carta

A universal document converter written in Rust. Read a markup format, render it back out in another. A performant and lightweight reimplementation of pandoc.

[!WARNING] carta is an early-stage alpha under active development. Not all of pandoc's formats are implemented yet and the API is still unstable.

Goals

• Performance and a smaller binary than pandoc.
• Feature parity with pandoc across all formats and extensions.
• A developer-friendly library, with the CLI as a thin shell over it.

Status

This tracks carta's status of all formats pandoc supports. See STATUS.md for a detailed per-format breakdown including extension coverage, and the full feature backlog.

✅ usable — basically done; any remaining parity gaps are minor and unlikely to affect regular use · 🚧 in development — large parity gaps or breaking issues (e.g. panics), not recommended for use yet · ❌ not started · ➖ not applicable (pandoc has no such direction)

Markdown family

HTML & slides

TeX & typesetting

Lightweight markup

Wikis

roff

Word processor, ebook & notebook

XML & publishing

Bibliography

Data, interchange & terminal

Installation

Prebuilt binaries

Download the archive for your platform from the latest release. Builds are provided for Linux (x86-64 gnu and static musl, arm64), macOS (Intel and Apple Silicon), and Windows (x86-64).

From crates.io

cargo install carta

This installs the carta binary. For a smaller build, pass --no-default-features with only the formats you need, e.g. --features cli,read-commonmark,write-html.

From source

git clone https://github.com/mfkrause/carta
cd carta
cargo build --release
# binary at target/release/carta

Usage

Command line

# CommonMark to HTML
carta -f commonmark -t html input.md -o output.html

# read from stdin, write to stdout
echo '# Hello' | carta -f commonmark -t html

# inspect the document model
carta -f commonmark -t json input.md

# standalone document with a table of contents and numbered sections
carta -f commonmark -t html -s --toc --number-sections input.md -o output.html

# render HTML math with MathJax (or --katex)
carta -f commonmark -t html -s --mathjax input.md -o output.html

# extract a notebook's embedded images to files, rewriting the references
carta -f ipynb -t markdown --extract-media=media notebook.ipynb -o notebook.md

# transform the document through a JSON filter before writing (repeatable, applied in order)
carta -f commonmark -t html -F ./my-filter.py input.md -o output.html

# discover what this build supports
carta --list-input-formats
carta --list-output-formats
carta --list-extensions          # extensions for the Markdown dialect
carta --list-extensions=gfm      # extensions and defaults for a given format

Library

use carta::{convert_text, ReaderOptions, WriterOptions};

let html = convert_text(
    "commonmark",
    "html",
    "# Hello, *world*",
    &ReaderOptions::default(),
    &WriterOptions::default(),
)?;

convert_text is the shortcut for text-to-text conversion. The general entry point is convert, which takes raw bytes and returns an Output that is text or bytes depending on the target format — use it when either side is a binary format.

You can select formats at compile time via per-direction features to make binaries even more lightweight for your individual needs.

cargo build -p carta --no-default-features --features read-commonmark,write-html

Development

cargo build                         # build the workspace
cargo nextest run --workspace       # run tests
cargo clippy --all-targets          # lint
cargo +nightly fuzz run commonmark  # fuzz a reader (see fuzz/README.md)

The workspace splits into carta-ast (the document model), carta-core (shared traits and options), carta-readers, carta-writers, and carta (the library facade, which also ships the command-line binary behind its cli feature).

License

Copyright © 2026 Maximilian Krause.

carta is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License, version 3, as published by the Free Software Foundation. See LICENSE for the full text.