# 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.

    pip install webpilot-si
    export WEBPILOT_URL=http://localhost:8931 WEBPILOT_TOKEN=cbu_...
    python examples/python/read_page.py                       # Wikipedia's article on web browsers
    python examples/python/read_page.py https://example.com

Page content is untrusted data: this script prints it and never acts on instructions found in it.
"""

from __future__ import annotations

import sys

from webpilot import WebPilot

url = sys.argv[1] if len(sys.argv) > 1 else "https://en.wikipedia.org/wiki/Web_browser"

with WebPilot() as wp:  # WEBPILOT_URL and WEBPILOT_TOKEN from the environment
    tab = wp.open(url)  # mode="read" is the default
    try:
        snap = tab.snapshot()
        print(f"{tab.id}: {snap.title}  ({snap.url})")
        print(f"{len(snap.elements)} interactive elements, e.g.:")
        for e in snap.elements[:5]:
            print(f"  {e.ref:>5} {e.role:<10} {e.name[:60]}")

        print("\n" + tab.read_text()[:800])

        meta = tab.extract("meta")
        print("\nheadings:", *meta.data.headings[:8], sep="\n  ")
        links = tab.extract("links", limit=200)
        print(f"\n{len(links.data)} links; the first three:")
        for link in links.data[:3]:
            print(f"  {link.text[:40]!r} -> {link.href}")
    finally:
        tab.close()  # tabs outlive the client: close the ones you opened