Skip to content

marker_pdfs

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

marker_pdfs

README.md

Marker PDFs MCP Server

This MCP server processes individual PDF files one at a time from numbered folders in /home/ash/matdatabase/PDFs and converts them to markdown with image extraction enabled.

Features

  • Processes PDFs one file at a time from numbered folders
  • Outputs markdown to corresponding folders in /home/ash/matdatabase/pdfs_markdown
  • Image extraction is enabled by default
  • Supports processing specific folders or all folders at once

Directory Structure

  • Input: /home/ash/matdatabase/PDFs/{folder_number}/
  • Output: /home/ash/matdatabase/pdfs_markdown/{folder_number}/

Tools Available

1. convert_pdfs_to_markdown

Convert PDFs from specific folders to markdown.

Parameters:

  • folder_numbers: List of folder numbers to process (e.g., ['1', '2', '12'])
  • num_devices: Number of devices for marker conversion (default: 3)
  • num_workers: Number of workers for marker conversion (default: 7)
  • enable_images: Whether to enable image extraction (default: True)

2. convert_all_pdfs_to_markdown

Convert ALL PDFs from all folders automatically.

Parameters:

  • num_devices: Number of devices for marker conversion (default: 3)
  • num_workers: Number of workers for marker conversion (default: 7)
  • enable_images: Whether to enable image extraction (default: True)

Installation

Dependencies have been installed using uv sync. The virtual environment is located at .venv/.

Usage

Run the server:

cd /home/ash/matdatabase/marker_pdfs
export MARKER_CHUNK_CONVERT_BIN=/home/ash/matdatabase/marker/.venv/bin/marker_chunk_convert
.venv/bin/python main.py

Important: Always use .venv/bin/python to ensure the virtual environment is active.

The server will expose the MCP tools for PDF to markdown conversion.