// ─────────────────────────────────────────────────────────────────────────────
// Competitor news fetcher → Persian, with source + link
//
// For each steel company it pulls the latest headline from Google News RSS,
// translates it to Persian, and writes public/competitor-news.json. The
// dashboard ("آخرین خبر و اقدام" / رصد رقبا) reads that file at runtime.
//
// No API key required (uses Google News RSS + Google translate gtx endpoint).
//
// Usage:
//   node scripts/fetch-competitor-news.mjs            # run once now
//   node scripts/fetch-competitor-news.mjs --daemon   # run now, then daily 16:00 (Asia/Tehran)
//
// Schedule daily at 16:00 without the daemon:
//   • Linux/macOS cron:   30 12 * * *  cd /path/to/repo && node scripts/fetch-competitor-news.mjs   (12:30 UTC = 16:00 Tehran)
//   • Windows Task Scheduler: daily trigger 16:00 → program "node", args "scripts/fetch-competitor-news.mjs", start-in = repo root
//
// Requires Node 18+ (global fetch).
// ─────────────────────────────────────────────────────────────────────────────

import { readFile, writeFile, mkdir } from 'node:fs/promises'
import { dirname, resolve } from 'node:path'
import { fileURLToPath } from 'node:url'

const __dirname = dirname(fileURLToPath(import.meta.url))
const OUT = resolve(__dirname, '..', 'public', 'competitor-news.json')
const HISTORY = resolve(__dirname, 'news-history.json') // remembers topics already shown so they never repeat

// keys MUST match competitor.name in src/pages/SteelDashboard/sections/IndustrySection.tsx
const COMPANIES = [
  { name: 'Foulad Mobarakeh',         q: 'Mobarakeh Steel Company' },
  { name: 'China Baowu',              q: 'China Baowu steel' },
  { name: 'ArcelorMittal',            q: 'ArcelorMittal' },
  { name: 'Ansteel Group',            q: 'Ansteel Group steel' },
  { name: 'Nippon Steel',             q: 'Nippon Steel' },
  { name: 'HBIS Group',               q: 'HBIS Group steel' },
  { name: 'POSCO Holdings',           q: 'POSCO steel' },
  { name: 'Tata Steel',               q: 'Tata Steel' },
  { name: 'JSW Steel',                q: 'JSW Steel' },
  { name: 'Nucor',                    q: 'Nucor steel' },
  { name: 'Khouzestan Steel',         q: 'Khouzestan Steel Company' },
  { name: 'Esfahan Steel (Zob Ahan)', q: 'Esfahan Steel Zob Ahan' },
  { name: 'Erdemir',                  q: 'Erdemir steel Turkey' },
  { name: 'Tosyalı Holding',          q: 'Tosyali steel' },
  { name: 'EMSTEEL',                  q: 'EMSTEEL Emirates Steel' },
  { name: 'Ezz Steel',                q: 'Ezz Steel Egypt' },
]

const UA = { 'User-Agent': 'Mozilla/5.0 (compatible; SteelForesightNewsBot/1.0)' }

async function fetchT(url, ms = 9000) {
  const ctrl = new AbortController()
  const id = setTimeout(() => ctrl.abort(), ms)
  try { return await fetch(url, { headers: UA, signal: ctrl.signal }) }
  finally { clearTimeout(id) }
}

function decode(s = '') {
  return s
    .replace(/<!\[CDATA\[([\s\S]*?)\]\]>/g, '$1')
    .replace(/&amp;/g, '&').replace(/&lt;/g, '<').replace(/&gt;/g, '>')
    .replace(/&quot;/g, '"').replace(/&#39;/g, "'").replace(/&apos;/g, "'")
    .trim()
}

async function fetchItems(query) {
  // last 30 days only; pull several so we can skip topics that are already taken
  const url = `https://news.google.com/rss/search?q=${encodeURIComponent(query + ' when:30d')}&hl=en-US&gl=US&ceid=US:en`
  const res = await fetchT(url)
  if (!res.ok) throw new Error(`news HTTP ${res.status}`)
  const xml = await res.text()
  const blocks = [...xml.matchAll(/<item>([\s\S]*?)<\/item>/g)].map((m) => m[1])
  return blocks
    .map((block) => {
      const rawTitle = decode(block.match(/<title>([\s\S]*?)<\/title>/)?.[1] || '')
      const link = decode(block.match(/<link>([\s\S]*?)<\/link>/)?.[1] || '')
      const pubDate = decode(block.match(/<pubDate>([\s\S]*?)<\/pubDate>/)?.[1] || '')
      const source = decode(block.match(/<source[^>]*>([\s\S]*?)<\/source>/)?.[1] || '')
      // Google News titles end with " - Source" — strip it for a clean headline
      const title = source && rawTitle.endsWith(` - ${source}`) ? rawTitle.slice(0, -(source.length + 3)) : rawTitle
      return { title, link, pubDate, source }
    })
    .filter((it) => it.title)
}

