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: "Switzerland (all)", keywords: KEYWORDS, location: "Switzerland", workType: ["onsite", "hybrid", "remote"], timePosted: "month" },
  { label: "Zürich Area", keywords: KEYWORDS, geoId: "90009888", location: "Zürich Area", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Basel Metro", keywords: KEYWORDS, geoId: "90009885", location: "Basel Metropolitan Area", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Bern", keywords: KEYWORDS, location: "Bern, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Geneva", keywords: KEYWORDS, location: "Geneva, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Lausanne", keywords: KEYWORDS, location: "Lausanne, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Zug", keywords: KEYWORDS, location: "Zug, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Lucerne", keywords: KEYWORDS, location: "Lucerne, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "St. Gallen", keywords: KEYWORDS, location: "St. Gallen, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
  { label: "Lugano", keywords: KEYWORDS, location: "Lugano, Switzerland", workType: ["onsite", "hybrid"], timePosted: "month" },
];

const seen = new Map<string, any>();
for (const s of searches) {
  const jobs = await fetchJobs(s);
  for (const j of jobs) if (!seen.has(j.id) && !existingIds.has(j.id)) seen.set(j.id, j);
  console.log(`${s.label}: ${jobs.length} raw, ${seen.size} unique new so far`);
}

const MS_RE = /\b(\.net|c#|azure|sharepoint|dynamics|power\s?(bi|platform|apps)|sccm|intune|windows)\b/i;
const candidates = [...seen.values()].filter((j) => !MS_RE.test(j.title));

console.log(`\n${candidates.length} candidates. Checking English content…\n`);
const results = [];
for (const j of candidates) {
  const ok = await isEnglishContent(j.id);
  if (ok) results.push(j);
  else console.log(`  dropped (non-English): ${j.title}`);
}

console.log(`\n=== ${results.length} NEW SWISS JOBS ===\n`);
for (const j of results) {
  console.log(`${j.title} | ${j.company} | ${j.location} | ${j.postedAt}`);
  console.log(`  ${j.link}`);
}