import { KEYWORDS } from "../config.ts";
import type { SearchConfig } 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 searches: SearchConfig[] = [
  {
    label: "Zürich (city)",
    keywords: KEYWORDS,
    geoId: "102436504",
    location: "Zürich, Switzerland",
    workType: ["onsite", "hybrid"],
    timePosted: "month",
  },
  {
    label: "Zürich Metropolitan Area",
    keywords: KEYWORDS,
    geoId: "107814425",
    location: "Zürich Metropolitan Area",
    workType: ["onsite", "hybrid"],
    timePosted: "month",
  },
  {
    label: "Zürich Area",
    keywords: KEYWORDS,
    geoId: "90009888",
    location: "Zürich Area",
    workType: ["onsite", "hybrid"],
    timePosted: "month",
  },
];

const seen = new Set<string>();
const allJobs = [];
for (const cfg of searches) {
  const jobs = await fetchJobs(cfg);
  for (const j of jobs) {
    if (seen.has(j.id) || existingIds.has(j.id)) continue;
    seen.add(j.id);
    allJobs.push(j);
  }
}

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

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