read_pdf

Instructions

Implementation Reference

• src/pdf-tools.ts:9-37 (handler)

  The core handler function 'extractTextFromPDF' that reads a PDF file and extracts all text content along with metadata (title, author, subject, etc.) and page count. This is the actual implementation called when 'read_pdf' is invoked.

  export async function extractTextFromPDF(filePath: string): Promise<PDFExtractionResult> {
    try {
      const dataBuffer = await fs.readFile(filePath);
      const parser = new PDFParse({ data: dataBuffer });
      const textResult = await parser.getText();
      const infoResult = await parser.getInfo();
      await parser.destroy();
  
      const metadata: PDFMetadata = {
        title: infoResult.info?.Title,
        author: infoResult.info?.Author,
        subject: infoResult.info?.Subject,
        creator: infoResult.info?.Creator,
        producer: infoResult.info?.Producer,
        creationDate: infoResult.info?.CreationDate,
        modificationDate: infoResult.info?.ModDate,
        keywords: infoResult.info?.Keywords,
        totalPages: infoResult.total,
      };
  
      return {
        text: textResult.text,
        metadata,
        pageCount: infoResult.total,
      };
    } catch (error) {
      throw new Error(`Failed to read PDF: ${error instanceof Error ? error.message : String(error)}`);
    }
  }

• src/types.ts:27-31 (schema)

  The PDFExtractionResult interface defines the return type of the read_pdf handler, containing text, optional metadata, and pageCount.

  export interface PDFExtractionResult {
    text: string;
    metadata?: PDFMetadata;
    pageCount: number;
  }

• src/index.ts:22-41 (registration)

  Tool registration definition for 'read_pdf' in the TOOLS array, including its name, description, and inputSchema (filePath required, includeMetadata optional).

  const TOOLS: Tool[] = [
    {
      name: 'read_pdf',
      description: 'Extract all text content from a PDF file',
      inputSchema: {
        type: 'object',
        properties: {
          filePath: {
            type: 'string',
            description: 'Absolute path to the PDF file',
          },
          includeMetadata: {
            type: 'boolean',
            description: 'Whether to include PDF metadata in the response',
            default: false,
          },
        },
        required: ['filePath'],
      },
    },

• src/index.ts:175-202 (handler)

  The CallToolRequestSchema handler case for 'read_pdf' in the main server - dispatches to extractTextFromPDF and formats the response based on includeMetadata flag.

  case 'read_pdf': {
    const { filePath, includeMetadata } = args as {
      filePath: string;
      includeMetadata?: boolean;
    };
    
    const result = await extractTextFromPDF(filePath);
    
    if (includeMetadata) {
      return {
        content: [
          {
            type: 'text',
            text: JSON.stringify(result, null, 2),
          },
        ],
      };
    }
    
    return {
      content: [
        {
          type: 'text',
          text: result.text,
        },
      ],
    };
  }

• src/types.ts:4-14 (schema)

  The PDFMetadata interface defines the metadata structure returned as part of the read_pdf result.

  export interface PDFMetadata {
    title?: string;
    author?: string;
    subject?: string;
    creator?: string;
    producer?: string;
    creationDate?: string;
    modificationDate?: string;
    keywords?: string;
    totalPages?: number;
  }