import type { ConnectorConfig, ConnectorContext, DiscoveredUrl, ExtractedRecord, Parser, RawDocument } from "@dci/connectors";
import { GenericConnector } from "@dci/connectors";
import { cleanText } from "@dci/core";
import { EXTRACTOR_VERSION } from "./extract-project.js";
import { detectOperators } from "./operators-lexicon.js";
import { US_STATES } from "./locations-lexicon.js";

/**
 * SEC EDGAR full-text search (efts.sec.gov/LATEST/search-index) — public JSON API, no key. The API ignores
 * `dateRange=30d`, so the connector computes `dateRange=custom&startdt=…&enddt=…` at discovery time
 * (`implementation: news_edgar_fts`, params.queries / params.lookbackDays / params.forms). One `news_event`
 * per filing (exhibit preferred), no fetch of the filing itself — links point to the SEC archive.
 * SEC fair-access policy: identify the crawler with a contact address (DEFAULT_UA does), ≤ 10 req/s.
 */
interface EdgarHit {
  _id: string; // "0001079973-26-000696:ex99x2.htm"
  _source: { ciks?: string[]; display_names?: string[]; file_date?: string; form?: string; root_forms?: string[]; file_type?: string; file_description?: string; adsh?: string; items?: string[]; biz_locations?: string[]; biz_states?: string[]; period_ending?: string; sics?: string[] };
}
interface EdgarResponse { hits?: { total?: { value?: number }; hits?: EdgarHit[] }; query?: unknown }

const iso = (d: Date): string => d.toISOString().slice(0, 10);

export class EdgarFtsConnector extends GenericConnector {
  async discover(_ctx: ConnectorContext): Promise<DiscoveredUrl[]> {
    const p = (this.cfg.params ?? {}) as { queries?: unknown; forms?: unknown; lookbackDays?: unknown };
    const queries = Array.isArray(p.queries) && p.queries.length ? p.queries.map(String) : ['"data center"'];
    const forms = typeof p.forms === "string" && p.forms ? p.forms : "8-K";
    const lookback = Number.isFinite(Number(p.lookbackDays)) && Number(p.lookbackDays) > 0 ? Number(p.lookbackDays) : 30;
    const end = new Date();
    const start = new Date(end.getTime() - lookback * 86_400_000);
    return queries.map((q, i) => ({
      url: `https://efts.sec.gov/LATEST/search-index?q=${encodeURIComponent(q)}&forms=${encodeURIComponent(forms)}&dateRange=custom&startdt=${iso(start)}&enddt=${iso(end)}`,
      group: "filings",
      pageType: "financial_disclosure",
      priority: 90 - i,
      discoveredFrom: null,
      meta: { query: q, forms, lookbackDays: lookback },
    }));
  }
}

export function edgarFilingUrl(hit: EdgarHit): { url: string; adsh: string; file: string } | null {
  const [adshRaw, file] = hit._id.split(":");
  const adsh = hit._source.adsh ?? adshRaw;
  const cik = hit._source.ciks?.[0];
  if (!adsh || !cik) return null;
  const folder = adsh.replace(/-/g, "");
  const url = file ? `https://www.sec.gov/Archives/edgar/data/${Number(cik)}/${folder}/${file}` : `https://www.sec.gov/Archives/edgar/data/${Number(cik)}/${folder}/`;
  return { url, adsh, file: file ?? "" };
}

function companyName(display: string | undefined): string | null {
  if (!display) return null;
  return cleanText(display.replace(/\s*\([A-Z0-9.]{1,10}\)\s*\(CIK \d+\)\s*$/, "").replace(/\s*\(CIK \d+\)\s*$/, ""));
}

/** Rank exhibits above the cover 8-K so one filing yields the press-release exhibit when it exists. */
function rank(hit: EdgarHit): number {
  const ft = (hit._source.file_type ?? "").toUpperCase();
  if (/^EX-99/.test(ft)) return 3;
  if (/^EX-/.test(ft)) return 2;
  return 1;
}

export const edgarFtsParser: Parser = {
  name: "news_edgar_fts_v1",
  version: EXTRACTOR_VERSION,
  parse(doc: RawDocument, ctx: ConnectorContext): ExtractedRecord[] {
    let res: EdgarResponse;
    try { res = JSON.parse(doc.text) as EdgarResponse; } catch { ctx.log("warn", `news_edgar_fts_v1: response is not JSON (${doc.finalUrl})`); return []; }
    const hits = res.hits?.hits ?? [];
    const query = typeof doc.meta?.query === "string" ? doc.meta.query : new URL(doc.finalUrl).searchParams.get("q") ?? "";
    const byFiling = new Map<string, EdgarHit>();
    for (const h of hits) {
      const adsh = h._source.adsh ?? h._id.split(":")[0]!;
      const prev = byFiling.get(adsh);
      if (!prev || rank(h) > rank(prev)) byFiling.set(adsh, h);
    }
    const out: ExtractedRecord[] = [];
    for (const h of byFiling.values()) {
      const link = edgarFilingUrl(h);
      if (!link) continue;
      const s = h._source;
      const company = companyName(s.display_names?.[0]);
      if (!company) continue;
      const form = s.root_forms?.[0] ?? s.form ?? "8-K";
      const fileType = s.file_type && s.file_type.toUpperCase() !== form.toUpperCase() ? s.file_type : null;
      const title = `${company}: Form ${form}${fileType ? ` (${fileType}${s.file_description && !/^EXHIBIT/i.test(s.file_description) ? ` — ${s.file_description}` : ""})` : ""}`;
      const loc = s.biz_locations?.[0] ?? null;
      const state = s.biz_states?.[0] ?? null;
      const summary = cleanText(`${form} filed ${s.file_date ?? "(date unknown)"} by ${(s.display_names ?? [company]).map((d) => companyName(d)).filter(Boolean).join(", ")}${s.items?.length ? `; items ${s.items.join(", ")}` : ""}${loc ? `; registrant located in ${loc}` : ""}. Matched EDGAR full-text search ${query}.`);
      const operators = detectOperators(company).map((o) => o.name);
      const isUs = state ? Object.prototype.hasOwnProperty.call(US_STATES, state.toUpperCase()) : false;
      const city = loc ? cleanText(loc.split(",")[0]) : null;
      out.push({
        kind: "news_event",
        key: `${ctx.connectorId}:news:${link.adsh}`,
        url: link.url,
        pageType: "financial_disclosure",
        certainty: 0.7,
        data: {
          title, url: link.url, publishedAt: s.file_date ?? null, summary, pageType: "financial_disclosure", eventType: "news",
          mw: [], status: null, money: null, text: null,
          operators, countries: isUs ? ["US"] : [], cities: city && isUs ? [city] : [],
          filing: { adsh: link.adsh, form, fileType: s.file_type ?? null, items: s.items ?? [], cik: s.ciks?.[0] ?? null, periodEnding: s.period_ending ?? null, query },
          relevance: "strong", extractor: EXTRACTOR_VERSION,
        },
        methods: { title: "edgar:display_names", publishedAt: "edgar:file_date", summary: "edgar:metadata", operators: "lexicon:operator" },
      });
    }
    ctx.log("info", `news_edgar_fts_v1: ${hits.length} hits → ${out.length} filings (${query})`);
    return out;
  },
};

export function edgarImplementation(cfg: ConnectorConfig): EdgarFtsConnector { return new EdgarFtsConnector(cfg); }