import { KEYWORDS } from "../config.ts";
import { fetchJobs, isEnglishContent } from "../fetch.ts";
import { readFileSync } from "fs";

const csv = readFileSync("./results/jobs.csv", "utf8");
const existingIds = new Set(
  [...csv.matchAll(/jobs\/view\/(\d+)\//g)].map((m) => m[1])
);

const MICROSOFT_RE = /\.net\b|\bc#|sharepoint|dynamics 365|power platform|windows system|\bazure\b|\bintune\b|\bsccm\b/i;

const jobs = await fetchJobs({
  label: "Allschwil",
  keywords: KEYWORDS,
  location: "Allschwil, Switzerland",
  workType: ["onsite", "hybrid"],
  timePosted: "month",
});

const seen = new Set<string>();
const newJobs = jobs.filter((j) => {
  if (seen.has(j.id) || existingIds.has(j.id)) return false;
  seen.add(j.id);
  if (MICROSOFT_RE.test(j.title)) {
    console.log(`  [ms-stack] dropped: ${j.title} — ${j.company}`);
    return false;
  }
  return true;
});

console.log(`\n${newJobs.length} new Allschwil jobs. Checking English content…\n`);

const results = [];
for (const j of newJobs) {
  process.stdout.write(`  ${j.title} — ${j.company} … `);
  const ok = await isEnglishContent(j.id);
  console.log(ok ? "ok" : "dropped (non-English)");
  if (ok) results.push(j);
}

console.log(`\n=== ${results.length} new English Allschwil jobs ===\n`);
for (const j of results) {
  console.log(`${j.title} | ${j.company} | ${j.link}`);
}