Track a web page for new items and get a task to review each

from urllib.parse import urljoin
from bs4 import BeautifulSoup
from google.oauth2.credentials import Credentials
from googleapiclient.discovery import build
from clickruntime import on_trigger, http, state, ai, credentials

PAGE = "https://example.gov/publications"
SHEET_ID = "your-spreadsheet-id"

@on_trigger
def run(event):
    links = {urljoin(PAGE, a["href"]): a.get_text(" ", strip=True)
             for a in BeautifulSoup(http.get(PAGE).text, "html.parser").select("main a[href]") if a.get_text(strip=True)}
    seen = set(state.get("seen") or [])
    new = [(url, title) for url, title in links.items() if url not in seen][:20]
    if not new:
        return {"new": 0}
    creds = Credentials(credentials.connection("google", "work"))
    sheets = build("sheets", "v4", credentials=creds, cache_discovery=False).spreadsheets()
    tasks = build("tasks", "v1", credentials=creds, cache_discovery=False).tasks()
    rows = []
    for url, title in new:
        summary = ai.generate(f"Summarise in two plain sentences:\n{title}\n" + BeautifulSoup(http.get(url).text, "html.parser")
                              .get_text(" ", strip=True)[:3000], max_output_tokens=120).text.strip()
        rows.append([(event.scheduled_time or "")[:10], title, url, summary])
        tasks.insert(tasklist="@default", body={"title": f"Review: {title}", "notes": f"{summary}\n{url}"}).execute()
    sheets.values().append(spreadsheetId=SHEET_ID, range="Updates!A:D", valueInputOption="USER_ENTERED",
                           body={"values": rows}).execute()
    state.set("seen", sorted(seen | {u for u, _ in new})[-2000:])
    return {"new": len(new)}