DocReaderMCP Server

A FastMCP server designed to read and stream various file formats commonly used in organizations (PDF, DOCX, Excel XLSX, CSV, TSV, TXT). It outputs formatted Markdown and supports both segment/page limits and item streaming.

Features

• Document Formats Supported:
  • PDF: Pages parsed and converted to text.
  • DOCX: Paragraphs segmented into logical page blocks.
  • Excel (XLSX/XLS): Targeted sheet names or first sheet parsed.
  • CSV & TSV: Tables outputted in Markdown format.
  • TXT: Plain text paginated into logical blocks.
• Reading Options:
  • Optional page_range parameter (e.g. 1-3, 2) to select specific pages.
  • Optional sheet_name parameter for Excel files.
• Streaming Options:
  • Text-Style Documents (PDF, DOCX, TXT): Streamed sentence-by-sentence.
  • Tabular Documents (Excel, CSV, TSV): Streamed row-by-row as Markdown table rows.
• Docker Ready: Built with python:3.13.3-slim.

---

Tools Reference

Tool Name	Parameters	Description
read_pdf	file_path (str), page_range (Optional[str])	Read PDF file pages
stream_pdf	file_path (str), page_range (Optional[str])	Stream PDF sentences
read_docx	file_path (str), page_range (Optional[str])	Read DOCX paragraphs by page
stream_docx	file_path (str), page_range (Optional[str])	Stream DOCX sentences
read_excel	file_path (str), sheet_name (Optional[str])	Read sheet to Markdown table
stream_excel	file_path (str), sheet_name (Optional[str])	Stream sheet row-by-row
read_csv	file_path (str)	Read CSV to Markdown table
stream_csv	file_path (str)	Stream CSV row-by-row
read_tsv	file_path (str)	Read TSV to Markdown table
stream_tsv	file_path (str)	Stream TSV row-by-row
read_txt	file_path (str), page_range (Optional[str])	Read TXT by page
stream_txt	file_path (str), page_range (Optional[str])	Stream TXT sentences

---

Installation & Setup

Local Run (Venv / System Python)

1. Install dependencies:

   pip install -r requirements.txt
   

2. Start the server in standard mode:

   python main.py run
   

   Or in development mode:

   python main.py dev
   

Running with Docker

1. Build and start the container:

   docker compose up --build
   

2. By default, the container mounts a local ./data directory to /data in the container. Put your organizational documents in ./data to read them via the container.