/**
 * SEC EDGAR Hyperscaler CapEx Scraper
 *
 * Uses the SEC EDGAR XBRL API (free, no auth) to extract quarterly CapEx
 * from Amazon, Microsoft, Google/Alphabet, and Meta 10-Q/10-K filings.
 *
 * XBRL concept: us-gaap/PaymentsToAcquirePropertyPlantAndEquipment
 * API: https://data.sec.gov/api/xbrl/companyfacts/CIK{cik}.json
 *
 * Hyperscaler DC CapEx is the strongest 6-12 month leading indicator
 * for 400G/800G transceiver demand.
 */

import { pool } from "../utils/db";
import { logger } from "../utils/logger";

const COMPANIES: Record<string, { name: string; cik: string; dcPct: number }> = {
  amazon:    { name: "Amazon (AWS)",    cik: "0001018724", dcPct: 0.65 }, // ~65% of CapEx is AWS infra
  microsoft: { name: "Microsoft Azure", cik: "0000789019", dcPct: 0.55 },
  alphabet:  { name: "Google Cloud",    cik: "0001652044", dcPct: 0.60 },
  meta:      { name: "Meta AI/DC",      cik: "0001326801", dcPct: 0.85 }, // almost all meta capex is DC
};

const EDGAR_BASE = "https://data.sec.gov/api/xbrl/companyfacts";
const HEADERS = {
  "User-Agent": "TIP-DataCollector/1.0 contact@context-x.org",
  Accept: "application/json",
};

interface XbrlUnit {
  end: string;        // ISO date
  val: number;        // value in USD
  form: string;       // '10-Q' or '10-K'
  filed: string;
  frame?: string;     // 'CY2024Q1' etc
  accn: string;       // accession number
  fp?: string;        // Q1, Q2, Q3, FY
  fy?: number;        // fiscal year
}

async function fetchCapexData(cik: string): Promise<XbrlUnit[]> {
  const url = `${EDGAR_BASE}/CIK${cik}.json`;
  const res = await fetch(url, { headers: HEADERS });
  if (!res.ok) throw new Error(`EDGAR fetch failed for CIK ${cik}: ${res.status}`);
  const data = await res.json() as {
    facts: {
      "us-gaap"?: {
        PaymentsToAcquirePropertyPlantAndEquipment?: {
          units: { USD: XbrlUnit[] }
        }
      }
    }
  };

  return data.facts?.["us-gaap"]
    ?.PaymentsToAcquirePropertyPlantAndEquipment
    ?.units?.USD ?? [];
}

function labelFromFrame(unit: XbrlUnit): string {
  if (unit.frame) {
    const m = unit.frame.match(/CY(\d{4})(Q\d)?/);
    if (m) return m[2] ? `${m[2]} ${m[1]}` : `FY ${m[1]}`;
  }
  if (unit.fp && unit.fy) return `${unit.fp === "FY" ? "FY" : unit.fp} ${unit.fy}`;
  return unit.end.substring(0, 7); // YYYY-MM
}

export async function scrapeSecEdgar(): Promise<void> {
  logger.info("SEC EDGAR CapEx scraper starting");
  let inserted = 0;

  for (const [key, company] of Object.entries(COMPANIES)) {
    try {
      logger.info(`Fetching EDGAR data for ${company.name}`);
      await new Promise(r => setTimeout(r, 800)); // respect SEC rate limit: 10 req/sec

      const units = await fetchCapexData(company.cik);
      if (!units.length) { logger.warn(`No XBRL data for ${company.name}`); continue; }

      // Filter to quarterly 10-Q/10-K filings from last 3 years
      const cutoff = new Date();
      cutoff.setFullYear(cutoff.getFullYear() - 3);

      const recent = units
        .filter(u => (u.form === "10-Q" || u.form === "10-K") && new Date(u.end) >= cutoff)
        .sort((a, b) => new Date(b.end).getTime() - new Date(a.end).getTime());

      // Deduplicate by period end date — keep most recently filed
      const seen = new Map<string, XbrlUnit>();
      for (const u of recent) {
        if (!seen.has(u.end)) seen.set(u.end, u);
      }

      for (const unit of seen.values()) {
        const capexM = unit.val / 1_000_000; // convert to millions
        const dcCapexM = Math.round(capexM * company.dcPct * 10) / 10;
        const periodLabel = labelFromFrame(unit);
        const sourceUrl = `https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=${company.cik}&type=${unit.form}&dateb=&owner=include&count=40`;

        await pool.query(`
          INSERT INTO hyperscaler_capex
            (company, period_label, period_end, capex_usd_millions, dc_capex_est_millions, source_url, filing_type)
          VALUES ($1, $2, $3, $4, $5, $6, $7)
          ON CONFLICT (company, period_end) DO UPDATE SET
            capex_usd_millions    = EXCLUDED.capex_usd_millions,
            dc_capex_est_millions = EXCLUDED.dc_capex_est_millions,
            period_label          = EXCLUDED.period_label,
            filing_type           = EXCLUDED.filing_type
        `, [key, periodLabel, unit.end, Math.round(capexM * 10) / 10, dcCapexM, sourceUrl, unit.form]);
        inserted++;
      }

      // Compute YoY growth for most recent period
      await pool.query(`
        UPDATE hyperscaler_capex h1
        SET yoy_growth_pct = ROUND(
          (h1.capex_usd_millions - h2.capex_usd_millions) / NULLIF(h2.capex_usd_millions, 0) * 100, 1
        )
        FROM hyperscaler_capex h2
        WHERE h1.company = $1
          AND h2.company = $1
          AND h2.period_end = h1.period_end - INTERVAL '1 year'
          AND h1.yoy_growth_pct IS NULL
      `, [key]);

      logger.info(`${company.name}: ${seen.size} periods upserted`);
    } catch (err) {
      logger.error(`EDGAR scraper failed for ${company.name}`, { err });
    }
  }

  logger.info(`SEC EDGAR scraper done — ${inserted} records`);
}