mdict-rs

mdict-rs is a library-first Rust parser for MDict .mdx and .mdd dictionaries.

The project currently targets the widely used v2.0 layout with a strong bias toward:

checked parsing over panics
lazy key / record block decoding
library APIs before CLI tooling
clean-room interoperability instead of GPL code translation

Status

mdict-rs is ready for an early public release in the 0.1.x range.

What the current release means:

the v2.0 parser path is implemented and tested against a local corpus of real dictionaries
the public API is intentionally small
format coverage is still conservative rather than “all historical MDict variants”

Supported Scope

Currently supported:

MDX v2.0
MDD v2.0
lazy lookup and entry iteration
encrypted keyword-index dictionaries
passcode-aware keyword-header support in the API
block compression:
- uncompressed
- zlib
- optional lzo feature
encodings:
- UTF-8
- UTF-16LE
- GBK / GB18030-family
- Big5

Currently not supported:

version 1.x section layouts
writer support
HTML / stylesheet rewriting
mmap or sidecar indexes

Installation

[dependencies]
mdict-rs = "0.1"

If you need LZO-compressed blocks:

[dependencies]
mdict-rs = { version = "0.1", features = ["lzo"] }

Quick Start

MDX lookup

use mdict_rs::MdxFile;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let dict = MdxFile::open("dictionary.mdx")?;
    if let Some(record) = dict.lookup("apple")? {
        println!("{}", record.text);
    }
    Ok(())
}

MDD resource lookup

use mdict_rs::MddFile;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let dict = MddFile::open("dictionary.mdd")?;
    if let Some(resource) = dict.lookup("\\\\LM5style.css")? {
        println!("{} bytes", resource.data.len());
    }
    Ok(())
}

Iteration

use mdict_rs::MdxFile;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let dict = MdxFile::open("dictionary.mdx")?;
    for entry in dict.entries().take(10) {
        let entry = entry?;
        println!("{}", entry.key);
    }
    Ok(())
}

Validation

The current codebase is verified with:

unit tests for crypto, compression, and header parsing
malformed-input tests
mutation-style regression tests
ignored integration tests against a large local sample
ignored integration tests against the full local dictionary corpus
a separate fuzz/ crate with libFuzzer entrypoints

License

This repository is publicly available under AGPL-3.0-only.

Commercial licensing is also available from the copyright holder for users who want to embed mdict-rs without AGPL obligations.

See COMMERCIAL-LICENSING.md.

mdict-rs 0.1.1