import type { Parser, ExtractedRecord, RawDocument } from "@dci/connectors";
import { countryFromText, slugify } from "@dci/core";
import { loadHtml, txt, parenthetical, resolvePlace, regionRecord, type Place } from "./shared.js";
import { normKey } from "./city-coords.js";

/**
 * Oracle Cloud Infrastructure — https://www.oracle.com/cloud/public-cloud-regions/
 * Server-rendered HTML (needs a browser identity: the bot UA gets 403). Sections (h2): Commercial,
 * Sovereign (EU Sovereign Cloud), Government, Multicloud (Oracle Database@AWS/Azure/Google — NOT Oracle
 * regions, skipped). Each table: <td geo-data="us-ashburn-1">US East (Ashburn)</td> … <td>Live|Coming soon|Planned</td>.
 * Quirks seen live: a few rows lack geo-data, one code is duplicated ("Spain Central 2"), and two rows have
 * swapped codes (ap-batam-1 / ap-hyderabad-1). The parser trusts the display name and falls back to a slug
 * code when the geo-data code contradicts the city in the name.
 */
const PREFIX_ISO: Array<[RegExp, string]> = [
  [/^(US|United States)\b/i, "US"], [/^Canada\b/i, "CA"], [/^Mexico\b/i, "MX"], [/^Brazil\b/i, "BR"], [/^Chile\b/i, "CL"], [/^Colombia\b/i, "CO"],
  [/^France\b/i, "FR"], [/^Germany\b/i, "DE"], [/^Italy\b/i, "IT"], [/^Netherlands\b/i, "NL"], [/^Serbia\b/i, "RS"], [/^Spain\b/i, "ES"], [/^Sweden\b/i, "SE"], [/^Switzerland\b/i, "CH"], [/^UK\b/i, "GB"],
  [/^Israel\b/i, "IL"], [/^Kenya\b/i, "KE"], [/^Morocco\b/i, "MA"], [/^Saudi Arabia\b/i, "SA"], [/^South Africa\b/i, "ZA"], [/^UAE\b/i, "AE"],
  [/^Australia/i, "AU"], [/^India\b/i, "IN"], [/^Indonesia\b/i, "ID"], [/^Japan\b/i, "JP"], [/^Malaysia\b/i, "MY"], [/^Singapore\b/i, "SG"], [/^South Korea\b/i, "KR"],
];

/** OCI codes legitimately named after a different town than the one displayed (uk-cardiff-1 is "UK West (Newport)"). */
const CODE_CITY_ALIASES: Record<string, string[]> = { "uk-cardiff-1": ["newp"], "sa-vinhedo-1": ["vinh"], "ap-chuncheon-1": ["chun"] };

function statusOf(s: string): "operational" | "announced" {
  return /^live$/i.test(s.trim()) ? "operational" : "announced";
}

export const oracleRegionsParser: Parser = {
  name: "cloud_oracle_regions_v1",
  version: "1.0.0",
  pageTypes: ["cloud_region"],
  parse(doc: RawDocument, ctx): ExtractedRecord[] {
    const $ = loadHtml(doc);
    const out: ExtractedRecord[] = [];
    const seenCodes = new Set<string>();
    // Walk section headers (h2 inside the accordion list) → tables that follow within the same <li>
    $("li").each((_, li) => {
      const section = txt($(li).find("h2").first());
      if (!/^(Commercial|Sovereign|Government)$/i.test(section)) return;
      const realm = section.toLowerCase();
      $(li).find("table").each((__, table) => {
        const area = txt($(table).find("thead th").first());
        $(table).find("tbody tr").each((___, tr) => {
          const tds = $(tr).find("td");
          if (tds.length < 2) return;
          const first = tds.first();
          const name = txt(first);
          if (!name || /^Oracle (AI )?Database@/i.test(name)) return;
          const statusCell = tds.toArray().map((c) => txt($(c))).find((t) => /^(Live|Coming soon|Planned)$/i.test(t)) ?? "";
          const geoCode = (first.attr("geo-data") ?? "").trim();
          const label = parenthetical(name);
          const iso = PREFIX_ISO.find(([re]) => re.test(name))?.[1] ?? (realm === "sovereign" ? null : countryFromText(name));
          let place: Place = resolvePlace(label, iso);
          if (!place.countryIso2) place.countryIso2 = iso ?? (label ? countryFromText(label) : null) ?? countryFromText(name);
          // Code: trust geo-data unless missing, duplicated, or contradicting the named city (site typo)
          let code = geoCode;
          const cityToken = label ? normKey(label).replace(/\s+/g, "").slice(0, 4) : null;
          const codeBody = code.replace(/-/g, "");
          const contradicts = Boolean(code && cityToken && !codeBody.includes(cityToken) && !(CODE_CITY_ALIASES[code] ?? []).includes(cityToken));
          if (!code || seenCodes.has(code) || contradicts) code = slugify(name);
          seenCodes.add(code);
          const externalIds: Record<string, string | number> = { oci_realm: realm, oci_area: area };
          if (geoCode) externalIds.oci_geo_data = geoCode;
          if (label) externalIds.published_location = label;
          const status = statusOf(statusCell);
          out.push(
            regionRecord({
              connectorId: ctx.connectorId,
              keyPrefix: "oracle:region",
              providerName: "Oracle Cloud",
              code,
              name,
              place,
              availabilityZones: null,
              status: realm === "government" && status === "operational" ? "restricted" : status,
              isSovereign: realm === "sovereign",
              externalIds,
              url: doc.finalUrl,
              methods: { code: geoCode && code === geoCode ? "html:td@geo-data" : "slug:name", name: "html:td", status: "html:td.Status", city: "curated-city-table" },
              certainty: code === geoCode ? 0.95 : 0.85,
            }),
          );
        });
      });
    });
    ctx.log("info", `oracle: ${out.length} regions (${out.filter((r) => r.data.status === "announced").length} announced, ${out.filter((r) => r.data.isSovereign).length} sovereign)`);
    return out;
  },
};