import type { Parser, ExtractedRecord, RawDocument } from "@dci/connectors";
import { countryFromText, cleanText } from "@dci/core";
import { loadHtml, tableRows, parenthetical, resolvePlace, regionRecord, type Place } from "./shared.js";

/**
 * Tencent Cloud — https://www.tencentcloud.com/document/product/213/6091 (docs table, server-rendered for the
 * SDK's direct fetcher; curl with a bot UA gets an anti-bot JS page, so `looksBlockedOrEmpty` escalation covers it):
 *   Region | Region ID | Number of AZs | AZ  — the first row of each region carries name/id/count, following rows only AZs.
 * Names look like "North China (Beijing)", "Asia Pacific and the Middle East (Singapore) Coverage: Southeast Asia",
 * "Beijing Finance Available only to finance institutions…", "Shanghai Self-driving Cloud".
 */
export const tencentRegionsParser: Parser = {
  name: "cloud_tencent_regions_v1",
  version: "1.0.0",
  pageTypes: ["cloud_region"],
  parse(doc: RawDocument, ctx): ExtractedRecord[] {
    const $ = loadHtml(doc);
    const out: ExtractedRecord[] = [];
    const seen = new Set<string>();
    for (const row of tableRows($, $("table"))) {
      if (row.length < 3) continue;
      const code = row[1]!.trim();
      if (!/^(ap|na|eu|sa|me|af)-[a-z]+(-[a-z]+)*$/.test(code) || seen.has(code)) continue;
      seen.add(code);
      const raw = row[0]!;
      const restricted = /available only|upon application/i.test(raw);
      const name = cleanText(raw.replace(/\s*Coverage:.*$/i, "").replace(/\s*Available only.*$/i, "")) ?? raw;
      const azs = Number.parseInt(row[2]!.replace(/[^\d]/g, ""), 10);
      // "Hong Kong (China), Macao (China), and Taiwan (China)" → drop the "(China)" qualifiers before reading the city
      const bare = name.replace(/\s*\(China\)/g, "");
      const label = parenthetical(bare)?.trim() ?? bare.replace(/\s+(Finance|Self-driving Cloud).*$/i, "").split(/,|\band\b/)[0]!.trim();
      const chinaRegion = /^(North|East|South|Southwest|Northwest|Central) China\b/i.test(name) || /Finance|Self-driving/i.test(name) || /^cn|^ap-(beijing|shanghai|nanjing|guangzhou|chengdu|chongqing|zhongwei|shenzhen|hongkong)/.test(code);
      const isoHint = /hong kong/i.test(label) ? "HK" : chinaRegion ? "CN" : null;
      let place: Place = resolvePlace(label, isoHint);
      if (!place.countryIso2) place.countryIso2 = isoHint ?? countryFromText(label) ?? countryFromText(name);
      const externalIds: Record<string, string | number> = { tencent_region_id: code, published_location: label };
      const coverage = raw.match(/Coverage:\s*(.+)$/i)?.[1];
      if (coverage) externalIds.coverage = cleanText(coverage) ?? coverage;
      if (/Finance/i.test(name)) externalIds.realm = "finance";
      if (/Self-driving/i.test(name)) externalIds.realm = "self_driving_cloud";
      out.push(
        regionRecord({
          connectorId: ctx.connectorId,
          keyPrefix: "tencent:region",
          providerName: "Tencent Cloud",
          code,
          name,
          place,
          availabilityZones: Number.isFinite(azs) && azs > 0 ? azs : null,
          status: restricted || /Self-driving/i.test(name) ? "restricted" : "operational",
          externalIds,
          url: doc.finalUrl,
          methods: { code: "html:table.RegionID", name: "html:table.Region", availabilityZones: "html:table.NumberOfAZs", city: "curated-city-table" },
        }),
      );
    }
    ctx.log("info", `tencent: ${out.length} regions`);
    return out;
  },
};