hire-me/src/match.ts

import type { Match, NormalizedJob } from "./types.ts";

const EU_COUNTRIES = [
  "france",
  "germany",
  "spain",
  "italy",
  "portugal",
  "netherlands",
  "belgium",
  "luxembourg",
  "austria",
  "switzerland",
  "ireland",
  "united kingdom",
  "sweden",
  "norway",
  "denmark",
  "finland",
  "poland",
  "czech",
  "czechia",
  "romania",
  "hungary",
  "greece",
  "croatia",
  "slovenia",
  "slovakia",
  "estonia",
  "latvia",
  "lithuania",
  "bulgaria",
  "europe",
  "european",
  "emea",
];

const AGENCY_DENY = [
  "lemon.io",
  "g2i",
  "distantjob",
  "talent sam",
  "kdci",
  "2am.tech",
  "photon",
  "turing",
  "toptal",
  "andela",
  "telus digital",
  "a.team",
  "white cloak",
  "creative chaos",
  "dacodes",
  "coalition technologies",
  "jobs for humanity",
  "jobsforlebanon",
  "consulting1x1",
];

const JUNIOR = /\b(intern|internship|trainee|stage\b|apprentice|junior|entry-level|entry level|software engineer i\b)\b/i;
const WRONG_ROLE =
  /\b(qa engineer|quality assurance|copywriter|marketing|sales|account executive|customer support|support engineer|devops|data engineer|scraping engineer|documentation engineer|ux designer|head of engineering|game developer)\b/i;
const WRONG_STACK =
  /\b(reactjs|react\.js|\breact\b|next\.js|angular|ruby|rails|django|\.net|asp\.net|spring ?boot|\bjava\b|\bphp\b|laravel|symfony|kotlin|golang|flutter)\b/i;
const VUE = /\bvue(?:\.?js)?\b/i;
const NUXT = /\bnuxt(?:\.?js)?\b/i;

function tokens(location: string): string[] {
  const parts = location
    .toLowerCase()
    .split(/[,/;|]+/)
    .map((part) => part.trim())
    .filter(Boolean);
  if (parts.length === 0 && location.trim()) return [location.toLowerCase().trim()];
  return parts;
}

function isEuToken(token: string): boolean {
  if (token === "uk" || token === "gb" || token === "eu") return true;
  return EU_COUNTRIES.some((country) => token === country || token.startsWith(`${country} `));
}

function geoEligible(job: NormalizedJob): { ok: boolean; reason?: string; bonus: number } {
  const loc = job.location.toLowerCase().trim();
  if (!loc || loc === "worldwide" || loc === "anywhere" || loc === "remote") {
    return { ok: true, bonus: 1, reason: "remote worldwide" };
  }

  const parts = tokens(job.location);
  const euParts = parts.filter(isEuToken);
  if (euParts.length === 0) return { ok: false, bonus: 0 };

  const france = parts.some((part) => part === "france" || part.includes("france"));
  return {
    ok: true,
    bonus: france ? 6 : 3,
    reason: france ? "France" : "Europe",
  };
}

function timezoneOk(job: NormalizedJob): boolean {
  if (job.timezones.length === 0) return true;
  return job.timezones.some((tz) => tz >= 0 && tz <= 3);
}

export function matchJob(job: NormalizedJob): Match | null {
  const company = job.company.toLowerCase();
  if (AGENCY_DENY.some((name) => company.includes(name))) return null;

  const title = job.title;
  if (JUNIOR.test(title) || WRONG_ROLE.test(title)) return null;
  if (/\b(hybrid|présentiel|on-?site|short term|fluent ukrainian)\b/i.test(title)) return null;
  if (!job.remote) return null;
  if (!timezoneOk(job)) return null;

  const geo = geoEligible(job);
  if (!geo.ok) return null;

  const vueTitle = VUE.test(title);
  const nuxtTitle = NUXT.test(title) || /\bsidebase\b/i.test(title);
  const conflictingTitle = WRONG_STACK.test(title) && !vueTitle && !nuxtTitle;

  if (conflictingTitle) return null;
  if (!vueTitle && !nuxtTitle) return null;

  const body = `${job.excerpt} ${job.description}`.toLowerCase().slice(0, 4000);
  const reasons: string[] = [];
  let score = geo.bonus;
  if (geo.reason) reasons.push(geo.reason);

  if (nuxtTitle) {
    score += 8;
    reasons.push("Nuxt");
  }
  if (vueTitle) {
    score += 7;
    reasons.push("Vue");
  }
  if (/\btypescript\b/i.test(`${title} ${job.tags.join(" ")} ${body}`)) {
    score += 2;
    reasons.push("TypeScript");
  }
  if (/\bnode(?:\.?js)?\b/i.test(`${title} ${job.tags.join(" ")} ${body}`)) {
    score += 2;
    reasons.push("Node.js");
  }
  if (/\bfull[\s-]?stack\b/i.test(title)) {
    score += 2;
    reasons.push("Fullstack");
  } else if (/\bfront[\s-]?end\b/i.test(title)) {
    score += 1;
    reasons.push("Frontend");
  }
  if (/\b(travel|booking|ndc|gds|fintech|b2b)\b/i.test(`${title} ${body}`)) {
    score += 2;
    reasons.push("domaine proche");
  }

  if (score < 8) return null;
  return { job, score, reasons };
}