sase.tr/apps/api/src/integrations/carcatonline/carcatonline.matcher.ts

/**
 * Map our PL24 `catalog_vehicles` (brand + model label) onto carcatonline
 * catalogs/models. carcatonline's `pl_*` catalogs mirror PartsLink24, so most
 * model labels are identical (measured 2026-09-25 on prod's empty catalogs:
 * Hyundai 226/226, Kia 145/145, Volvo 59/59, Mitsubishi 56/56, Toyota 39/39,
 * Alfa 23/23 exact; Renault 45/65, VW 38/58, Porsche 26/64 exact). The rest
 * differ by region suffixes ("ALTIMA (EL)" vs "ALTIMA"), year ranges
 * ("B-MAX (2012- 2017)" vs "B-MAX"), or body-code-only labels (Mercedes "C107"
 * vs "Mercedes-Benz R107 / C107 …"), handled by the relaxed / code tiers.
 */
import type { CarcatModel } from "./carcatonline.client";

/** Ordered catalog candidates per brand; the first one with a model match wins. */
export const BRAND_CATALOGS: Record<string, string[]> = {
  abarth: ["pl_abarth"],
  "alfa romeo": ["pl_alfa_romeo"],
  alpine: ["pl_alpine"],
  audi: ["pl_audi"],
  bentley: ["pl_bentley"],
  bmw: ["pl_bmw"],
  citroen: ["pl_citroen"],
  citroën: ["pl_citroen"],
  cupra: ["pl_cupra", "pl_seat"],
  dacia: ["pl_dacia", "pl_renault"],
  ds: ["pl_ds"],
  fiat: ["pl_fiat", "pl_fiat_professional"],
  ford: ["pl_ford", "pl_ford_pro"],
  hyundai: ["pl_hyundai"],
  infiniti: ["pl_infiniti", "pl_nissan"],
  iveco: ["pl_iveco"],
  jaguar: ["pl_jaguar"],
  jeep: ["pl_jeep"],
  kia: ["pl_kia"],
  lancia: ["pl_lancia"],
  "land rover": ["pl_land_rover"],
  lexus: ["pl_lexus", "pl_toyota"],
  man: ["pl_man"],
  "mercedes-benz": ["pl_mercedes", "pl_mercedes_vans", "pl_mercedes_truck"],
  mercedes: ["pl_mercedes", "pl_mercedes_vans", "pl_mercedes_truck"],
  mini: ["pl_mini"],
  mitsubishi: ["pl_mitsubishi"],
  nissan: ["pl_nissan"],
  opel: ["pl_opel", "pl_psa_opel", "pl_vauxhall"],
  peugeot: ["pl_peugeot"],
  polestar: ["pl_polestar"],
  porsche: ["pl_porsche", "pl_porsche_classic"],
  renault: ["pl_renault"],
  seat: ["pl_seat", "pl_cupra"],
  skoda: ["pl_skoda"],
  smart: ["pl_smart"],
  suzuki: ["pl_suzuki"],
  toyota: ["pl_toyota"],
  vauxhall: ["pl_vauxhall", "pl_psa_vauxhall"],
  volkswagen: ["pl_volkswagen", "pl_volkswagen_commercial", "pl_volkswagen_classic"],
  volvo: ["pl_volvo"],
};

export function catalogsForBrand(brandName: string | null | undefined): string[] {
  const key = (brandName ?? "").trim().toLowerCase();
  return BRAND_CATALOGS[key] ?? [];
}

/** PL24 pseudo-models that are not vehicles (VW group "Kimyasal maddeler", "Özel kataloglar" …). */
const NON_VEHICLE_LABELS = new Set<string>([
  "KIMYASAL MADDELER",
  "OZEL KATALOGLAR",
  "DIGER URUNLER",
  "URETICI SAYFALARI",
  "KISALTMALAR",
  "AOS",
  "ELEKTRIKL BAGLANTI",
  "KAMP ARACI",
  "KAMYONET",
  "KOMB ISI GUC SANTR",
]);

export function isNonVehicleModel(label: string): boolean {
  return NON_VEHICLE_LABELS.has(normalizeLabel(label));
}

/** Uppercase, diacritic-folded, punctuation collapsed, PL24 backslash escapes removed. */
export function normalizeLabel(s: string | null | undefined): string {
  return (s ?? "")
    .toUpperCase()
    .normalize("NFD")
    .replace(/\p{M}/gu, "")
    .replace(/\\/g, "")
    .replace(/[^A-Z0-9]+/g, " ")
    .trim();
}

/** Drop region suffixes "(EL)/(ER)", year ranges "(2012- 2017)", "2004 - 2014", and 4-digit years. */
export function relaxLabel(s: string | null | undefined): string {
  const stripped = (s ?? "")
    .replace(/\((EL|ER|EUR|JPN|USA|GCC|RUS)\)/gi, " ")
    .replace(/\((19|20)\d\d\s*-\s*(19|20)?\d{0,2}\)/g, " ")
    .replace(/\b(19|20)\d\d\s*-\s*(19|20)?\d{0,2}\b/g, " ")
    .replace(/\b(19|20)\d\d\b/g, " ");
  return normalizeLabel(stripped);
}

export interface ModelMatch {
  model: CarcatModel;
  tier: "exact" | "relaxed" | "code";
}

/**
 * Pick the carcatonline model for one of our PL24 model labels.
 *  - exact: normalized labels identical
 *  - relaxed: identical after stripping region/year decorations on both sides
 *  - code: our label is a short body/chassis code (≤6 chars, has a digit) that
 *    appears as a whole token in exactly ONE carcatonline label (Mercedes C107)
 * Returns null when ambiguous or absent — the caller records "unmatched".
 */
export function matchCarcatModel(ourLabel: string, models: CarcatModel[]): ModelMatch | null {
  const target = normalizeLabel(ourLabel);
  if (!target) return null;
  const exact = models.filter((m) => normalizeLabel(m.name) === target);
  if (exact.length === 1) return { model: exact[0], tier: "exact" };
  if (exact.length > 1) return { model: exact[0], tier: "exact" }; // identical labels → same PL24 model, first wins

  const relaxedTarget = relaxLabel(ourLabel);
  if (relaxedTarget) {
    const relaxed = models.filter((m) => relaxLabel(m.name) === relaxedTarget);
    if (relaxed.length === 1) return { model: relaxed[0], tier: "relaxed" };
  }

  if (/^[A-Z]{1,2}\d{2,4}[A-Z]?$/.test(target) && target.length <= 6) {
    const code = models.filter((m) => normalizeLabel(m.name).split(" ").includes(target));
    if (code.length === 1) return { model: code[0], tier: "code" };
  }
  return null;
}