watermaps/packages/shared/tests/marine-poi-clustering.test.ts

import { describe, expect, it } from "vitest";
import {
  canonicalizeMarinePois,
  normalizedMarineFacilityName,
  type MarinePoiCandidate
} from "../src/marine-poi-clustering.js";

const lock = (
  id: string,
  name: string | null,
  lon: number,
  properties: Record<string, unknown> = {},
  overrides: Partial<MarinePoiCandidate> = {}
): MarinePoiCandidate => ({
  id,
  layer: "locks",
  source: "osm",
  sourceId: id,
  name,
  coordinate: { lon, lat: 51.695 },
  properties,
  ...overrides
});

describe("marine POI canonicalization", () => {
  it("normalizes facility type words but retains facility ordinals", () => {
    expect(normalizedMarineFacilityName("Schleuse Werries")).toBe("werries");
    expect(normalizedMarineFacilityName("Schleusengebiet Werries")).toBe("werries");
    expect(normalizedMarineFacilityName("Werries Lock")).toBe("werries");
    expect(normalizedMarineFacilityName("W.S.V. Sixhaven")).toBe("sixhaven");
    expect(normalizedMarineFacilityName("Große Kammer Schleuse Ahsen")).toBe("ahsen");
    expect(normalizedMarineFacilityName("Schleuse Ahsen Kammer 2")).toBe("ahsen");
    expect(normalizedMarineFacilityName("Schleuse 55")).toBe("55");
    expect(normalizedMarineFacilityName("Außentor")).toBeNull();
  });

  it("merges matching OSM and EuRIS locks and combines their contacts", () => {
    const result = canonicalizeMarinePois([
      lock("w-osm", "Schleuse Werries", 7.867, {
        website: "https://osm.example/werries",
        phone: "+49 2381 9019290"
      }),
      lock(
        "w-euris",
        "Werries",
        7.86708,
        { "ref:EU:RIS": "DEHMM00301LOCKS00404", phone: "+49 2381 9019-290", vhf: "22" },
        { source: "euris", sourceId: "DEHMM00301LOCKS00404" }
      )
    ]);

    expect(result).toHaveLength(1);
    expect(result[0]).toMatchObject({
      entityId: "marine-poi:locks:euris:DEHMM00301LOCKS00404",
      canonicalSource: "euris",
      memberCount: 2,
      properties: {
        phone: "+49 2381 9019-290",
        website: "https://osm.example/werries",
        vhf: "22",
        source: "EuRIS + OpenStreetMap"
      }
    });
  });

  it("does not merge neighbouring facilities with different specific names", () => {
    const result = canonicalizeMarinePois([
      lock("a", "Wilhelminasluis", 4.724),
      lock("b", "Hondsbossche Sluis", 4.72418)
    ]);
    expect(result).toHaveLength(2);
  });

  it("merges duplicate facilities with the exact same name despite differing source websites", () => {
    const result = canonicalizeMarinePois([
      lock("relation", "Kesselschleuse Emden", 7.2, { website: "https://city.example/lock" }),
      lock("way", "Kesselschleuse Emden", 7.20008, { website: "https://operator.example/lock" })
    ]);

    expect(result).toHaveLength(1);
    expect(result[0]?.properties.websites).toEqual([
      "https://city.example/lock",
      "https://operator.example/lock"
    ]);
  });

  it("does not merge equal names carrying conflicting official IDs", () => {
    const result = canonicalizeMarinePois([
      lock("a", "Schleuse 55", 7.1, { "ref:EU:RIS": "DE-55" }),
      lock("b", "Schleuse 55", 7.1001, { "ref:EU:RIS": "DE-56" })
    ]);
    expect(result).toHaveLength(2);
  });

  it("assigns an unnamed gate to exactly the nearest anchor without bridging facilities", () => {
    const result = canonicalizeMarinePois([
      lock("west", "Schleuse West", 7),
      lock("east", "Schleuse Ost", 7.0018),
      lock("gate", "Außentor", 7.0002, { waterway: "lock_gate" })
    ]);
    expect(result).toHaveLength(2);
    expect(result.find((poi) => poi.name === "Schleuse West")?.memberIds).toContain("gate");
    expect(result.find((poi) => poi.name === "Schleuse Ost")?.memberIds).not.toContain("gate");
  });

  it("groups a named harbour anchor with its dock component", () => {
    const result = canonicalizeMarinePois([
      {
        id: "marina",
        layer: "harbours",
        source: "osm",
        sourceId: "w1",
        name: "Stadthafen",
        coordinate: { lon: 7.7, lat: 52 },
        properties: { leisure: "marina" }
      },
      {
        id: "dock",
        layer: "harbours",
        source: "osm",
        sourceId: "w2",
        name: "Stadthafen",
        coordinate: { lon: 7.7004, lat: 52 },
        properties: { waterway: "dock" }
      }
    ]);
    expect(result).toHaveLength(1);
    expect(result[0]?.memberCount).toBe(2);
  });

  it("is independent of input order and adding a gate keeps the anchor ID stable", () => {
    const anchor = lock("anchor", "Kesselschleuse Emden", 7.2, { lock: "yes" });
    const gate = lock("gate", null, 7.2001, { waterway: "lock_gate" });
    const forward = canonicalizeMarinePois([anchor, gate]);
    const reverse = canonicalizeMarinePois([gate, anchor]);
    const anchorOnly = canonicalizeMarinePois([anchor]);
    expect(forward).toEqual(reverse);
    expect(forward[0]?.entityId).toBe(anchorOnly[0]?.entityId);
  });

  it("keeps same-named facilities apart beyond the layer radius", () => {
    const result = canonicalizeMarinePois([
      lock("one", "Keersluis", 5),
      lock("two", "Keersluis", 5.02)
    ]);
    expect(result).toHaveLength(2);
  });
});