Jobber/extractors/bctenet/manifest.ts

/**
 * BC T-Net — public RSS aggregate of BC tech jobs.
 *
 * Default: https://www.bctechnology.com/rss/jobs/tnetjobs.xml
 *
 * Feeds may embed `<![CDATA[&]]>` inside `<link>` URLs — normalized before fetch.
 */

import type {
  ExtractorManifest,
  ExtractorRunResult,
} from "@shared/types/extractors";
import type { CreateJobInput } from "@shared/types/jobs";

interface BcItem {
  title?: string;
  link?: string;
  guid?: string;
  description?: string;
  pubDate?: string;
  category?: string;
}

function xmlText(xml: string, tag: string): string | undefined {
  const pattern = new RegExp(`<${tag}[^>]*>([\\s\\S]*?)</${tag}>`);
  const match = xml.match(pattern);
  if (!match?.[1]) return undefined;
  return (
    match[1].replace(/<!\[CDATA\[([\s\S]*?)\]\]>/g, "$1").trim() || undefined
  );
}

function normalizeFeedLink(raw: string): string {
  return raw.replace(/<!\[CDATA\[&\]\]>/g, "&").trim();
}

function parseItems(xml: string): BcItem[] {
  const items: BcItem[] = [];
  const blocks = xml.match(/<item>([\s\S]*?)<\/item>/g) ?? [];

  for (const raw of blocks) {
    const block = raw.replace(/^<item>/, "").replace(/<\/item>$/, "");
    const linkRaw = xmlText(block, "link");
    items.push({
      title: xmlText(block, "title"),
      link: linkRaw ? normalizeFeedLink(linkRaw) : undefined,
      guid: xmlText(block, "guid"),
      description: xmlText(block, "description"),
      pubDate: xmlText(block, "pubDate"),
      category: xmlText(block, "category"),
    });
  }

  return items;
}

function readUrls(raw: string | undefined): string[] {
  if (!raw) return [];
  try {
    const parsed = JSON.parse(raw);
    if (Array.isArray(parsed)) {
      return parsed
        .map((entry) => (typeof entry === "string" ? entry.trim() : ""))
        .filter(Boolean);
    }
  } catch {
    // fall through
  }
  return raw
    .split(/[\n|]+/)
    .map((entry) => entry.trim())
    .filter(Boolean);
}

function decodeHtmlEntities(html: string): string {
  return html
    .replace(/&amp;/g, "&")
    .replace(/&lt;/g, "<")
    .replace(/&gt;/g, ">")
    .replace(/&#x2f;/gi, "/")
    .replace(/&#x26;/gi, "&");
}

function matchesTerm(item: BcItem, term: string): boolean {
  const lower = term.toLowerCase();
  const hay =
    `${item.title ?? ""} ${item.description ?? ""} ${item.category ?? ""}`.toLowerCase();
  return hay.includes(lower);
}

function mapJob(item: BcItem): CreateJobInput | null {
  const jobUrl = item.link?.trim();
  if (!jobUrl) return null;

  const title = item.title ? decodeHtmlEntities(item.title) : "Unknown Title";
  const employer = item.category?.trim() || "Unknown Employer";

  return {
    source: "bctenet",
    sourceJobId: item.guid ?? jobUrl,
    title,
    employer,
    jobUrl,
    applicationLink: jobUrl,
    location: "British Columbia, Canada",
    datePosted: item.pubDate,
    jobDescription: item.description
      ? decodeHtmlEntities(item.description)
      : undefined,
  };
}

export const manifest: ExtractorManifest = {
  id: "bctenet",
  displayName: "BC T-Net (RSS)",
  providesSources: ["bctenet"],
  async run(context): Promise<ExtractorRunResult> {
    if (context.shouldCancel?.()) return { success: true, jobs: [] };

    const defaults = ["https://www.bctechnology.com/rss/jobs/tnetjobs.xml"];
    const configured = readUrls(context.settings.bctenetRssUrls);
    const urls = configured.length > 0 ? configured : defaults;

    const maxJobs = context.settings.bctenetMaxJobsPerTerm
      ? Number.parseInt(context.settings.bctenetMaxJobsPerTerm, 10)
      : 400;
    const cap = Number.isFinite(maxJobs)
      ? Math.min(Math.max(maxJobs, 1), 2000)
      : 400;

    const terms = context.searchTerms.length > 0 ? context.searchTerms : [];
    const maxTotal = cap * Math.max(terms.length, 1);
    const seen = new Set<string>();
    const out: CreateJobInput[] = [];

    try {
      for (let i = 0; i < urls.length; i += 1) {
        if (context.shouldCancel?.()) break;
        const rssUrl = urls[i];

        context.onProgress?.({
          phase: "list",
          termsProcessed: i,
          termsTotal: urls.length,
          currentUrl: rssUrl,
          detail: `BC T-Net: fetching (${i + 1}/${urls.length})`,
        });

        const response = await fetch(rssUrl, {
          headers: {
            Accept: "application/rss+xml, application/xml, text/xml",
            "User-Agent":
              "Mozilla/5.0 (compatible; JobOps/1.0) BC T-Net RSS consumer",
          },
        });
        if (!response.ok) {
          throw new Error(`BC T-Net RSS failed: ${response.status}`);
        }

        const xml = await response.text();
        const items = parseItems(xml);

        for (const item of items) {
          if (out.length >= maxTotal) break;
          if (terms.length > 0 && !terms.some((t) => matchesTerm(item, t))) {
            continue;
          }
          const mapped = mapJob(item);
          if (!mapped) continue;
          const key = mapped.sourceJobId || mapped.jobUrl;
          if (seen.has(key)) continue;
          seen.add(key);
          out.push(mapped);
        }

        context.onProgress?.({
          phase: "list",
          termsProcessed: i + 1,
          termsTotal: urls.length,
          currentUrl: rssUrl,
          jobPagesProcessed: out.length,
          detail: `BC T-Net: ${items.length} items (${out.length} kept total)`,
        });
      }

      return { success: true, jobs: out };
    } catch (error) {
      const message = error instanceof Error ? error.message : "Unknown error";
      return { success: false, jobs: out, error: message };
    }
  },
};

export default manifest;