Word Document Reader MCP Server

Instructions

Implementation Reference

• server.js:759-795 (handler)

  Executes the search_documents tool: parses args, calls documentIndexer.search(query), filters by documentType, limits results, formats and returns search results.

  case "search_documents": {
    const { query, documentType, limit = 10 } = args;
  
    const searchResults = documentIndexer.search(query);
  
    // 按文档类型过滤
    const filteredResults = documentType
      ? searchResults.filter(result => result.document.documentType === documentType)
      : searchResults;
  
    const limitedResults = filteredResults.slice(0, limit);
  
    if (limitedResults.length === 0) {
      return {
        content: [
          {
            type: "text",
            text: `未找到包含关键词 "${query}" 的文档`
          }
        ]
      };
    }
  
    const resultsText = limitedResults.map((result, index) => {
      const doc = result.document;
      return `${index + 1}. 相关度: ${result.score}\n   内存键: ${doc.memoryKey}\n   文件: ${doc.filePath}\n   类型: ${doc.documentType}\n   表格数: ${doc.tablesCount}\n   图片数: ${doc.imagesCount}\n   最后索引: ${new Date(doc.lastIndexed).toLocaleString()}`;
    }).join('\n\n');
  
    return {
      content: [
        {
          type: "text",
          text: `搜索结果 "${query}" (找到 ${limitedResults.length} 个匹配，共 ${filteredResults.length} 个):\n\n${resultsText}`
        }
      ]
    };
  }

• server.js:518-541 (registration)

  Registers the search_documents tool in the ListTools response, including name, description, and input schema.

  {
    name: "search_documents",
    description: "全文索引搜索，支持中英文混合搜索",
    inputSchema: {
      type: "object",
      properties: {
        query: {
          type: "string",
          description: "搜索关键词"
        },
        documentType: {
          type: "string",
          description: "限制搜索的文档类型",
          enum: ["ui-component", "api-doc", "common-doc", "other"]
        },
        limit: {
          type: "number",
          description: "返回结果数量限制",
          default: 10
        }
      },
      required: ["query"]
    }
  },

• server.js:106-130 (helper)

  Core search logic in DocumentIndexer: extracts words from query, computes relevance scores using inverted index, sorts and returns results with metadata.

  search(query) {
    const queryWords = this.extractWords(query.toLowerCase());
    const documentScores = new Map();
  
    queryWords.forEach(word => {
      const docs = this.index.get(word);
      if (docs) {
        docs.forEach(docId => {
          const score = documentScores.get(docId) || 0;
          documentScores.set(docId, score + 1);
        });
      }
    });
  
    // 按相关性排序
    const results = Array.from(documentScores.entries())
      .sort((a, b) => b[1] - a[1])
      .map(([docId, score]) => ({
        documentId,
        score,
        document: this.documents.get(docId)
      }));
  
    return results;
  }

• server.js:75-102 (helper)

  Helper method used by search: tokenizes text into Chinese n-grams (2-4 chars), lowercase English words, and numbers.

  extractWords(text) {
    // 中英文分词
    const chinesePattern = /[\u4e00-\u9fff]+/g;
    const englishPattern = /[a-zA-Z]+/g;
    const numberPattern = /\d+/g;
  
    const words = [];
  
    // 提取中文词汇（简单分词，后续可优化为更智能的分词）
    const chineseMatches = text.match(chinesePattern) || [];
    chineseMatches.forEach(match => {
      // 简单的中文二字、三字、四字词分词
      for (let i = 0; i < match.length; i++) {
        for (let len = 2; len <= 4 && i + len <= match.length; len++) {
          words.push(match.substr(i, len));
        }
      }
    });
  
    // 提取英文单词
    const englishMatches = text.match(englishPattern) || [];
    words.push(...englishMatches.map(word => word.toLowerCase()));
  
    // 提取数字
    const numberMatches = text.match(numberPattern) || [];
    words.push(...numberMatches);
  
    return words.filter(word => word.length > 0);