import { describe, test, expect } from "bun:test";
import { matchCachedModels } from "../src/lib/modelCache";

// Pure URL matching (D-005 downloaded badge); the `caches` wrapper is
// browser-only and exercised manually.

const TINY = "onnx-community/whisper-tiny.en";
const BASE = "onnx-community/whisper-base";

function hubUrl(modelId: string, file: string): string {
  return `https://huggingface.co/${modelId}/resolve/main/${file}`;
}

describe("matchCachedModels", () => {
  test("matches models with cached .onnx weights", () => {
    const urls = [
      hubUrl(TINY, "onnx/encoder_model_quantized.onnx"),
      hubUrl(TINY, "onnx/decoder_model_merged_quantized.onnx"),
      hubUrl(TINY, "tokenizer.json"),
    ];
    expect(matchCachedModels(urls, [TINY, BASE])).toEqual(new Set([TINY]));
  });

  test("configs/tokenizer alone do not count as downloaded", () => {
    const urls = [hubUrl(BASE, "tokenizer.json"), hubUrl(BASE, "config.json")];
    expect(matchCachedModels(urls, [TINY, BASE])).toEqual(new Set());
  });

  test("ignores query strings when checking the .onnx extension", () => {
    const urls = [`${hubUrl(BASE, "onnx/encoder_model_q4.onnx")}?download=true`];
    expect(matchCachedModels(urls, [BASE])).toEqual(new Set([BASE]));
  });

  test("does not confuse '.en' variants with their multilingual sibling", () => {
    // "onnx-community/whisper-tiny.en/..." contains the substring
    // "whisper-tiny" but must only match the .en id thanks to the trailing /.
    const urls = [hubUrl(TINY, "onnx/encoder_model_quantized.onnx")];
    expect(
      matchCachedModels(urls, [TINY, "onnx-community/whisper-tiny"]),
    ).toEqual(new Set([TINY]));
  });

  test("empty inputs yield an empty set", () => {
    expect(matchCachedModels([], [TINY])).toEqual(new Set());
    expect(matchCachedModels([hubUrl(TINY, "onnx/encoder_model.onnx")], [])).toEqual(
      new Set(),
    );
  });
});