Jobber/extractors/careerspages/manifest.ts

/**
 * Direct company career pages — auto-detect ATS backends and pull public job feeds.
 */

import type {
  ExtractorManifest,
  ExtractorRunResult,
} from "@shared/types/extractors";
import type { CreateJobInput } from "@shared/types/jobs";
import {
  detectAtsTargets,
  fetchCareersPage,
  fetchJobsForTarget,
  readUrlList,
} from "./src/resolve-ats.js";

function matchesTerm(job: CreateJobInput, term: string): boolean {
  const lower = term.toLowerCase();
  if (job.title.toLowerCase().includes(lower)) return true;
  if (job.employer.toLowerCase().includes(lower)) return true;
  if (job.jobDescription?.toLowerCase().includes(lower)) return true;
  return false;
}

export const manifest: ExtractorManifest = {
  id: "careerspages",
  displayName: "Career Pages",
  providesSources: ["careerspages"],
  async run(context): Promise<ExtractorRunResult> {
    if (context.shouldCancel?.()) return { success: true, jobs: [] };

    const pageUrls = readUrlList(context.settings.careersPageUrls);
    if (pageUrls.length === 0) {
      return {
        success: true,
        jobs: [],
        error:
          "No career page URLs configured. Set CAREERS_PAGE_URLS or careersPageUrls (JSON array or comma/newline-separated URLs).",
      };
    }

    const terms = context.searchTerms.length > 0 ? context.searchTerms : [];
    const seen = new Set<string>();
    const out: CreateJobInput[] = [];

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

        context.onProgress?.({
          phase: "list",
          termsProcessed: i,
          termsTotal: pageUrls.length,
          currentUrl: pageUrl,
          detail: `Career pages: resolving ATS for ${pageUrl}`,
        });

        const html = await fetchCareersPage(pageUrl);
        const targets = detectAtsTargets(html, pageUrl);
        if (targets.length === 0) continue;

        for (const target of targets) {
          if (context.shouldCancel?.()) break;
          const jobs = await fetchJobsForTarget(target);
          for (const job of jobs) {
            if (
              terms.length > 0 &&
              !terms.some((term) => matchesTerm(job, term))
            ) {
              continue;
            }
            const key = job.jobUrl;
            if (seen.has(key)) continue;
            seen.add(key);
            out.push(job);
          }
        }
      }

      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;