Skip to content

README.md

Latest commit

 

History

History
95 lines (67 loc) · 3.43 KB

README.md

File metadata and controls

95 lines (67 loc) · 3.43 KB

MuPDF.NET4LLM

LLM/RAG helpers for MuPDF.NET: PDF-to-Markdown conversion, layout parsing, document structure analysis. Designed for use with RAG pipelines and integration with LLMs.

Installation

dotnet add package MuPDF.NET4LLM

MuPDF.NET4LLM depends on MuPDF.NET; it is installed automatically.

Features

  • PDF-to-Markdown — Convert PDF pages to Markdown with layout awareness (tables, headers, images)
  • Layout parsing — Extract document structure (pages, boxes, tables, images) as JSON or structured objects
  • Plain text extraction — Same layout analysis as Markdown, without syntax
  • LlamaIndex integration — PDFMarkdownReader for compatibility with LlamaIndex document loading
  • OCR support — Optional OCR for scanned or image-heavy pages
  • Form fields — Extract key/value pairs from interactive PDF forms

Quick Start

Convert PDF to Markdown

using MuPDF.NET;
using MuPDF.NET4LLM;

Document doc = new Document("document.pdf");
string markdown = MuPDF4LLM.ToMarkdown(doc);
doc.Close();

Convert to plain text

string text = MuPDF4LLM.ToText(doc);

Get layout as JSON

string json = MuPDF4LLM.ToJson(doc);

Use with LlamaIndex-style loading

var reader = MuPDF4LLM.LlamaMarkdownReader();
var docs = reader.LoadData("document.pdf", extraInfo: new Dictionary<string, object>());
foreach (var d in docs)
{
    Console.WriteLine($"Page {d.ExtraInfo["page"]}: {d.Text}");
}

Extract form field values

var keyValues = MuPDF4LLM.GetKeyValues(doc);

API Overview

Method Description
ToMarkdown() Convert document (or selected pages) to Markdown with optional images
ToText() Convert to plain text using layout analysis
ToJson() Export layout structure as JSON
ParseDocument() Return a ParsedDocument with pages, boxes, tables, images
LlamaMarkdownReader() Create a LlamaIndex-compatible PDF reader
GetKeyValues() Extract form field name/value pairs and page locations

Options

ToMarkdown, ToText, and ToJson support options such as:

  • pages — Restrict to specific pages (0-based)
  • writeImages / embedImages — Save or embed images
  • imagePath, imageFormat — Where and how to store images
  • useOcr, ocrLanguage — OCR for scanned content
  • showProgress — Log progress while processing
  • forceText — Prefer text extraction over image backgrounds

Requirements

  • .NET Standard 2.0 or later (net461, net472, net48, net5.0, net6.0, net7.0, net8.0)
  • MuPDF.NET 3.2.13 or newer

Note: If you see "An assembly with the same simple name 'MuPDF.NET4LLM' has already been imported", the MuPDF.NET package you have includes MuPDF.NET4LLM. Use either MuPDF.NET alone (which has 4LLM bundled) or add only MuPDF.NET4LLM (which brings MuPDF.NET). Do not add both if MuPDF.NET already bundles 4LLM. A future MuPDF.NET release will exclude the bundle so MuPDF.NET4LLM can be used as a separate package without conflict.

License

MuPDF.NET4LLM is part of MuPDF.NET and is available under the Artifex Community License and commercial license agreements. For commercial use, please contact Artifex.