@mcpcn/markdown-converter-mcp 1.0.1

package/src/core/WordConverter.ts

@@ -0,0 +1,470 @@
+/**
+ * Word转换器
+ * 基于md-to-word-mcp项目的成功实现，继承BaseConverter
+ */
+
+import { writeFile } from 'fs/promises';
+import { createRequire } from 'module';
+const require = createRequire(import.meta.url);
+
+const docx = require('docx');
+const {
+  Document,
+  Paragraph,
+  TextRun,
+  Packer,
+  AlignmentType,
+  BorderStyle,
+  Footer,
+  PageNumber,
+  PageOrientation
+} = docx;
+
+// 类型定义  
+type ParagraphType = any;
+type TextRunType = any;
+
+import { BaseConverter, ConversionResult, BaseConversionOptions } from './BaseConverter.js';
+import { 
+  WordStyleConfig, 
+  CodeBlockTheme, 
+  WordConversionOptions,
+  defaultWordStyleConfig, 
+  codeBlockThemes 
+} from '../config/word.js';
+
+export class WordConverter extends BaseConverter {
+  private styleConfig: WordStyleConfig;
+  private codeBlockTheme: CodeBlockTheme;
+
+  constructor(options: WordConversionOptions & { cacheEnabled?: boolean; cacheDir?: string } = {}) {
+    super(options);
+    
+    // 合并样式配置
+    this.styleConfig = { ...defaultWordStyleConfig, ...options.styleConfig };
+    
+    // 设置代码块主题
+    if (typeof options.codeBlockTheme === 'string') {
+      this.codeBlockTheme = codeBlockThemes[options.codeBlockTheme] || codeBlockThemes.default;
+    } else if (options.codeBlockTheme) {
+      this.codeBlockTheme = options.codeBlockTheme;
+    } else {
+      this.codeBlockTheme = codeBlockThemes.default;
+    }
+  }
+
+  /**
+   * 转换Markdown为Word文档
+   */
+  async convert(
+    markdownContent: string, 
+    outputPath: string, 
+    options: BaseConversionOptions & WordConversionOptions = {}
+  ): Promise<ConversionResult> {
+    const startTime = Date.now();
+
+    try {
+      // 检查缓存
+      const cacheKey = this.generateCacheKey(markdownContent, options);
+      const cachedResult = await this.loadFromCache(cacheKey, 'docx');
+      
+      if (cachedResult) {
+        this.reportProgress({
+          stage: 'complete',
+          progress: 100,
+          message: '使用缓存结果'
+        }, options.progress);
+
+        await this.ensureOutputDir(outputPath);
+        await writeFile(outputPath, cachedResult);
+        
+        const fileSize = await this.getFileSize(outputPath);
+        return {
+          success: true,
+          filePath: outputPath,
+          fileSize,
+          stats: {
+            processingTime: Date.now() - startTime,
+            headings: 0,
+            codeBlocks: 0,
+            paragraphs: 0,
+            tables: 0,
+            fileSize
+          }
+        };
+      }
+
+      this.reportProgress({
+        stage: 'parsing',
+        progress: 10,
+        message: '解析Markdown内容'
+      }, options.progress);
+
+      // 解析内容并获取统计信息
+      const { content, stats: parseStats } = this.parseMarkdown(markdownContent);
+      
+      this.reportProgress({
+        stage: 'converting',
+        progress: 30,
+        message: '转换为Word文档'
+      }, options.progress);
+
+      // 转换为Word文档
+      const docChildren = this.processMarkdownToWord(content);
+      const doc = this.createDocument(docChildren, options);
+
+      this.reportProgress({
+        stage: 'saving',
+        progress: 80,
+        message: '生成Word文件'
+      }, options.progress);
+
+      // 生成文件
+      const buffer = await Packer.toBuffer(doc);
+      
+      // 保存文件
+      await this.ensureOutputDir(outputPath);
+      await writeFile(outputPath, buffer);
+
+      // 保存到缓存
+      await this.saveToCache(cacheKey, 'docx', buffer);
+
+      const fileSize = await this.getFileSize(outputPath);
+      const processingTime = Date.now() - startTime;
+
+      this.reportProgress({
+        stage: 'complete',
+        progress: 100,
+        message: `转换完成: ${this.formatFileSize(fileSize)}`
+      }, options.progress);
+
+      return {
+        success: true,
+        filePath: outputPath,
+        fileSize,
+        stats: {
+          processingTime,
+          headings: parseStats.headings || 0,
+          codeBlocks: parseStats.codeBlocks || 0,
+          paragraphs: parseStats.paragraphs || 0,
+          tables: parseStats.tables || 0,
+          fileSize
+        }
+      };
+
+    } catch (error) {
+      console.error('Word转换失败:', error);
+      return {
+        success: false,
+        error: error instanceof Error ? error.message : String(error)
+      };
+    }
+  }
