spb/datacenterindex

1import { describe, expect, it } from "vitest";2import { hasFilters, interpretQuery } from "./interpret.js";34describe("interpretQuery", () => {5  it("returns empty for blank input", () => {6    expect(interpretQuery("   ")).toEqual({});7    expect(hasFilters(interpretQuery(""))).toBe(false);8  });910  it("extracts a country from a name", () => {11    const i = interpretQuery("data centers in France");12    expect(i.countryIso2).toBe("FR");13    expect(i.text).toBeUndefined();14  });1516  it("extracts a bare ISO2 code in caps", () => {17    expect(interpretQuery("hyperscale US").countryIso2).toBe("US");18    expect(interpretQuery("UK colocation").countryIso2).toBe("GB");19  });2021  it("extracts MW thresholds", () => {22    expect(interpretQuery("over 100 MW").minMw).toBe(100);23    expect(interpretQuery("1.2 GW campus").minMw).toBe(1200);24    expect(interpretQuery("at least 50mw germany")).toMatchObject({ minMw: 50, countryIso2: "DE" });25  });2627  it("extracts maximum MW thresholds", () => {28    expect(interpretQuery("<500 MW")).toMatchObject({ maxMw: 500 });29    expect(interpretQuery("under 500 MW announced")).toMatchObject({ maxMw: 500, status: "announced" });30    expect(interpretQuery("below 1 GW").maxMw).toBe(1000);31    expect(interpretQuery("less than 50mw in france")).toMatchObject({ maxMw: 50, countryIso2: "FR" });32    expect(interpretQuery("less than 50mw").minMw).toBeUndefined();33    expect(interpretQuery(">500 MW announced")).toMatchObject({ minMw: 500, status: "announced" });34    expect(interpretQuery("500+ MW").minMw).toBe(500);35  });3637  it("extracts status and type words", () => {38    expect(interpretQuery("under construction data centers")).toMatchObject({ status: "under_construction" });39    expect(interpretQuery("planned AI data center")).toMatchObject({ status: "announced", facilityType: "ai" });40    expect(interpretQuery("operational colocation in Canada")).toMatchObject({ status: "operational", facilityType: "colocation", countryIso2: "CA" });41  });4243  it("flags AI queries", () => {44    expect(interpretQuery("ai data centers in texas")).toMatchObject({ ai: true, facilityType: "ai" });45    expect(interpretQuery("gpu campus")).toMatchObject({ ai: true });46    expect(interpretQuery("colocation in Paris").ai).toBeUndefined();47  });4849  it("switches to projects when the query says projects", () => {50    const i = interpretQuery("projects over 500 MW in texas");51    expect(i).toMatchObject({ entity: "project", minMw: 500 });52    expect(i.text).toBe("texas");53    expect(interpretQuery("announced data centers").entity).toBeUndefined();54    expect(interpretQuery("planned campus in Ohio").entity).toBeUndefined();55  });5657  it("extracts opening year phrases", () => {58    expect(interpretQuery("projects opening 2028")).toMatchObject({ entity: "project", year: 2028, yearOp: "in" });59    expect(interpretQuery("before 2028")).toMatchObject({ year: 2028, yearOp: "before" });60    expect(interpretQuery("after 2027 hyperscale")).toMatchObject({ year: 2027, yearOp: "after", facilityType: "hyperscale" });61    expect(interpretQuery("opens in 2026").year).toBe(2026);62    expect(interpretQuery("by 2030")).toMatchObject({ year: 2030, yearOp: "before" });63    expect(interpretQuery("500 MW").year).toBeUndefined();64  });6566  it("matches metro names and aliases supplied by the caller", () => {67    const metros = [{ slug: "northern-virginia", name: "Northern Virginia", aliases: ["NoVA", "Ashburn"] }, { slug: "dallas-fort-worth", name: "Dallas-Fort Worth", aliases: ["DFW"] }];68    expect(interpretQuery("hyperscale in Northern Virginia", { metros })).toMatchObject({ metro: "northern-virginia", facilityType: "hyperscale" });69    expect(interpretQuery("ashburn colocation", { metros })).toMatchObject({ metro: "northern-virginia" });70    expect(interpretQuery("DFW over 50 MW", { metros })).toMatchObject({ metro: "dallas-fort-worth", minMw: 50 });71    expect(interpretQuery("virginia", { metros }).metro).toBeUndefined();72  });7374  it("recognises an operator supplied by the caller and keeps the rest as text", () => {75    const i = interpretQuery("Equinix Paris operational", { operators: [{ slug: "equinix", name: "Equinix", similarity: 0.5 }] });76    expect(i.operator).toBe("equinix");77    expect(i.status).toBe("operational");78    expect(i.text).toBe("Paris");79  });8081  it("uses the countries table names when aliases miss", () => {82    const i = interpretQuery("Luxembourg City edge", { countries: [{ iso2: "LU", name: "Luxembourg" }] });83    expect(i.countryIso2).toBe("LU");84    expect(i.facilityType).toBe("edge");85    expect(i.text).toBe("City");86  });8788  it("drops generic words from the remaining text", () => {89    expect(interpretQuery("montreal data center").text).toBe("montreal");90    expect(interpretQuery("the campus of Ashburn").text).toBe("Ashburn");91  });9293  it("combines everything", () => {94    const i = interpretQuery("equinix operational 100 MW in france", { operators: [{ slug: "equinix", name: "Equinix", similarity: 0.4 }] });95    expect(i).toEqual({ minMw: 100, operator: "equinix", countryIso2: "FR", status: "operational" });96    expect(hasFilters(i)).toBe(true);97    expect(hasFilters({ entity: "project" })).toBe(true);98    expect(hasFilters({ year: 2028 })).toBe(true);99  });100});101