import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
import path from 'path';
import os from 'os';

describe('pipeline cache + deterministic IDs', () => {
  const tmpCache = () => path.join(os.tmpdir(), `cache-${Date.now()}`);

  beforeEach(() => {
    vi.resetModules();
    process.env.PIPELINE_CHUNK_SOURCE = 'jsonl';
    process.env.PIPELINE_CACHE_DIR = tmpCache();
  });

  afterEach(() => {
    delete process.env.PIPELINE_CHUNK_SOURCE;
    delete process.env.PIPELINE_CACHE_DIR;
    vi.restoreAllMocks();
    vi.unmock('../src/retrieval/jsonl_chunks.mjs');
    vi.unmock('../src/pipeline/step.mjs');
  });

  it('uses cached reward to skip generator/verifier/reward work', async () => {
    const chunk = { content: 'hello world', sourceId: 's1' };

    // Deterministic IDs and cache helpers
    const {
      chunkIdFromContent,
      questionId,
      saveQuestions,
      saveReward,
    } = await import('../src/pipeline/cache.mjs');

    const chunkId = chunkIdFromContent(chunk.content, chunk.sourceId);
    const question = 'What is said?';
    const qId = questionId(chunkId, question);

    // Pre-populate cache with questions and an accepted reward
    await saveQuestions(chunkId, [question], {
      provider: 'mock',
      model: 'mock',
    });
    await saveReward(
      chunkId,
      qId,
      'gen-1',
      { score: 0.9, ok: true, raw: '0.9' },
      { provider: 'mock', model: 'mock' },
    );

    // Mock chunk loader and pipeline step to observe bypass
    vi.doMock('../src/retrieval/jsonl_chunks.mjs', () => ({
      loadRagChunks: vi.fn(async () => [{ ...chunk, id: chunkId }]),
    }));

    const runPipelineStep = vi.fn();
    vi.doMock('../src/pipeline/step.mjs', () => ({ runPipelineStep }));

    const { runPipelineBatch } = await import('../src/pipeline/batch.mjs');

    const result = await runPipelineBatch({
      seedMode: 'question-first',
      limit: 5,
      verbose: false,
      logger: { log() {}, error() {} },
    });

    expect(runPipelineStep).not.toHaveBeenCalled();
    expect(result.processed).toBe(1);
    expect(result.accepted).toBe(1);
    expect(result.statusCounts.cached_reward).toBe(1);
  });

  it('stable chunk ids despite whitespace differences', async () => {
    const { chunkIdFromContent } = await import('../src/pipeline/ids.mjs');

    const idA = chunkIdFromContent('Hello   world', 'doc1');
    const idB = chunkIdFromContent('Hello world', 'doc1');
    const idC = chunkIdFromContent('Hello world!', 'doc1');

    expect(idA).toBe(idB);
    expect(idA).not.toBe(idC); // punctuation changes content hash
  });
});