+
+  /**
+   * 处理Markdown内容转换为Word段落
+   */
+  private processMarkdownToWord(markdownContent: string): any[] {
+    const lines = markdownContent.split('\n');
+    const docChildren: any[] = [];
+    
+    let inCodeBlock = false;
+    let codeBlockContent = '';
+    let codeBlockLanguage = '';
+    
+    for (let i = 0; i < lines.length; i++) {
+      const line = lines[i];
+      const trimmedLine = line.trim();
+
+      // 处理代码块开始/结束
+      if (trimmedLine.startsWith('```')) {
+        if (!inCodeBlock) {
+          // 代码块开始
+          inCodeBlock = true;
+          codeBlockLanguage = trimmedLine.substring(3).trim();
+          codeBlockContent = '';
+        } else {
+          // 代码块结束
+          inCodeBlock = false;
+          
+          // 创建优化的代码块
+          const codeBlockParagraphs = this.createOptimizedCodeBlock(
+            codeBlockContent,
+            codeBlockLanguage
+          );
+          
+          docChildren.push(...codeBlockParagraphs);
+          
+          // 在代码块后添加分隔段落
+          docChildren.push(new Paragraph({ text: '' }));
+          
+          codeBlockContent = '';
+          codeBlockLanguage = '';
+        }
+        continue;
+      }
+
+      if (inCodeBlock) {
+        // 收集代码块内容
+        codeBlockContent += (codeBlockContent ? '\n' : '') + line;
+        continue;
+      }
+
+      // 处理标题
+      if (trimmedLine.startsWith('#')) {
+        const match = trimmedLine.match(/^#+/);
+        if (match) {
+          const level = match[0].length;
+          const headingText = trimmedLine.substring(level).trim();
+          
+          if (level >= 1 && level <= 6) {
+            docChildren.push(this.createHeading(headingText, level));
+            continue;
+          }
+        }
+      }
+
+      // 处理普通段落
+      if (trimmedLine) {
+        docChildren.push(this.createParagraph(trimmedLine));
+      } else {
+        // 空行
+        docChildren.push(new Paragraph({ text: '' }));
+      }
+    }
+
+    return docChildren;
+  }
+
+  /**
+   * 创建优化的代码块 - 基于成功的实现
+   */
+  private createOptimizedCodeBlock(code: string, language?: string): any[] {
+    const lines = code.split('\n');
+    const paragraphs: any[] = [];
+
+    // 如果有语言标识，添加语言标签段落
+    if (language && this.codeBlockTheme.showLanguage) {
+      paragraphs.push(
+        new Paragraph({
+          children: [
+            new TextRun({
+              text: `语言: ${language}`,
+              font: this.codeBlockTheme.fontFamily,
+              size: this.codeBlockTheme.fontSize - 4, // 稍微小一点
+              bold: true,
+              color: '666666',
+              shading: { fill: 'E8E8E8' }
+            })
+          ],
+          spacing: { before: this.styleConfig.spacing.codeBlock / 2, after: 60 },
+          shading: { fill: 'E8E8E8' },
+          border: {
+            top: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+            bottom: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+            left: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+            right: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+          },
+          indent: { left: this.styleConfig.margins.codeBlock.left }
+        })
+      );
+    }
+
+    // 为每一行代码创建单独的段落 - 这是关键的成功因素
+    lines.forEach((line, index) => {
+      // 处理空行
+      const displayText = line.trim() === '' ? ' ' : line;
+      
+      // 保持原有的缩进
+      const leadingSpaces = line.match(/^(\s*)/)?.[1] || '';
+      const indentLevel = leadingSpaces.length * 50; // 每个空格50单位缩进
+      
+      paragraphs.push(
+        new Paragraph({
+          children: [
+            new TextRun({
+              text: displayText,
+              font: this.codeBlockTheme.fontFamily,
+              size: this.codeBlockTheme.fontSize,
+              color: this.codeBlockTheme.textColor
+            })
+          ],
+          spacing: {
+            line: this.styleConfig.spacing.line, // 固定行高
+            lineRule: 'exact'
+          },
+          shading: { fill: this.codeBlockTheme.background },
+          border: {
+            left: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+            right: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border },
+            ...(index === 0 && !this.codeBlockTheme.showLanguage ? 
+              { top: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border } } : {}),
+            ...(index === lines.length - 1 ? 
