WordCraft

Fast DOCX to PDF conversion for Ruby. Ruby FFI bindings for WordCraft, a pure Rust word processor engine.

• In-process. No LibreOffice, no headless browser, no conversion service.
• Fast. A typical document converts in tens of milliseconds.
• Thread-friendly. Conversions release the GVL, so Puma and Sidekiq threads convert in parallel.
• Self-contained. Fonts are bundled; no system packages needed.
• Prebuilt. Native libraries for macOS and Linux (glibc and musl) ship inside the platform gems.

# Gemfile
gem 'wordcraft'

Requires Ruby 3.1+.

require 'wordcraft'

pdf = WordCraft.convert(File.binread('contract.docx'))
File.binwrite('contract.pdf', pdf)

WordCraft.convert(docx, markup: false) → String

Converts a DOCX document to PDF and returns the PDF bytes.

Raises WordCraft::Error when the document can't be opened or converted. RTF and .doc files are converted too.

pdf = WordCraft.convert(docx)
pdf = WordCraft.convert(docx, markup: true)

WordCraft.convert_file(input_path, output_path, markup: false) → String

Reads input_path, converts it, and writes the PDF to output_path. Returns output_path.

WordCraft.convert_file('in.docx', 'out.pdf')

WordCraft.add_font(bytes) / WordCraft.add_font_file(path) make a font available to every later conversion. Both return the number of faces added.

WordCraft.load_system_fonts loads the installed system fonts up front instead of on the first conversion. Call it at boot.

WordCraft.add_font_file('/app/fonts/Brand-Regular.ttf')
WordCraft.load_system_fonts

WordCraft::Error is raised for every conversion failure.

begin
  WordCraft.convert(upload.read)
rescue WordCraft::Error => e
  Rails.logger.warn("DOCX conversion failed: #{e.message}")
end

Send a converted upload from a Rails controller:

def create
  pdf = WordCraft.convert(params[:file].read)

  send_data pdf, type: 'application/pdf', filename: 'document.pdf'
rescue WordCraft::Error => e
  render plain: e.message, status: :unprocessable_entity
end

Convert an Active Storage attachment:

pdf = WordCraft.convert(document.file.download)

document.pdf.attach(io: StringIO.new(pdf), filename: "#{document.file.filename.base}.pdf",
                    content_type: 'application/pdf')

WordCraft uses the fonts embedded in the document, then installed system fonts, then its bundled fonts. For Word's proprietary fonts it picks metric-compatible open ones when they are installed. For page breaks closest to Word, install them:

# Alpine
RUN apk add --no-cache font-liberation font-carlito
# Debian / Ubuntu
RUN apt-get install -y --no-install-recommends fonts-liberation2 fonts-crosextra-carlito

musl's malloc slows conversions down, especially in parallel. Preload jemalloc:

RUN apk add --no-cache jemalloc
ENV LD_PRELOAD=/usr/lib/libjemalloc.so.2

Using this library to convert untrusted documents (user uploads, files from the internet) is not recommended. The conversion runs native code inside your Ruby process with no sandbox.

If you convert untrusted documents, you must use an isolated, sandboxed process.

MIT. The native library bundles WordCraft (MIT OR Apache-2.0) and the Source Sans 3, Source Serif 4, Inter and JetBrains Mono fonts (OFL-1.1); their licenses and WordCraft's attribution list are in licenses/, and WordCraft's notices in NOTICE.