// normalised "topic" so near-identical headlines collapse to one key (for de-duping)
function topicKey(title) {
  return title
    .toLowerCase()
    .replace(/[^a-z0-9؀-ۿ\s]/g, ' ')
    .replace(/\s+/g, ' ')
    .trim()
    .split(' ')
    .slice(0, 8)
    .join(' ')
}

async function loadHistory() {
  try { return new Set(JSON.parse(await readFile(HISTORY, 'utf8'))) } catch { return new Set() }
}
async function saveHistory(seen) {
  try { await writeFile(HISTORY, JSON.stringify([...seen].slice(-600)), 'utf8') } catch { /* ignore */ }
}

async function toPersian(text) {
  if (!text) return ''
  const url = `https://translate.googleapis.com/translate_a/single?client=gtx&sl=en&tl=fa&dt=t&q=${encodeURIComponent(text)}`
  const res = await fetchT(url)
  if (!res.ok) throw new Error(`translate HTTP ${res.status}`)
  const data = await res.json()
  return (data?.[0] || []).map((seg) => seg[0]).join('')
}

function faDate(iso) {
  try {
    return new Intl.DateTimeFormat('fa-IR', { year: 'numeric', month: 'long', day: 'numeric' }).format(new Date(iso))
  } catch {
    return ''
  }
}

const sleep = (ms) => new Promise((r) => setTimeout(r, ms))

async function runOnce() {
  const history = await loadHistory()  // topics shown on previous days
  const usedThisRun = new Set()        // topics already taken in this run
  const items = {}
  for (const c of COMPANIES) {
    try {
      const list = await fetchItems(c.q)
      // first headline whose topic is new — not taken this run AND not shown before
      let pick = list.find((it) => { const k = topicKey(it.title); return k && !usedThisRun.has(k) && !history.has(k) })
      // if everything is a repeat, at least don't clash with another company in this same run
      if (!pick) pick = list.find((it) => !usedThisRun.has(topicKey(it.title)))
      if (!pick) { console.warn(`· no fresh news: ${c.name}`); continue }
      const key = topicKey(pick.title)
      usedThisRun.add(key); history.add(key)
      const newsFa = await toPersian(pick.title)
      items[c.name] = { newsFa, source: pick.source, link: pick.link, date: faDate(pick.pubDate) }
      console.log(`✓ ${c.name} — ${newsFa.slice(0, 64)}`)
    } catch (e) {
      console.warn(`✗ ${c.name}: ${e.message}`)
    }
    await sleep(1200) // stay polite to the free endpoints
  }
  await saveHistory(history)
  await mkdir(dirname(OUT), { recursive: true })
  await writeFile(OUT, JSON.stringify({ updatedAt: new Date().toISOString(), items }, null, 2), 'utf8')
  console.log(`\n📰 wrote ${Object.keys(items).length}/${COMPANIES.length} items (no repeated topics) → ${OUT}`)
}

// 16:00 Asia/Tehran = 12:30 UTC (Iran dropped DST in 2022, fixed at UTC+3:30)
function msUntilNextRun() {
  const now = new Date()
  const next = new Date(now)
  next.setUTCHours(12, 30, 0, 0)
  if (next <= now) next.setUTCDate(next.getUTCDate() + 1)
  return next.getTime() - now.getTime()
}

async function daemon() {
  console.log('🕓 news daemon up — runs now, then every day at 16:00 Asia/Tehran')
  await runOnce()
  const schedule = () => {
    const wait = msUntilNextRun()
    console.log(`next run in ${(wait / 3_600_000).toFixed(2)}h`)
    setTimeout(async () => { await runOnce(); schedule() }, wait)
  }
  schedule()
}

if (process.argv.includes('--daemon')) daemon()
else runOnce()