import { describe, it, expect, beforeAll } from 'vitest';
import { TextRankKeyword, TextRankSentence } from '../src/index';

describe('TextRank4ZH-TS 集成测试', () => {
  const sampleTexts = {
    news: `中新网北京12月1日电(记者 张曦) 30日晚，高圆圆和赵又廷在京举行答谢宴，诸多明星现身捧场，其中包括张杰、谢娜夫妇、何炅、蔡康永、徐克、张凯丽、黄轩等。高圆圆身穿粉色外套，看到大批记者在场露出娇羞神色，赵又廷则戴着鸭舌帽，十分淡定，两人快步走进电梯，未接受媒体采访。记者了解到，出席高圆圆、赵又廷答谢宴的宾客近百人，其中不少都是女方的高中同学。`,

    tech: `人工智能技术在自然语言处理领域取得了重大突破。机器学习和深度学习推动了人工智能的快速发展。神经网络模型能够理解复杂的语言结构和语义信息。文本分析、语音识别、图像处理等应用场景不断涌现。未来人工智能将在更多领域发挥重要作用。`,

    education: `中国的教育事业蓬勃发展，各级各类学校办学条件不断改善。高等教育进入普及化发展阶段，教育质量稳步提升。职业教育体系日趋完善，为经济社会发展培养了大批技能人才。基础教育均衡发展，城乡教育差距逐步缩小。`,

    short: `北京是中国的首都。`,
  };

  describe('新闻文本分析场景', () => {
    let tr4w: TextRankKeyword;
    let tr4s: TextRankSentence;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      tr4s = new TextRankSentence();

      // 等待分词器初始化
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('应该能够提取新闻关键词', () => {
      tr4w.analyze(sampleTexts.news, {
        lower: true,
        window: 2,
        pageRankConfig: { alpha: 0.85 },
      });

      const keywords = tr4w.getKeywords(10, 1);

      expect(keywords).toBeDefined();
      expect(keywords.length).toBeGreaterThan(0);
      expect(keywords.length).toBeLessThanOrEqual(10);

      // 检查关键词结构
      keywords.forEach((keyword) => {
        expect(keyword).toHaveProperty('word');
        expect(keyword).toHaveProperty('weight');
        expect(typeof keyword.word).toBe('string');
        expect(typeof keyword.weight).toBe('number');
        expect(keyword.weight).toBeGreaterThan(0);
        expect(keyword.word.length).toBeGreaterThanOrEqual(1);
      });

      // 权重应该递减
      for (let i = 1; i < keywords.length; i++) {
        const prev = keywords[i - 1];
        const current = keywords[i];
        expect(prev).toBeDefined();
        expect(current).toBeDefined();
        if (prev && current) {
          expect(prev.weight).toBeGreaterThanOrEqual(current.weight);
        }
      }
    });

    it('应该能够识别新闻中的关键短语', () => {
      tr4w.analyze(sampleTexts.news, { lower: true, window: 2 });
      const keyphrases = tr4w.getKeyphrases(15, 1);

      expect(Array.isArray(keyphrases)).toBe(true);
      keyphrases.forEach((phrase) => {
        expect(typeof phrase).toBe('string');
        expect(phrase.length).toBeGreaterThan(1);
      });
    });

    it('应该能够生成新闻摘要', () => {
      tr4s.analyze(sampleTexts.news, {
        lower: true,
        source: 'all_filters',
        pageRankConfig: { alpha: 0.85 },
      });

      const keySentences = tr4s.getKeySentences(2);

      expect(keySentences).toBeDefined();
      expect(keySentences.length).toBeGreaterThan(0);
      expect(keySentences.length).toBeLessThanOrEqual(2);

      keySentences.forEach((sentence) => {
        expect(sentence).toHaveProperty('index');
        expect(sentence).toHaveProperty('sentence');
        expect(sentence).toHaveProperty('weight');
        expect(typeof sentence.index).toBe('number');
        expect(typeof sentence.sentence).toBe('string');
        expect(typeof sentence.weight).toBe('number');
      });

      const summary = tr4s.getSummary(2, 10, true);
      expect(typeof summary).toBe('string');
      expect(summary.length).toBeGreaterThan(0);
    });
  });

  describe('技术文本分析场景', () => {
    let tr4w: TextRankKeyword;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('应该能够提取技术领域关键词', () => {
      tr4w.analyze(sampleTexts.tech, {
        lower: true,
        window: 3,
        vertexSource: 'all_filters',
        edgeSource: 'no_stop_words',
      });

      const keywords = tr4w.getKeywords(8, 1); // 改为最小长度1，适应单字分词

      expect(keywords.length).toBeGreaterThan(0);

