import { describe, expect, it } from "vitest";
import { classifyPage, eventTypeFor, isDataCenterRelevant } from "./classify.js";
import { projectMwFigures } from "./mw-figures.js";

describe("classifyPage — URL rules win", () => {
  it("a facility page mentioning an acquisition stays facility_page", () => {
    const c = classifyPage("https://op.com/data-centers/ashburn-va1/", "VA1 Ashburn", "VA1 was acquired in 2019 as part of the acquisition of Acme. 36 MW critical load, N+1, 2N power.");
    expect(c.pageType).toBe("facility_page");
    expect(c.eventType).toBe("facility_updated");
    expect(c.mw).toEqual([36]);
  });
  it("facility index, newsroom index, press release, cloud region, financial, planning by URL", () => {
    expect(classifyPage("https://op.com/data-centers/", null, "").pageType).toBe("facility_index");
    expect(classifyPage("https://op.com/newsroom/", null, "").pageType).toBe("news_index");
    expect(classifyPage("https://op.com/news/2026/big-campus", null, "").pageType).toBe("press_release");
    expect(classifyPage("https://cloud.com/global-infrastructure/regions/", null, "").pageType).toBe("cloud_region");
    expect(classifyPage("https://op.com/investors/annual-report", null, "").pageType).toBe("financial_disclosure");
    expect(classifyPage("https://county.gov/planning/applications/2026-001", null, "").pageType).toBe("planning_document");
    expect(classifyPage("https://op.de/rechenzentren/frankfurt-1/", null, "").pageType).toBe("facility_page");
  });
  it("PDFs are planning documents only with planning vocabulary in URL or title, otherwise unknown", () => {
    expect(classifyPage("https://county.gov/files/rezoning-application-2026.pdf", null, "").pageType).toBe("planning_document");
    expect(classifyPage("https://county.gov/files/ZMAP-2026-0001.pdf", "Planning Commission agenda", "")).toMatchObject({ pageType: "planning_document", rule: "url:pdf+planning" });
    expect(classifyPage("https://op.com/files/brochure.pdf", "Campus brochure", "We announce a new 100 MW data center")).toMatchObject({ pageType: "unknown", rule: "url:pdf" });
    expect(classifyPage("https://op.com/news/release.pdf", null, "").pageType).toBe("unknown"); // the news parser decides
  });
});

describe("classifyPage — text rules", () => {
  it("first matching text rule wins", () => {
    expect(classifyPage("https://x.com/p/1", "Acme completes acquisition of Beta", "").pageType).toBe("acquisition");
    expect(classifyPage("https://x.com/p/2", "Acme expands Dallas campus", "adds 40 MW").pageType).toBe("expansion");
    expect(classifyPage("https://x.com/p/3", "Acme breaks ground in Ohio", "").pageType).toBe("construction_update");
    expect(classifyPage("https://x.com/p/4", "Acme signs power purchase agreement", "").pageType).toBe("power_infrastructure");
    expect(classifyPage("https://x.com/p/5", "Acme launches new region in Madrid", "availability zones").pageType).toBe("cloud_region");
    expect(classifyPage("https://x.com/p/6", "Acme announces $2 billion campus", "").pageType).toBe("project_announcement");
    expect(classifyPage("https://x.com/p/7", "Quarterly newsletter", "nothing here").pageType).toBe("unknown");
  });
  it("understands German, French, Spanish, Portuguese and Dutch announcements", () => {
    expect(classifyPage("https://x.de/p/1", "Acme baut neues Rechenzentrum in Frankfurt", "").pageType).toBe("project_announcement");
    expect(classifyPage("https://x.fr/p/1", "Acme annonce un nouveau centre de données à Marseille", "communiqué de presse").pageType).toBe("project_announcement");
    expect(classifyPage("https://x.es/p/1", "Acme construirá un nuevo centro de datos en Madrid", "").pageType).toBe("project_announcement");
    expect(classifyPage("https://x.pt/p/1", "Acme anuncia novo centro de dados em Sines", "").pageType).toBe("project_announcement");
    expect(classifyPage("https://x.nl/p/1", "Acme bouwt nieuw datacentrum in Groningen", "").pageType).toBe("project_announcement");
  });
  it("filters portfolio totals out of the MW list", () => {
    const c = classifyPage("https://x.com/p/8", "Acme announces 60 MW data center in Lisbon", "The new facility brings the company's total capacity to 1.4 GW across Europe. The 60MW site opens in 2028.");
    expect(c.mw).toEqual([60]);
  });
});

describe("eventTypeFor", () => {
  it("maps page types to event types with text refinements", () => {
    expect(eventTypeFor("construction_update", "Acme breaks ground")).toBe("construction_started");
    expect(eventTypeFor("construction_update", "construction progress update")).toBe("project_status_changed");
    expect(eventTypeFor("planning_document", "the board approved the application")).toBe("planning_approved");
    expect(eventTypeFor("planning_document", "application filed")).toBe("planning_filed");
    expect(eventTypeFor("cloud_region", "Acme announces upcoming region")).toBe("cloud_region_announced");
    expect(eventTypeFor("cloud_region", "region is now available")).toBe("cloud_region_launched");
    expect(eventTypeFor("project_announcement", "to invest $5 billion in the state")).toBe("investment_announced");
    expect(eventTypeFor("project_announcement", "to invest $5 billion in a 300 MW campus")).toBe("project_announced");
    expect(eventTypeFor("project_announcement", "Acme investiert 1 Mrd. in ein neues Rechenzentrum")).toBe("project_announced");
    expect(eventTypeFor("press_release", "")).toBe("news");
    expect(eventTypeFor("unknown", "")).toBeUndefined();
  });
});

describe("isDataCenterRelevant", () => {
  it("accepts data-center vocabulary in every indexed language", () => {
    for (const t of ["new data center in Ohio", "colocation provider", "50 MW campus", "Rechenzentrum in Frankfurt", "centre de données à Paris", "centro de datos en Madrid", "centro de dados em Lisboa", "datacentrum in Amsterdam", "nytt datasenter i Oslo", "nytt datacenter i Stockholm", "serverhall i Luleå"]) expect(isDataCenterRelevant(t), t).toBe(true);
  });
  it("rejects unrelated text", () => {
    expect(isDataCenterRelevant("Quarterly dividend declared by the board")).toBe(false);
    expect(isDataCenterRelevant("Acme names Jane Doe as CFO")).toBe(false);
  });
});

describe("projectMwFigures", () => {
  it("keeps site figures, drops portfolio/company-wide figures, dedupes, sorts descending, bounds 0–20 000", () => {
    expect(projectMwFigures("a 100 MW campus with a first 20MW phase and 20 MW later; portfolio of 2 GW across Asia; 30,000 MW pipeline")).toEqual([100, 20]);
    expect(projectMwFigures("brings the total capacity to 500 MW")).toEqual([]);
    expect(projectMwFigures("now operates 800 MW globally")).toEqual([]);
    expect(projectMwFigures("1.5 GW of contracted capacity")).toEqual([]);
    expect(projectMwFigures("no figures here")).toEqual([]);
  });
});