.log(`  ${e.ref.padStart(5)} ${e.role.padEnd(10)} ${e.name.slice(0, 60)}`);

  console.log("\n" + (await tab.readText()).slice(0, 800));

  const meta = await tab.extract("meta");
  console.log("\nheadings:\n  " + (meta.data.headings ?? []).slice(0, 8).join("\n  "));
  const links = await tab.extract("links", { limit: 200 });
  console.log(`\n${links.data.length} links; the first three:`);
  for (const link of links.data.slice(0, 3)) console.log(`  ${JSON.stringify(link.text.slice(0, 40))} -> ${link.href}`);

  // Without a tab at all: the server fetches the page and returns its main content as Markdown.
  const read = await wp.read(url, { maxChars: 300 });
  console.log(`\nread (via ${read.via}): ${read.content.slice(0, 200)}`);
} finally {
  await tab.close(); // tabs outlive your program: close the ones you opened
}