import { defineConnector, extractTables, htmlFingerprints, raw, stripHtml, type NormalizedBatch, type ProposedEvent } from "@market-atlas/connector-sdk";

const URL = "https://www.nasdaqtrader.com/Trader.aspx?id=Calendar";
/** The Nasdaq schedule applies to all US equity and options markets. */
const US_EXCHANGES = ["xnas", "xnys", "xase", "arcx", "bats", "xcbo"];
const MONTHS: Record<string, number> = { january: 1, february: 2, march: 3, april: 4, may: 5, june: 6, july: 7, august: 8, september: 9, october: 10, november: 11, december: 12 };

/**
 * HTML change-detection connector: the Nasdaq Trader "Holiday Schedule" page. Section-level
 * fingerprints are stored in connector state; a change emits DOCUMENT_CHANGED and the parsed
 * table feeds the market-hours engine (closed days and 1:00 p.m. early closes).
 */
export const nasdaqMarketCalendar = defineConnector({
  metadata: {
    id: "nasdaq-market-calendar",
    name: "Nasdaq Trader — US market holiday schedule (HTML)",
    version: "1.0.0",
    sourceId: "nasdaq-trader",
    organization: "Nasdaq, Inc.",
    sourceType: "HTML",
    jurisdiction: "US",
    rightsStatus: "PUBLIC_ATTRIBUTED",
    realtimeStatus: "END_OF_DAY",
    expectedLatencyMs: null,
    supportsStreaming: false,
    supportsHistorical: false,
    assetClasses: ["EQUITY", "ETF", "INDEX"],
    exchanges: US_EXCHANGES,
    homepage: URL,
    description: "Parses the U.S. Equity and Options Markets Holiday Schedule table (date, holiday, status) and detects any change in the page (document + section hashes). Output feeds the exchange calendar for all US venues.",
    rightsNotes: "Public reference page; parsed for facts (dates), no content redistribution.",
    termsUrl: "https://www.nasdaqtrader.com/Trader.aspx?id=Terms",
    sourceFamily: "nasdaq",
    enabled: true,
  },
  rateLimits: { "www.nasdaqtrader.com": 1 },
  schedule: { intervalMs: 12 * 60 * 60_000 },
  async poll(ctx) {
    const res = await ctx.http.getText(URL, { timeoutMs: 30_000, headers: { accept: "text/html" } });
    const fp = htmlFingerprints(res.text);
    const prev = await ctx.state.get<{ document: string; sections: Record<string, string> }>("fingerprint");
    const changed = !prev || prev.document !== fp.document;
    const changedSections = prev ? Object.keys(fp.sections).filter((k) => prev.sections[k] !== fp.sections[k]) : Object.keys(fp.sections);
    await ctx.state.set("fingerprint", fp);
    await ctx.state.set("last_checked_at", new Date().toISOString());
    // Always emit the page (holidays are idempotent); mark whether the document changed so normalize can raise an event.
    return [raw("nasdaq-market-calendar", "nasdaq-trader", "page", res.text, { changed, first: !prev, changed_sections: changedSections.slice(0, 10), document_hash: fp.document })];
  },
  normalize(r): NormalizedBatch {
    if (r.kind !== "page" || typeof r.payload !== "string") return { observations: [] };
    const html = r.payload;
    const yearMatch = stripHtml(html).match(/Holiday Schedule (\d{4})/);
    const holidays: NonNullable<NormalizedBatch["holidays"]> = [];
    for (const table of extractTables(html)) {
      for (const row of table) {
        if (row.length < 3) continue;
        const date = parseLongDate(row[0]!);
        if (!date) continue;
        const name = row[1]!.replace(/\*+/g, "").trim();
        const status = row[2]!.trim();
        const early = status.match(/(\d{1,2})(?::(\d{2}))?\s*(a|p)\.?m\.?/i);
        const kind = /closed/i.test(status) ? "CLOSED" : early ? "EARLY_CLOSE" : null;
        if (!kind) continue;
        let closeTime: string | null = null;
        if (early) {
          let h = Number(early[1]);
          if (early[3]!.toLowerCase() === "p" && h < 12) h += 12;
          closeTime = `${String(h).padStart(2, "0")}:${early[2] ?? "00"}`;
        }
        const cleaned = name.replace(/early close/i, "").replace(/^[\s\-–]+|[\s\-–]+$/g, "").trim();
        const label = kind === "EARLY_CLOSE" ? (cleaned && !/^u\.s\.?$/i.test(cleaned) ? `${cleaned} (early close)` : "Early close (U.S. markets)") : name;
        for (const ex of US_EXCHANGES) holidays.push({ exchangeId: ex, date, name: label, kind, closeTime });
      }
    }
    const events: ProposedEvent[] = [];
    if (r.meta?.changed && !r.meta?.first) {
      events.push({
        type: "DOCUMENT_CHANGED",
        timestamp: r.receivedAt,
        severity: "NOTICE",
        confidence: 1,
        title: `Nasdaq Trader holiday schedule page changed${yearMatch ? ` (${yearMatch[1]})` : ""}`,
        summary: `Sections changed: ${(r.meta.changed_sections as string[])?.join(", ") || "n/a"}`,
        dedupeKey: `nasdaq-calendar:${r.meta.document_hash}`,
        data: { url: URL, document_hash: r.meta.document_hash, changed_sections: r.meta.changed_sections, holidays_parsed: holidays.length / US_EXCHANGES.length },
      });
    }
    return { observations: [], holidays, events, stats: { holidays: holidays.length / US_EXCHANGES.length } };
  },
  fixturesDir: "fixtures",
});

/** "January 19, 2026" → "2026-01-19" */
export function parseLongDate(s: string): string | null {
  const m = s.trim().match(/^([A-Za-z]+)\s+(\d{1,2}),\s*(\d{4})$/);
  if (!m) return null;
  const mo = MONTHS[m[1]!.toLowerCase()];
  if (!mo) return null;
  return `${m[3]}-${String(mo).padStart(2, "0")}-${m[2]!.padStart(2, "0")}`;
}