Skip to main content

​
Overview

ToMarkdown() is the primary extraction method in PDF4LLM. It reads a document and returns its content as a Markdown string, preserving headings, lists, tables, code blocks, images, and reading order as closely as possible. 
using MuPDF.NET;
using PDF4LLM;

string mdText = PdfExtractor.ToMarkdown("document.pdf");

​
Common options

​
Page selection

Extract only specific pages by passing a list of zero-based page indices: 
// Extract pages 1, 2, and 3 (zero-based: 0, 1, 2)
string mdText = PdfExtractor.ToMarkdown(
    "document.pdf",
    pages: new List<int> { 0, 1, 2 }
);
Extract every other page by building the page list with Linq: 
// Extract every other page
Document doc         = new Document("document.pdf");
var      everyOther  = Enumerable.Range(0, doc.PageCount)
                                 .Where(i => i % 2 == 0)
                                 .ToList();

string mdText = PdfExtractor.ToMarkdown(doc, pages: everyOther);
doc.Close();
For large documents, limiting extraction to the pages you need can dramatically reduce processing time — especially when OCR is involved.

​
Per-page chunks

Use LlamaMarkdownReader to return one document object per page instead of a single concatenated string. Each chunk includes the page’s Markdown text and associated metadata: 
var reader = PdfExtractor.LlamaMarkdownReader();
var chunks = reader.LoadData("document.pdf");

foreach (var chunk in chunks)
{
    int    page = (int)chunk.ExtraInfo["page"];
    string text = chunk.Text;

    Console.WriteLine($"Page {page}");
    Console.WriteLine(text);
}

​
Headers and footers

PDF4LLM uses bounding box position to identify and exclude repeating page headers and footers. Filter them by building the page list and using ToJson to identify the margin bands, or exclude them at the chunking stage by filtering short leading and trailing lines from each page chunk. For documents with consistent header and footer heights, the most reliable approach is to filter blocks by their bounding box position using ParseDocument: 
ParsedDocument parsed = PdfExtractor.ParseDocument("document.pdf");

foreach (ParsedPage page in parsed.Pages)
{
    // Exclude blocks in the top and bottom 60pt margin bands
    var bodyBlocks = page.Blocks
        .Where(b => b.BoundingBox.Y0 > 60 && b.BoundingBox.Y1 < (page.Height - 60))
        .ToList();

    // Render body blocks only
}

​
Images

To extract embedded images and reference them inline in the Markdown output: 
string mdText = PdfExtractor.ToMarkdown(
    "document.pdf",
    writeImages:  true,
    imagePath:    "assets/images/",
    imageFormat:  "png"
);
Image references are embedded as standard Markdown image syntax: 
![image](assets/images/document.pdf-0-1.png)
See Image extraction for a full breakdown of image options.

​
Tables

Table extraction runs automatically. PDF4LLM renders detected tables as GitHub-flavoured Markdown tables: 
| Column A | Column B | Column C |
|----------|----------|----------|
| Value 1  | Value 2  | Value 3  |

​
Full example

A more complete call combining several options: 
using MuPDF.NET;
using PDF4LLM;
using System.IO;

// Ensure image output directory exists
Directory.CreateDirectory("assets/");

// Extract the first five pages with images
string mdText = PdfExtractor.ToMarkdown(
    "report.pdf",
    pages:       new List<int> { 0, 1, 2, 3, 4 },   // first five pages only
    writeImages: true,                                  // extract images to disk
    imagePath:   "assets/",                             // image output directory
    imageFormat: "png"                                  // image format
);

// Save the full output as a single Markdown file
File.WriteAllText("output/report.md", mdText, System.Text.Encoding.UTF8);
To save each page as a separate file, use LlamaMarkdownReader for per-page output: 
var reader = PdfExtractor.LlamaMarkdownReader();
var chunks = reader.LoadData("report.pdf");

Directory.CreateDirectory("output");

foreach (var chunk in chunks)
{
    int    pageNum  = (int)chunk.ExtraInfo["page"];
    string filePath = $"output/page-{pageNum}.md";

    File.WriteAllText(filePath, chunk.Text, System.Text.Encoding.UTF8);
}
For the full API signature including all parameters and return types, see the ToMarkdown() API reference.

​
Next steps

Extract JSON

Bounding boxes and layout data for custom pipelines.

Extract Text

Get clean, plain text output.