Add dictionaries transform with filtering and partitioning rules

2026-04-17 03:13:18 +03:00
parent dc373553d2
commit e5b49acecf
2 changed files with 232 additions and 0 deletions
@@ -0,0 +1,96 @@
+import { describe, it, expect } from "vitest";
+import { transformDictionaries } from "./transform.js";
+import type { IRawDictionaries } from "./types.js";
+
+function makeRaw(overrides: Partial<IRawDictionaries> = {}): IRawDictionaries {
+  return {
+    regions: [],
+    countries: [],
+    cities: [],
+    airports: [],
+    ...overrides,
+  };
+}
+
+describe("transformDictionaries — filtering", () => {
+  it("drops cities whose title.ru is purely ASCII (garbage-data guard)", () => {
+    const raw = makeRaw({
+      cities: [
+        { code: "MOW", title: { ru: "Москва", en: "Moscow" }, country_code: "RU", has_afl_flights: true, location: { lat: 55, lon: 37 } },
+        { code: "BAD", title: { ru: "Moscow", en: "Moscow" }, country_code: "RU", has_afl_flights: true, location: { lat: 0, lon: 0 } },
+      ],
+      airports: [
+        { code: "SVO", city_code: "MOW", title: { ru: "Шереметьево", en: "SVO" }, has_afl_flights: true, location: { lat: 55, lon: 37 } },
+      ],
+    });
+
+    const d = transformDictionaries(raw, "ru");
+
+    expect(d.cities.map((c) => c.code)).toEqual(["MOW"]);
+  });
+
+  it("drops airports with has_afl_flights=false", () => {
+    const raw = makeRaw({
+      cities: [
+        { code: "MOW", title: { ru: "Москва" }, country_code: "RU", has_afl_flights: true, location: { lat: 55, lon: 37 } },
+      ],
+      airports: [
+        { code: "SVO", city_code: "MOW", title: { ru: "Шереметьево" }, has_afl_flights: true, location: { lat: 55, lon: 37 } },
+        { code: "DME", city_code: "MOW", title: { ru: "Домодедово" }, has_afl_flights: false, location: { lat: 55, lon: 37 } },
+      ],
+    });
+
+    const d = transformDictionaries(raw, "ru");
+
+    expect(d.airports.map((a) => a.code)).toEqual(["SVO"]);
+  });
+
+  it("drops airports whose title.ru is purely ASCII", () => {
+    const raw = makeRaw({
+      cities: [
+        { code: "MOW", title: { ru: "Москва" }, country_code: "RU", has_afl_flights: true, location: { lat: 55, lon: 37 } },
+      ],
+      airports: [
+        { code: "SVO", city_code: "MOW", title: { ru: "SVO" }, has_afl_flights: true, location: { lat: 55, lon: 37 } },
+      ],
+    });
+
+    const d = transformDictionaries(raw, "ru");
+
+    expect(d.airports).toHaveLength(0);
+  });
+
+  it("drops cities that end up with no afl airports", () => {
+    const raw = makeRaw({
+      cities: [
+        { code: "MOW", title: { ru: "Москва" }, country_code: "RU", has_afl_flights: true, location: { lat: 55, lon: 37 } },
+        { code: "NOW", title: { ru: "Никогорск" }, country_code: "RU", has_afl_flights: true, location: { lat: 0, lon: 0 } },
+      ],
+      airports: [
+        { code: "SVO", city_code: "MOW", title: { ru: "Шереметьево" }, has_afl_flights: true, location: { lat: 55, lon: 37 } },
+      ],
+    });
+
+    const d = transformDictionaries(raw, "ru");
+
+    expect(d.cities.map((c) => c.code)).toEqual(["MOW"]);
+  });
+
+  it("partitions city codes into ruCityCodes and otherCityCodes", () => {
+    const raw = makeRaw({
+      cities: [
+        { code: "MOW", title: { ru: "Москва" }, country_code: "RU", has_afl_flights: true, location: { lat: 55, lon: 37 } },
+        { code: "PAR", title: { ru: "Париж" }, country_code: "FR", has_afl_flights: true, location: { lat: 48, lon: 2 } },
+      ],
+      airports: [
+        { code: "SVO", city_code: "MOW", title: { ru: "Шереметьево" }, has_afl_flights: true, location: { lat: 55, lon: 37 } },
+        { code: "CDG", city_code: "PAR", title: { ru: "Шарль-де-Голль" }, has_afl_flights: true, location: { lat: 48, lon: 2 } },
+      ],
+    });
+
+    const d = transformDictionaries(raw, "ru");
+
+    expect([...d.ruCityCodes]).toEqual(["MOW"]);
+    expect([...d.otherCityCodes]).toEqual(["PAR"]);
+  });
+});
@@ -0,0 +1,136 @@
+/**
+ * Pure transform from raw dictionary responses to consumer-facing shapes.
