mcp-pdf-reader

Instructions

Implementation Reference

• src/pdf_reader_mcp/server.py:521-548 (handler)

  Implements the core logic for the 'get-pdf-page-text' tool: validates inputs, retrieves the PDF reader, extracts text from the specified page using PyPDF2, and returns the text content.

  elif name == "get-pdf-page-text":
      pdf_id = arguments.get("pdf_id")
      if not pdf_id or pdf_id not in pdfs:
          raise ValueError("Invalid PDF ID")
  
      page_number = arguments.get("page_number")
      if page_number is None:
          raise ValueError("Missing page number")
  
      reader = pdfs[pdf_id]
  
      # Check if page number is valid
      if page_number < 0 or page_number >= len(reader.pages):
          raise ValueError(f"Invalid page number. PDF has {len(reader.pages)} pages (0-{len(reader.pages)-1})")
  
      # Extract text from the specified page
      page = reader.pages[page_number]
      page_text = page.extract_text()
  
      if not page_text:
          page_text = f"No extractable text found on page {page_number}"
  
      return [
          types.TextContent(
              type="text",
              text=f"Text from page {page_number} of '{os.path.basename(pdf_paths[pdf_id])}':\n\n{page_text}",
          )
      ]

• src/pdf_reader_mcp/server.py:395-406 (registration)

  Registers the 'get-pdf-page-text' tool with MCP server, providing name, description, and JSON schema for input validation (pdf_id and page_number required).

  types.Tool(
      name="get-pdf-page-text",
      description="Get the text content of a specific page in a PDF",
      inputSchema={
          "type": "object",
          "properties": {
              "pdf_id": {"type": "string", "description": "ID of the PDF to get page text from"},
              "page_number": {"type": "integer", "description": "Page number (0-based index)"},
          },
          "required": ["pdf_id", "page_number"],
      },
  ),