ironpress

Pure Rust HTML/CSS/Markdown to PDF converter. No browser, no system dependencies.

Other Rust PDF crates shell out to headless Chrome or wkhtmltopdf. ironpress does it natively with a built-in layout engine. No C libraries, no binaries to install, just cargo add ironpress.

Table of Contents

• Quick Start
• API Reference
• Markdown to PDF
• HTML Elements
• CSS Support
• Images
• Tables
• Fonts
• Streaming Output
• Async API
• Security
• How It Works
• What's Next
• License

Quick Start

use ironpress::html_to_pdf;

let pdf_bytes = html_to_pdf("<h1>Hello</h1><p>World</p>").unwrap();
std::fs::write("output.pdf", pdf_bytes).unwrap();

API Reference

One-liner functions

// HTML string to PDF bytes
let pdf = ironpress::html_to_pdf("<h1>Title</h1><p>Content</p>").unwrap();

// Markdown string to PDF bytes
let pdf = ironpress::markdown_to_pdf("# Title\n\nContent").unwrap();

// HTML file to PDF file
ironpress::convert_file("input.html", "output.pdf").unwrap();

// Markdown file to PDF file
ironpress::convert_markdown_file("input.md", "output.pdf").unwrap();

Builder API

use ironpress::{HtmlConverter, PageSize, Margin};

let pdf = HtmlConverter::new()
    .page_size(PageSize::LETTER)        // default: A4
    .margin(Margin::uniform(54.0))      // default: 72pt (1 inch)
    .sanitize(false)                    // default: true
    .convert("<h1>Custom page</h1>")
    .unwrap();

Custom fonts

use ironpress::HtmlConverter;

let ttf_data = std::fs::read("fonts/MyFont.ttf").unwrap();
let pdf = HtmlConverter::new()
    .add_font("MyFont", ttf_data)
    .convert(r#"<p style="font-family: MyFont">Custom font text</p>"#)
    .unwrap();

Page sizes

use ironpress::PageSize;

PageSize::A4         // 595.28 x 841.89 pt (default)
PageSize::LETTER     // 612.0 x 792.0 pt
PageSize::LEGAL      // 612.0 x 1008.0 pt
PageSize::new(width_pt, height_pt)  // custom

Margins

use ironpress::Margin;

Margin::default()                    // 72pt on all sides (1 inch)
Margin::uniform(54.0)               // same value on all sides
Margin::new(top, right, bottom, left)  // individual values in pt

Markdown to PDF

Built-in Markdown parser with zero external dependencies.

let pdf = ironpress::markdown_to_pdf(r#"
# Project Title

Some **bold** and *italic* text with `inline code`.

## Features

- Item one
- Item two
- Item three

1. First
2. Second

> A wise quote

---

[Link text](https://example.com)
"#).unwrap();

Supported Markdown syntax: headings (# to ######), bold (**), italic (*), bold+italic (***), inline code, fenced code blocks, links, images, unordered lists (-, *, +), ordered lists, blockquotes, and horizontal rules.

HTML Elements

CSS Support

Properties

All shorthand properties are supported. Margin and padding accept 1, 2, 3, or 4 values. Border accepts width style color shorthand.

<style> blocks

<style>
  p { color: navy; font-size: 14pt }
  .highlight { background-color: yellow; font-weight: bold }
  #title { font-size: 24pt }
  h1, h2 { color: darkblue }

  @media print {
    .screen-only { display: none }
  }
</style>

Selectors

Values

Media queries

@media print rules are applied (since PDF is print output). @media screen rules are ignored.

@page rule

Control page size and margins from CSS:

<style>
  @page { size: letter landscape; margin: 0.5in; }
</style>

Supported values: A4, letter, legal, landscape, custom dimensions (210mm 297mm), and individual margins.

Images

JPEG and PNG images are supported via data URIs and local file paths.

<!-- Data URI -->
<img src="data:image/jpeg;base64,/9j/4AAQ..." width="200" height="150">

<!-- Local file -->
<img src="photo.jpg" width="300" height="200">

Images are embedded directly in the PDF. JPEG uses DCTDecode, PNG uses FlateDecode with PNG predictors. Width and height attributes are converted from px to pt.