+ *
+ * Ports the rules from Angular's DictionariesService.handleLoading:
+ * - drop cities whose title.ru is purely ASCII (garbage-data guard)
+ * - drop airports without afl flights or with ASCII title.ru
+ * - attach airports to their city (sorted by localized title)
+ * - drop cities that end up with no afl airports
+ * - enrich city.name, city.countryName from localized titles
+ * - build lookup maps keyed by uppercase code
+ * - partition cities into RU vs non-RU sets
+ * - flatten regions (Russia first, Australia filtered out)
+ */
+
+import type {
+  IAirport,
+  ICity,
+  ICountry,
+  IDictionaries,
+  IRegion,
+  IRawAirport,
+  IRawDictionaries,
+  IRawRegion,
+} from "./types.js";
+
+const ASCII_ONLY = /^[a-zA-Z.,:; ]+$/;
+const AUSTRALIA_REGION_ID = 500373;
+const RUSSIA_REGION_ID = 500374;
+
+export function transformDictionaries(
+  raw: IRawDictionaries,
+  lang: string,
+): IDictionaries {
+  const citiesAfterAsciiFilter = raw.cities.filter(
+    (c) => !ASCII_ONLY.test(c.title["ru"] ?? ""),
+  );
+
+  const airportsFiltered = raw.airports.filter(
+    (a) =>
+      a.has_afl_flights === true &&
+      !ASCII_ONLY.test(a.title["ru"] ?? ""),
+  );
+
+  const countriesByCode = new Map(
+    raw.countries.map((c) => [c.code, c]),
+  );
+
+  const citiesEnriched: ICity[] = citiesAfterAsciiFilter.map((c) => {
+    const airports = airportsFiltered
+      .filter((a) => a.city_code === c.code)
+      .map((a) => shapeAirport(a, lang))
+      .sort((a, b) => a.name.localeCompare(b.name));
+
+    const country = countriesByCode.get(c.country_code);
+
+    return {
+      code: c.code,
+      name: c.title[lang] ?? c.code,
+      location: c.location ?? { lat: 0, lon: 0 },
+      country_code: c.country_code,
+      countryName: country?.title[lang] ?? "",
+      has_afl_flights: airports.length > 0,
+      airports,
+    };
+  });
+
+  const cities = citiesEnriched.filter((c) => c.has_afl_flights);
+
+  const airports = cities.flatMap((c) => c.airports);
+
+  const cityByCode = new Map<string, ICity>(
+    cities.map((c) => [c.code.toUpperCase(), c]),
+  );
+  const airportByCode = new Map<string, IAirport>(
+    airports.map((a) => [a.code.toUpperCase(), a]),
+  );
+
+  const ruCityCodes = new Set<string>();
+  const otherCityCodes = new Set<string>();
+  for (const c of cities) {
+    if (c.country_code === "RU") ruCityCodes.add(c.code);
+    else otherCityCodes.add(c.code);
+  }
+
+  const countries: ICountry[] = raw.countries.map((c) => ({
+    code: c.code,
+    name: c.title[lang] ?? c.code,
+    world_region_id: c.world_region_id,
+  }));
+
+  const regions = buildRegions(raw.regions, countries, lang);
+
+  return {
+    regions,
+    countries,
+    cities,
+    airports,
+    cityByCode,
+    airportByCode,
+    ruCityCodes,
+    otherCityCodes,
+  };
+}
+
+function shapeAirport(a: IRawAirport, lang: string): IAirport {
+  return {
+    code: a.code,
+    name: a.title[lang] ?? a.code,
+    city_code: a.city_code,
+    location: a.location ?? { lat: 0, lon: 0 },
+    has_afl_flights: a.has_afl_flights,
+  };
+}
+
+function buildRegions(
+  raw: IRawRegion[],
+  countries: ICountry[],
+  lang: string,
+): IRegion[] {
+  const filtered = raw.filter((r) => r.world_region_id !== AUSTRALIA_REGION_ID);
+
+  const sorted = [...filtered].sort((a, b) =>
+    (a.title[lang] ?? "").localeCompare(b.title[lang] ?? ""),
+  );
+  const ruIdx = sorted.findIndex((r) => r.world_region_id === RUSSIA_REGION_ID);
+  if (ruIdx > 0) {
+    const [ru] = sorted.splice(ruIdx, 1);
+    if (ru) sorted.unshift(ru);
+  }
+
+  return sorted.map((r) => ({
+    id: r.world_region_id,
+    name: r.title[lang] ?? "",
+    countries: countries.filter((c) => c.world_region_id === r.world_region_id),
+  }));
+}