Merge pull request 'fix(pcat): robust year extraction across key spelling + date format' (#138) from dev into main

Reviewed-on: #138

apps/api/src/vehicles/vehicles.service.spec.ts

@@ -654,19 +654,33 @@ describe("VehiclesService", () => {
       expect(year).toBe(2021);
     });
 
-    it("falls back to production_date when no numeric Year param exists", () => {
+    const extractYear = (parameters: unknown[]) => {
       const { service } = createService();
-      const year = (
+      return (
         service as unknown as {
           extractYearFromPcatCar: (c: { parameters: unknown[] }) => number | null;
         }
-      ).extractYearFromPcatCar({
-        parameters: [
+      ).extractYearFromPcatCar({ parameters });
+    };
+
+    it("falls back to year-first production date (space-keyed)", () => {
+      expect(
+        extractYear([
           { key: "year", name: "Year", value: "All" },
-          { key: "production_date", name: "Production date", value: "2020/10/05" },
-        ],
-      });
-      expect(year).toBe(2020);
+          { key: "production date", name: "Production date", value: "2020/10/05" },
+        ]),
+      ).toBe(2020);
+    });
+
+    it("reads the year from a DAY-first production date (DD/MM/YYYY)", () => {
+      // parseInt("08/03/2016") would have returned 8 (the day) — regex saves us.
+      expect(
+        extractYear([{ key: "production date", name: "Production date", value: "08/03/2016" }]),
+      ).toBe(2016);
+    });
+
+    it("ignores the year=All axis with no concrete year/date", () => {
+      expect(extractYear([{ key: "year", name: "Year", value: "All" }])).toBeNull();
     });
   });
 

apps/api/src/vehicles/vehicles.service.ts

@@ -1186,25 +1186,31 @@ export class VehiclesService {
 
   private extractYearFromPcatCar(car: PcatCar): number | null {
     if (!car.parameters) return null;
-    // pcat returns a faceting AXIS param `year` (often value "All") BEFORE the
-    // concrete per-car `Year` param, so a naive find(includes("year")) grabs the
-    // axis → parseInt("All") = NaN → null. Scan every year-ish param and take the
-    // first that parses to a plausible year; only then fall back to a
-    // production_date like "2020/10/05".
-    const plausible = (raw: string | undefined): number | null => {
-      const num = Number.parseInt(raw ?? "", 10);
-      return num > 1900 && num < 2100 ? num : null;
+    // pcat is messy about year. Three traps:
+    //  1. A faceting AXIS param key="year" value="All" precedes the concrete
+    //     per-car key="Year" value="2021" — a naive find(includes("year")) grabs
+    //     the axis and parseInt("All")=NaN.
+    //  2. The build date lives under key "production date" (space) on most cars,
+    //     "production_date" (underscore) on others.
+    //  3. Those dates come in BOTH "2006/02/27" (year-first) and "08/03/2016"
+    //     (day-first), so parseInt would read the day. Pull a 4-digit 19xx/20xx
+    //     token by regex instead — robust to either ordering.
+    const yearIn = (raw: string | undefined): number | null => {
+      const m = (raw ?? "").match(/\b(19|20)\d{2}\b/);
+      return m ? Number.parseInt(m[0], 10) : null;
     };
+    // Prefer an explicit year param (skips the "All" axis, which has no digits).
     for (const p of car.parameters) {
       const key = p.key.toLowerCase();
       if (key.includes("year") || key.includes("model_year")) {
-        const year = plausible(p.value);
+        const year = yearIn(p.value);
         if (year) return year;
       }
     }
+    // Fall back to the specific car's build date (either key spelling / format).
     for (const p of car.parameters) {
-      if (p.key.toLowerCase().includes("production_date")) {
-        const year = plausible(p.value);
+      if (p.key.toLowerCase().includes("production") && p.key.toLowerCase().includes("date")) {
+        const year = yearIn(p.value);
         if (year) return year;
       }
     }