+              { bottom: { style: BorderStyle.SINGLE, size: 1, color: this.codeBlockTheme.border } } : {})
+          },
+          indent: { 
+            left: this.styleConfig.margins.codeBlock.left + indentLevel
+          }
+        })
+      );
+    });
+
+    return paragraphs;
+  }
+
+  /**
+   * 创建标题段落
+   */
+  private createHeading(text: string, level: number): any {
+    let fontSize: number;
+    
+    switch (level) {
+      case 1: fontSize = this.styleConfig.fontSizes.heading1; break;
+      case 2: fontSize = this.styleConfig.fontSizes.heading2; break;
+      case 3: fontSize = this.styleConfig.fontSizes.heading3; break;
+      case 4: fontSize = this.styleConfig.fontSizes.heading4; break;
+      case 5: fontSize = this.styleConfig.fontSizes.heading5; break;
+      default: fontSize = this.styleConfig.fontSizes.heading5; break;
+    }
+
+    return new Paragraph({
+      children: [
+        new TextRun({
+          text: text,
+          font: this.styleConfig.fonts.heading,
+          size: fontSize,
+          bold: true,
+          color: this.styleConfig.colors.heading
+        })
+      ],
+      heading: `Heading${Math.min(level, 6)}` as any,
+      spacing: { 
+        before: this.styleConfig.spacing.heading, 
+        after: this.styleConfig.spacing.paragraph 
+      },
+      alignment: level === 1 ? AlignmentType.CENTER : AlignmentType.LEFT
+    });
+  }
+
+  /**
+   * 创建普通段落
+   */
+  private createParagraph(text: string): any {
+    // 清理和标准化文本 - 修复乱码问题
+    const cleanedText = this.cleanText(text);
+    
+    // 处理内联代码和其他Markdown语法
+    const children = this.processInlineMarkdown(cleanedText);
+    
+    return new Paragraph({
+      children,
+      spacing: { after: this.styleConfig.spacing.paragraph }
+    });
+  }
+
+  /**
+   * 处理行内Markdown语法
+   */
+  private processInlineMarkdown(text: string): any[] {
+    const runs: any[] = [];
+    const codeRegex = /`([^`]+)`/g;
+    let lastIndex = 0;
+    let match;
+
+    // 处理内联代码
+    while ((match = codeRegex.exec(text)) !== null) {
+      // 添加代码前的普通文本
+      if (match.index > lastIndex) {
+        const normalText = text.substring(lastIndex, match.index);
+        if (normalText.trim()) {
+          runs.push(new TextRun({
+            text: normalText,
+            font: this.styleConfig.fonts.default,
+            size: this.styleConfig.fontSizes.paragraph,
+            color: this.styleConfig.colors.text
+          }));
+        }
+      }
+
+      // 添加代码文本
+      runs.push(new TextRun({
+        text: match[1],
+        font: this.styleConfig.fonts.code,
+        size: this.styleConfig.fontSizes.code,
+        color: this.styleConfig.colors.code,
+        shading: { fill: this.styleConfig.colors.codeBackground },
+        border: {
+          style: BorderStyle.SINGLE,
+          size: 1,
+          color: this.styleConfig.colors.codeBorder
+        }
+      }));
+
+      lastIndex = codeRegex.lastIndex;
+    }
+
+    // 添加剩余的普通文本
+    if (lastIndex < text.length) {
+      const remainingText = text.substring(lastIndex);
+      if (remainingText.trim()) {
+        runs.push(new TextRun({
+          text: remainingText,
+          font: this.styleConfig.fonts.default,
+          size: this.styleConfig.fontSizes.paragraph,
+          color: this.styleConfig.colors.text
+        }));
+      }
+    }
+
+    // 如果没有找到任何特殊格式，返回普通文本
+    if (runs.length === 0) {
+      runs.push(new TextRun({
+        text: text,
+        font: this.styleConfig.fonts.default,
+        size: this.styleConfig.fontSizes.paragraph,
+        color: this.styleConfig.colors.text
+      }));
+    }
+
+    return runs;
+  }
+
+  /**
+   * 创建Word文档
+   */
+  private createDocument(children: any[], options: WordConversionOptions): any {
+    const docOptions: any = {
+      sections: [{
+        properties: {
+          page: {
+            margin: this.styleConfig.margins.page,
+            size: {
+              orientation: PageOrientation.PORTRAIT,
+            },
+          },
+        },
+        children,
+      }],
+    };
+
+    // 添加页脚（页码）
+    if (options.addFooter !== false) {
+      docOptions.sections[0].footers = {
+        default: new Footer({
+          children: [
+            new Paragraph({
+              alignment: AlignmentType.CENTER,
+              children: [
+                new TextRun({
+                  children: [PageNumber.CURRENT],
+                }),
+              ],
+            }),
+          ],
+        }),
+      };
+    }
+
+    return new Document(docOptions);
+  }
+}