// author: simon-pierre boucher <contact@spboucher.ai>
import { describe, expect, it } from "vitest";
import { crawlDelay, isAllowed, parseRobots } from "./robots.js";

describe("parseRobots / isAllowed", () => {
  it("allows everything when robots is empty", () => {
    const r = parseRobots("");
    expect(isAllowed(r, "/anything")).toBe(true);
  });

  it("honors a wildcard Disallow", () => {
    const r = parseRobots("User-agent: *\nDisallow: /private/");
    expect(isAllowed(r, "/private/x")).toBe(false);
    expect(isAllowed(r, "/public/x")).toBe(true);
  });

  it("applies longest-match-wins between Allow and Disallow", () => {
    const r = parseRobots("User-agent: *\nDisallow: /a/\nAllow: /a/b");
    expect(isAllowed(r, "/a/b/c")).toBe(true);
    expect(isAllowed(r, "/a/x")).toBe(false);
  });

  it("supports * and $ wildcards", () => {
    const r = parseRobots("User-agent: *\nDisallow: /*.pdf$");
    expect(isAllowed(r, "/docs/file.pdf")).toBe(false);
    expect(isAllowed(r, "/docs/file.pdf?x=1")).toBe(true);
    expect(isAllowed(r, "/docs/file.html")).toBe(true);
  });

  it("selects the most specific matching user-agent group", () => {
    const r = parseRobots(
      "User-agent: *\nDisallow: /\n\nUser-agent: Tendril\nAllow: /\nDisallow: /secret",
    );
    expect(isAllowed(r, "/page", "Tendril/1.0")).toBe(true);
    expect(isAllowed(r, "/secret", "Tendril/1.0")).toBe(false);
    expect(isAllowed(r, "/page", "SomeOtherBot")).toBe(false);
  });

  it("treats an empty Disallow as allow-all", () => {
    const r = parseRobots("User-agent: *\nDisallow:");
    expect(isAllowed(r, "/anything")).toBe(true);
  });

  it("groups consecutive user-agent lines together", () => {
    const r = parseRobots("User-agent: A\nUser-agent: Tendril\nDisallow: /x");
    expect(isAllowed(r, "/x", "Tendril")).toBe(false);
  });

  it("collects Sitemap directives and Crawl-delay", () => {
    const r = parseRobots("Sitemap: https://e.com/sitemap.xml\nUser-agent: *\nCrawl-delay: 5\nDisallow: /y");
    expect(r.sitemaps).toEqual(["https://e.com/sitemap.xml"]);
    expect(crawlDelay(r)).toBe(5);
  });
});