Tables

Full table support with sections, spanning, auto-sized columns, and styling.

<table>
  <thead>
    <tr><th>Name</th><th>Role</th><th>Status</th></tr>
  </thead>
  <tbody>
    <tr>
      <td rowspan="2">Alice</td>
      <td>Engineer</td>
      <td>Active</td>
    </tr>
    <tr>
      <td colspan="2">On project X</td>
    </tr>
    <tr>
      <td>Bob</td>
      <td>Designer</td>
      <td>Active</td>
    </tr>
  </tbody>
</table>

Column widths are automatically calculated based on content. Features: <thead>, <tbody>, <tfoot> sections, colspan and rowspan attributes, bold headers in <th>, cell borders, background colors, and padding.

Fonts

Standard fonts

ironpress includes the 14 standard PDF fonts (no embedding required). CSS font-family values are mapped to the closest match:

Each family includes regular, bold, italic, and bold-italic variants (12 fonts total).

Custom fonts (TrueType)

Embed any TTF font for pixel-perfect rendering:

use ironpress::HtmlConverter;

let font = std::fs::read("fonts/Inter.ttf").unwrap();
let pdf = HtmlConverter::new()
    .add_font("Inter", font)
    .convert(r#"<p style="font-family: Inter">Rendered with Inter</p>"#)
    .unwrap();

The TTF parser extracts character metrics for accurate text wrapping and embeds the font directly in the PDF.

Streaming Output

Write PDF output directly to any std::io::Write implementation instead of allocating a Vec<u8>:

use std::fs::File;

let mut file = File::create("output.pdf").unwrap();
ironpress::html_to_pdf_writer("<h1>Hello</h1>", &mut file).unwrap();

Also available on the builder:

use ironpress::HtmlConverter;
use std::fs::File;

let mut file = File::create("output.pdf").unwrap();
HtmlConverter::new()
    .convert_to_writer("<h1>Hello</h1>", &mut file)
    .unwrap();

Async API

Enable the async feature for async file I/O:

ironpress = { version = "0.5", features = ["async"] }

ironpress::convert_file_async("input.html", "output.pdf").await.unwrap();
ironpress::convert_markdown_file_async("input.md", "output.pdf").await.unwrap();

The HTML parsing, layout, and rendering remain synchronous (CPU-bound). Async is used for file reads and writes via tokio.

Security

HTML is sanitized by default before conversion:

• <script>, <iframe>, <object>, <embed>, <form> tags are stripped
• <style> tags are preserved but dangerous CSS (@import, external url(), expression()) is removed
• Event handlers (onclick, onload, etc.) are removed
• javascript: URLs are neutralized
• Input size (10 MB) and nesting depth (100 levels) are limited

Sanitization can be disabled with .sanitize(false) if you trust the input.

How It Works

Input --> Sanitize --> Parse (html5ever) --> Extract <style> --> Style cascade --> Layout engine --> PDF

1. Sanitize the input HTML to remove dangerous elements
2. Parse HTML into a DOM tree using html5ever, extracting <style> blocks
3. Resolve styles by cascading: tag defaults, then @media print rules, then stylesheet rules, then inline CSS
4. Layout elements with text wrapping, float positioning, page breaks, tables, lists, images, and the CSS box model
5. Render to PDF 1.4 with text, graphics, link annotations, embedded images, and custom fonts

For Markdown input, a built-in parser converts Markdown to HTML first (no external dependencies).

What's Next

ironpress focuses on being the best HTML/CSS/Markdown to PDF engine in Rust. Other input formats (SVG, DOCX, EPUB, CSV) will be available as separate crates in the ironpress ecosystem.

Remaining work for HTML/CSS rendering:

• Rounded corners rendering (border-radius is parsed but not yet drawn as curves)
• z-index stacking order
• list-style-type, list-style-position
• background-position, background-size, background-repeat
• text-overflow: ellipsis
• Hyphenation and advanced text shaping
• PDF bookmarks from heading structure
• WASM support for browser-side PDF generation

License

MIT