A MoonBit-native document-to-Markdown converter with multi-format parsing, metadata, assets, batch conversion, and validation tooling

moonbit

markdown

document-to-markdown

document-conversion

pdf

docx

xlsx

pptx

epub

rag

content-ingestion

cli

Download zip

Author

ZSeanYves

Version

0.8.0

License

Apache-2.0

Repository

github.com/ZSeanYves/markitdown.git

Last updated

yesterday

Downloads

149

#markitdown

Native document-to-Markdown conversion in MoonBit for text extraction, content ingestion and RAG preparation. Supports Linux x86_64 and macOS arm64, with a command-line tool and the ZSeanYves/markitdown/lib library.

#Features

Text and data: TXT, CSV/TSV, JSON/JSONL/NDJSON, YAML, TOML, IPYNB and SRT/VTT subtitles.
Markup: Markdown, HTML, XML, TeX, RST and AsciiDoc.
Office documents: DOCX, XLSX, PPTX, ODT, ODS and ODP.
Containers and messages: ZIP, EPUB and MIME/EML.
PDF text-layer extraction, layout information, links and supported document image assets.
Markdown, debug JSON and RAG output, with metadata, source locations, provenance and recoverable image assets.
Single-file, batch and stdin conversion; asynchronous library conversion from paths, text, bytes and readers.

balance is the default mode. accurate enables additional Office/ODF semantics and PDF text processing. stream is available only for formats with a supported streaming route; ZIP accepts balance only. OCR, scanned-page recognition, audio transcription and image recognition are not supported.

#Install and use

Requires MoonBit 0.1.20261021 (moonc v0.10.17+093cb8f9c) and a C compiler. Install the Native CLI:

moon install ZSeanYves/markitdown@0.8.0

Ensure MoonBit's executable directory is on your PATH, then convert a file:

markitdown input.docx output.md
markitdown accurate input.pdf output.md
markitdown --rag input.pdf chunks.json
markitdown batch balance input_dir output_dir
cat input.jsonl | markitdown --format jsonl - output.md

Omit the output path to write to stdout. Use --format to specify the input format, --provenance-out provenance.json to save single-file provenance, and --max-temporary-bytes <bytes> to cap conversion spool storage.

markitdown --help
markitdown --capabilities

#Library usage

Add the library dependency from your consumer directory:

moon add ZSeanYves/markitdown@0.8.0

Set preferred_target = "native" in moon.mod and import ZSeanYves/markitdown/lib in moon.pkg:

async fn convert_example() -> Result[@lib.Output, @lib.ConvertError] {
  let input = @lib.Input::from_path("document.docx")
  @lib.convert(input)
}

Source Files

main.mbt

Site source Report issue Packages Build queue Skills Statistics