      // 验证是否包含技术相关字符（适应轻量级分词器）
      const keywordTexts = keywords.map((k) => k.word);
      const hasTechWords = keywordTexts.some(
        (word) =>
          word.includes('智') ||
          word.includes('学') ||
          word.includes('技') ||
          word.includes('络') ||
          word.includes('工') ||
          word.includes('理') ||
          word.includes('语')
      );
      expect(hasTechWords).toBe(true);
    });
  });

  describe('配置参数影响测试', () => {
    let tr4w: TextRankKeyword;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('不同窗口大小应该产生不同结果', () => {
      // 窗口大小为2
      tr4w.analyze(sampleTexts.tech, { window: 2 });
      const keywords2 = tr4w.getKeywords(5);

      // 窗口大小为4
      tr4w.analyze(sampleTexts.tech, { window: 4 });
      const keywords4 = tr4w.getKeywords(5);

      // 权重分布应该有差异
      const weights2 = keywords2.map((k) => k.weight);
      const weights4 = keywords4.map((k) => k.weight);

      expect(weights2).not.toEqual(weights4);
    });

    it('不同PageRank参数应该影响结果', () => {
      // alpha = 0.5
      tr4w.analyze(sampleTexts.tech, {
        pageRankConfig: { alpha: 0.5 },
      });
      const keywords1 = tr4w.getKeywords(5);

      // alpha = 0.9
      tr4w.analyze(sampleTexts.tech, {
        pageRankConfig: { alpha: 0.9 },
      });
      const keywords2 = tr4w.getKeywords(5);

      // 权重应该有差异
      const weights1 = keywords1.map((k) => k.weight);
      const weights2 = keywords2.map((k) => k.weight);

      expect(weights1).not.toEqual(weights2);
    });
  });

  describe('边界情况处理', () => {
    let tr4w: TextRankKeyword;
    let tr4s: TextRankSentence;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      tr4s = new TextRankSentence();
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('应该能够处理短文本', () => {
      tr4w.analyze(sampleTexts.short);
      const keywords = tr4w.getKeywords(5);

      expect(Array.isArray(keywords)).toBe(true);

      tr4s.analyze(sampleTexts.short);
      const sentences = tr4s.getKeySentences(2);

      expect(Array.isArray(sentences)).toBe(true);
      expect(sentences.length).toBeGreaterThan(0);
    });

    it('应该能够处理空文本', () => {
      tr4w.analyze('');
      const keywords = tr4w.getKeywords(5);
      const keyphrases = tr4w.getKeyphrases(5);

      expect(Array.isArray(keywords)).toBe(true);
      expect(Array.isArray(keyphrases)).toBe(true);
      expect(keywords.length).toBe(0);
      expect(keyphrases.length).toBe(0);
    });

    it('应该能够处理纯标点符号文本', () => {
      tr4w.analyze('！！！？？？。。。');
      const keywords = tr4w.getKeywords(5);

      expect(Array.isArray(keywords)).toBe(true);
      expect(keywords.length).toBe(0);
    });
  });

  describe('性能测试', () => {
    let tr4w: TextRankKeyword;
    let tr4s: TextRankSentence;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      tr4s = new TextRankSentence();
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('处理中等长度文本应该在合理时间内完成', () => {
      const longText = sampleTexts.news.repeat(5); // 约1000字

      const startTime = Date.now();
      tr4w.analyze(longText, { lower: true, window: 2 });
      const keywords = tr4w.getKeywords(10);
      const keywordsTime = Date.now() - startTime;

      const sentenceStartTime = Date.now();
      tr4s.analyze(longText, { lower: true, source: 'all_filters' });
      const sentences = tr4s.getKeySentences(3);
      const sentencesTime = Date.now() - sentenceStartTime;

      // 关键词提取应该在2秒内完成
      expect(keywordsTime).toBeLessThan(2000);
      expect(keywords.length).toBeGreaterThan(0);

      // 句子摘要应该在2秒内完成
      expect(sentencesTime).toBeLessThan(2000);
      expect(sentences.length).toBeGreaterThan(0);
    });
  });

  describe('多语言混合文本处理', () => {
    let tr4w: TextRankKeyword;

    beforeAll(async () => {
      tr4w = new TextRankKeyword();
      await new Promise((resolve) => setTimeout(resolve, 100));
    });

    it('应该能够处理中英文混合文本', () => {
      const mixedText =
        '这是一个关于 Machine Learning 和人工智能 AI 的测试。Deep Learning 深度学习正在改变世界。';

      tr4w.analyze(mixedText, { lower: true });
      const keywords = tr4w.getKeywords(8);

      expect(keywords.length).toBeGreaterThan(0);

      // 应该能提取中文和英文词汇
      const keywordTexts = keywords.map((k) => k.word);
      const hasChineseWords = keywordTexts.some((word) => /[\u4e00-\u9fa5]/.test(word));
      const hasEnglishWords = keywordTexts.some((word) => /[a-zA-Z]/.test(word));

      expect(hasChineseWords || hasEnglishWords).toBe(true);
    });

    it('应该能够处理包含数字的文本', () => {
      const numberText = '2023年中国GDP增长5.2%，比2022年提高了0.8个百分点。';

      tr4w.analyze(numberText, { lower: true });
      const keywords = tr4w.getKeywords(5);

      expect(keywords.length).toBeGreaterThan(0);
    });
  });
});