import { load } from "cheerio";
import type { ExtractedRecord, Parser, RawDocument } from "@dci/connectors";
import { certificationsFrom, cleanText, coolingFrom, description, firstText, lastSlug, mwWithContext, parseCityLine, parseMw, record, renewableFrom, statusFromText } from "./shared.js";

/**
 * CyrusOne facility pages (www.cyrusone.com/data-centers/<region>/<slug>). HubSpot CMS:
 *   <h1>Frankfurt, Germany</h1><h3>FRA5</h3>, breadcrumb #hs_cos_wrapper_name "Frankfurt, Germany: FRA5",
 *   "Key Features" tiles: <h3>54 MW</h3><p>Total IT Capacity</p>, <h3>18,000 sq. m.</h3><p>Technical IT Space</p>,
 *   accordions "Technical Specifications" / "Sustainability" (li/span items).
 * No street address and no coordinates are published → geo left null (city-level from other sources).
 */
export const cyrusOneFacility: Parser = {
  name: "cyrusone_facility_v1",
  version: "1.0.0",
  pageTypes: ["facility_page"],
  parse(doc: RawDocument): ExtractedRecord[] {
    const url = doc.finalUrl;
    if (!/\/data-centers\/(north-america|emea|apac)\/[a-z0-9-]+\/?$/.test(url)) return [];
    const $ = load(doc.text);
    const h1 = firstText($, "h1");
    if (!h1) return [];
    const codeLine = cleanText($("h1").first().nextAll("h3").first().text());
    const codes = codeLine ? (codeLine.toUpperCase().match(/\b[A-Z]{2,4}\d{1,2}[A-Z]?\b/g) ?? []) : [];
    // h1 is normally "City, ST" / "City, Country"; campus pages ("Yorkville Technology Campus") name the place in the meta description ("… campus in Yorkville, Illinois")
    let cl = parseCityLine(h1);
    const metaDesc = description(doc.text);
    if (!h1.includes(",") || !cl.countryIso2) {
      const inLoc = metaDesc?.match(/\b(?:in|near|located in)\s+([A-Z][A-Za-z .'-]+?),\s*([A-Z][A-Za-z .]+?)(?:[,.;]|\s-|$)/);
      if (inLoc) { const alt = parseCityLine(`${inLoc[1]}, ${inLoc[2]}`); if (alt.countryIso2) cl = alt; }
      if (!h1.includes(",") && cl.city === h1) cl = { ...cl, city: null };
    }
    const slug = lastSlug(url);
    const key = codes.length === 1 ? `cyrusone:${codes[0]}` : `cyrusone:${slug}`;
    const name = codes.length ? `CyrusOne ${codes.join("/")} (${cl.city ?? h1})` : `CyrusOne ${h1}`;

    // Key Features tiles
    let itCapacityMw: number | null = null;
    let techSpace: string | null = null;
    $("h3").each((_, e) => {
      const v = cleanText($(e).text());
      const label = cleanText($(e).nextAll("p").first().text()) ?? cleanText($(e).parent().find("p").first().text());
      if (!v || !label || !/\d/.test(v)) return; // tile values are figures ("54 MW", "18,000 sq. m."); headings like "Key Features:" are not
      if (/IT capacity|critical (?:IT )?(?:load|power)/i.test(label) && itCapacityMw == null) itCapacityMw = parseMw(v);
      if (/technical (?:IT )?space|white ?space|data hall space|raised floor/i.test(label) && !techSpace) techSpace = v;
    });
    const text = cleanText($("main, body").first().text()) ?? "";
    const techSpaceStr = techSpace as string | null; // assigned inside the .each callback (TS narrows the let to null otherwise)
    itCapacityMw ??= mwWithContext(text, /\s*(?:of\s+)?(?:total\s+)?IT\s+capacity/i) ?? mwWithContext(text, /\s*(?:of\s+)?critical\s+(?:IT\s+)?load/i);

    const accordionItems = $("li, span").filter((_, e) => $(e).children().length === 0).map((_, e) => cleanText($(e).text()) ?? "").get().filter((t) => t.length > 2 && t.length < 160);
    const certifications = certificationsFrom(accordionItems.filter((t) => /ISO|SOC|PCI|HIPAA|LEED|BREEAM|Uptime|Tier|NIST|FedRAMP|HITRUST|EN\s?50600/i.test(t)));
    const sustText = accordionItems.filter((t) => /renewable|carbon|green|wind|solar|water|PUE|WUE/i.test(t)).join(". ");
    const h2 = firstText($, "h2");
    const firstP = cleanText($("h2").first().nextAll("div, p").find("p").first().text()) ?? cleanText($("h2").first().next("p").text());
    const desc = description(doc.text);
    const status = statusFromText(`${h1} ${codeLine ?? ""} ${h2 ?? ""} ${firstP ?? ""}`);

    const data: Record<string, unknown> = {
      name, code: codes.length === 1 ? codes[0] : null, aliases: codes.map((c) => `CyrusOne ${c}`),
      city: cl.city, regionName: cl.region, countryIso2: cl.countryIso2,
      itCapacityMw, certifications, renewableClaim: renewableFrom(sustText), coolingType: coolingFrom(`${sustText} ${accordionItems.join(". ")}`),
      status, description: [desc, techSpaceStr ? `Technical IT space: ${techSpaceStr.replace(/\.+$/, "")}.` : null].filter(Boolean).join(" ") || null,
      externalIds: { cyrusone_slug: slug },
    };
    const methods: Record<string, string> = {
      name: "selector:h1+h3", code: "selector:h1~h3", city: "selector:h1+parse", regionName: "selector:h1+parse", countryIso2: "selector:h1+parse>country",
      itCapacityMw: "selector:key-features(h3+p 'IT Capacity')>mw", certifications: "selector:accordion li+regex", renewableClaim: "selector:accordion li+regex", coolingType: "regex:cooling_v1",
      status: "regex:status_v1", description: "meta:description+key-features", externalIds: "url:slug",
    };
    return [record("facility", key, url, data, methods, codes.length ? 0.9 : 0.75)];
  },
};