All-in-MCP

Instructions

Implementation Reference

• src/apaper/server.py:124-171 (handler)

  MCP tool handler for 'apaper_read_iacr_paper' (inferred name from FastMCP('apaper') prefix). Converts args, calls IACRSearcher.read_paper, handles truncation and errors.

  @mcp.tool()
  def read_iacr_paper(
      paper_id: str,
      save_path: str = "./downloads",
      start_page: int | str | None = None,
      end_page: int | str | None = None,
  ) -> str:
      """
      Read and extract text content from an IACR ePrint paper PDF
      
      Args:
          paper_id: IACR paper ID (e.g., '2009/101')
          save_path: Directory where the PDF is/will be saved (default: './downloads')
          start_page: Starting page number (1-indexed, inclusive). Defaults to 1.
          end_page: Ending page number (1-indexed, inclusive). Defaults to last page.
      """
      try:
          # Convert string parameters to integers if needed
          start_page_int = None
          end_page_int = None
          
          if start_page is not None:
              start_page_int = int(start_page)
          
          if end_page is not None:
              end_page_int = int(end_page)
          
          result = iacr_searcher.read_paper(
              paper_id, save_path, start_page=start_page_int, end_page=end_page_int
          )
  
          if result.startswith("Error"):
              return result
          else:
              # Truncate very long text for display
              if len(result) > 5000:
                  truncated_result = (
                      result[:5000]
                      + f"\n\n... [Text truncated. Full text is {len(result)} characters long]"
                  )
                  return truncated_result
              else:
                  return result
      except ValueError as e:
          return f"Error: Invalid page number format. Please provide valid integers for start_page and end_page."
      except Exception as e:
          return f"Error reading IACR paper: {str(e)}"

• src/apaper/platforms/iacr.py:236-271 (helper)

  Core implementation of reading IACR paper: fetches details, extracts PDF text using utils.pdf_reader.read_pdf, prepends metadata.

  def read_paper(self, paper_id: str, save_path: str = "./downloads", start_page: int | None = None, end_page: int | None = None) -> str:
      """
      Download and extract text from IACR paper PDF
  
      Args:
          paper_id: IACR paper ID
          save_path: Directory to save downloaded PDF
          start_page: Starting page number (1-indexed, inclusive). Defaults to 1.
          end_page: Ending page number (1-indexed, inclusive). Defaults to last page.
  
      Returns:
          str: Extracted text from the PDF or error message
      """
      try:
          # First get paper details to get the PDF URL
          paper = self.get_paper_details(paper_id)
          if not paper or not paper.pdf_url:
              return f"Error: Could not find PDF URL for paper {paper_id}"
  
          # Use the read_pdf function to extract text
          text = read_pdf(paper.pdf_url, start_page, end_page)
  
          # Add paper metadata at the beginning
          metadata = f"Title: {paper.title}\n"
          metadata += f"Authors: {', '.join(paper.authors)}\n"
          metadata += f"Published Date: {paper.published_date}\n"
          metadata += f"URL: {paper.url}\n"
          metadata += f"PDF URL: {paper.pdf_url}\n"
          metadata += "=" * 80 + "\n\n"
  
          return metadata + text
  
      except Exception as e:
          logger.error(f"Error reading paper: {e}")
          return f"Error reading paper: {e}"

• src/apaper/server.py:22-23 (registration)

  FastMCP server initialization with 'apaper' prefix, implying tool names like 'apaper_read_iacr_paper'. Instantiates IACRSearcher used by the tool.

  mcp = FastMCP("apaper")