import { defineConnector, raw, type NormalizedBatch, type ProposedInstrument } from "@market-atlas/connector-sdk";

const URL = "https://www.sec.gov/files/company_tickers_exchange.json";
const EXCHANGE_MAP: Record<string, { exchangeId: string; mic: string } | null> = {
  Nasdaq: { exchangeId: "xnas", mic: "XNAS" },
  NYSE: { exchangeId: "xnys", mic: "XNYS" },
  CBOE: { exchangeId: "bats", mic: "BATS" },
  OTC: null,
};

/** SEC company ↔ ticker ↔ exchange directory (bulk JSON, refreshed daily). Enriches instruments with CIK + company; creates listed US equities. */
export const secCompanyTickers = defineConnector({
  metadata: {
    id: "sec-company-tickers",
    name: "SEC — company tickers & exchanges directory",
    version: "1.0.0",
    sourceId: "sec-edgar",
    organization: "U.S. Securities and Exchange Commission",
    sourceType: "BULK_FILE",
    jurisdiction: "US",
    rightsStatus: "OFFICIAL_OPEN_DATA",
    realtimeStatus: "END_OF_DAY",
    expectedLatencyMs: null,
    supportsStreaming: false,
    supportsHistorical: false,
    assetClasses: ["EQUITY"],
    exchanges: ["xnas", "xnys", "bats"],
    homepage: "https://www.sec.gov/files/company_tickers_exchange.json",
    description: "Bulk directory of ~10,000 SEC registrants with ticker, listing exchange and CIK. Used to resolve filings to instruments and to attach companies (CIK) to US equities.",
    rightsNotes: "U.S. government data (public domain).",
    termsUrl: "https://www.sec.gov/os/accessing-edgar-data",
    sourceFamily: "sec",
    enabled: true,
  },
  rateLimits: { "www.sec.gov": 2 },
  schedule: { intervalMs: 24 * 60 * 60_000 },
  async poll(ctx) {
    const res = await ctx.http.getText(URL, { timeoutMs: 60_000, headers: { accept: "application/json" } });
    if (res.notModified) return [];
    return [raw("sec-company-tickers", "sec-edgar", "directory", JSON.parse(res.text))];
  },
  normalize(r): NormalizedBatch {
    const p = r.payload as { fields?: string[]; data?: Array<[number, string, string, string | null]> };
    if (r.kind !== "directory" || !Array.isArray(p.data) || !Array.isArray(p.fields)) return { observations: [] };
    const idx = (n: string) => p.fields!.indexOf(n);
    const iCik = idx("cik"), iName = idx("name"), iTicker = idx("ticker"), iEx = idx("exchange");
    if ([iCik, iName, iTicker, iEx].some((i) => i < 0)) return { observations: [] };
    const instruments: ProposedInstrument[] = [];
    for (const row of p.data) {
      const ticker = String(row[iTicker] ?? "").trim();
      const name = String(row[iName] ?? "").trim();
      const cik = String(row[iCik] ?? "").trim();
      const ex = row[iEx] ? EXCHANGE_MAP[String(row[iEx])] : null;
      if (!ticker || !name || !cik) continue;
      const listed = ex != null;
      instruments.push({
        symbol: ticker.replace("-", "."),
        aliases: ticker.includes("-") ? [ticker] : [],
        // Listed names may be created (id derives from the exchange); OTC/unknown venues only enrich existing rows.
        createIfMissing: listed,
        hint: { assetClass: "EQUITY", name: titleCase(name), exchangeId: ex?.exchangeId ?? null, mic: ex?.mic ?? null, currency: "USD", country: "US", securityType: "COMMON_STOCK", companyName: titleCase(name), cik },
      });
    }
    return { observations: [], instruments, stats: { rows: p.data.length } };
  },
  fixturesDir: "fixtures",
});

/** "MICROSOFT CORP" → "Microsoft Corp" while preserving already mixed-case names like "Apple Inc.". */
export function titleCase(s: string): string {
  if (s !== s.toUpperCase()) return s;
  return s
    .toLowerCase()
    .replace(/(^|[\s\-/(&])([a-z])/g, (m, pre: string, c: string) => pre + c.toUpperCase())
    .replace(/\b(Llc|Inc|Corp|Ltd|Plc|Co|Sa|Nv|Ag|Lp)\b/g, (w) => w)
    .replace(/\b(Etf|Reit|Spac|Usa|Us|Adr|Ii|Iii|Iv)\b/g, (w) => w.toUpperCase());
}