pvs_chat/js/page-analyzer.js

// 确保类定义在全局作用域
window.PageAnalyzer = class PageAnalyzer {
  constructor() {
    this.readability = null;
  }

  /**
   * 分析页面内容
   * @returns {Object} 页面分析结果
   */
  analyzePage() {
    try {
      // 检查Readability是否可用
      if (typeof Readability === "undefined") {
        console.warn(
          "Readability not loaded, falling back to basic extraction"
        );
        return this.fallbackAnalysis();
      }

      // 创建文档副本以避免修改原始DOM
      const documentClone = document.cloneNode(true);

      // 初始化 Readability
      this.readability = new Readability(documentClone, {
        debug: false,
        charThreshold: 20,
      });

      // 解析页面
      const article = this.readability.parse();

      return {
        title: article.title || document.title,
        url: window.location.href,
        mainContent: article.textContent || article.excerpt || "",
        excerpt: article.excerpt || "",
        siteName: article.siteName || new URL(window.location.href).hostname,
        wordCount: article.length || 0,
      };
    } catch (error) {
      console.warn("Readability failed, using fallback:", error);
      return this.fallbackAnalysis();
    }
  }

  // 基础提取方法作为后备
  fallbackAnalysis() {
    return {
      title: document.title,
      url: window.location.href,
      mainContent: this.extractMainContent(),
      excerpt: "",
      siteName: new URL(window.location.href).hostname,
      wordCount: 0,
    };
  }

  // 基础的内容提取方法
  extractMainContent() {
    // 移除脚本、样式等
    const content = document.body.cloneNode(true);
    // content
    //   .querySelectorAll("script, style, iframe, nav, header, footer")
    //   .forEach((el) => el.remove());
    //提取页面的表单
    const form = document.querySelector("form");
    if (form) {
      content.textContent = form.outerHTML;
    }

    return content.textContent.trim().replace(/\s+/g, " ");
  }
};

// 创建全局实例
window.pageAnalyzer = new window.PageAnalyzer();