# chd-rs
[![Latest Version](https://img.shields.io/crates/v/chd.svg)](https://crates.io/crates/chd) [![Docs](https://docs.rs/chd/badge.svg)](https://docs.rs/chd) ![License](https://img.shields.io/crates/l/chd)
[![Minimum Supported Rust Version 1.59](https://img.shields.io/badge/rust-1.59%2B-orange.svg)](https://github.com/rust-lang/rust/blob/master/RELEASES.md#version-1590-2022-02-24)


Reimplementation of the CHD file format in pure Safe Rust, drop-in compatible with libchdr.

chd-rs aims to be a memory-safe, well documented, and clean from-scratch implementation of CHD, verifiable against 
[chd.cpp](https://github.com/mamedev/mame/blob/master/src/lib/util/chd.cpp) while being easier to read and use as
documentation to implement the format natively in other languages. It is standalone and can be built with just
a Rust compiler, without the need for a full C/C++ toolchain. 

Performance is competitive but a little slower than libchdr in benchmarks from using more immature (but fully correct) 
pure Rust implementations of compression codecs. Deflate (zlib) compression is backed by [flate2](https://crates.io/crates/flate2), 
LZMA is backed by [lzma-rs](https://crates.io/crates/lzma-rs), and FLAC decompression is backed by
[claxon](https://crates.io/crates/claxon). While performance is not ignored, the focus
is on readability and correctness.

## Usage
Open a `Chd` with `Chd::open`, then iterate hunks from 0 to `chd.header().hunk_count()` to
read hunks.

The size of the destination buffer must be exactly `chd.header().hunk_size()` to decompress with
`hunk.read_hunk_in`, which takes the output slice and a buffer to hold compressed data.

```rust
fn main() -> Result<()> {
    let mut f = BufReader::new(File::open("image.chd")?;
    let mut chd = Chd::open(&mut f, None)?;
    let hunk_count = chd.header().hunk_count();
    let hunk_size = chd.header().hunk_size();
    
    // buffer to store decompressed hunks
    let mut out_buf = chd.get_hunksized_buffer();
    
    // buffer for temporary compressed
    let mut temp_buf = Vec::new();
    for hunk_num in 0..hunk_count {
        let mut hunk = chd.hunk(hunk_num)?;
        hunk.read_hunk_in(&mut temp_buf, &mut out_buf)?;
    }
}
```
For more ergonomic but slower usage, [`chd::read`](https://github.com/SnowflakePowered/chd-rs/blob/master/chd-rs/src/read.rs) provides buffered adapters that implement `Read` and `Seek` at the
hunk level. A buffered adapter at the file level is also available.

### Lending Iterators
With `unstable_lending_iterators`, hunks and metadata can be slightly more ergonomically iterated over
albeit with a `while let` loop. This API is unstable until [Generic Associated Types](https://github.com/rust-lang/rust/pull/96709)
and the `LendingIterator` trait is stabilized.


```toml
[dependencies]
chd = { version = "0.2", features = ["unstable_lending_iterators"] }
```

Then hunks can be iterated like so.

```rust
fn main() -> Result<()> {
    let mut f = BufReader::new(File::open("image.chd")?);
    let mut chd = Chd::open(&mut f, None)?;
    
    // buffer to store decompressed hunks
    let mut out_buf = chd.get_hunksized_buffer();
    
    // buffer for temporary compressed
    let mut temp_buf = Vec::new();
    let mut hunk_iter = chd.hunks();
    while let Some(mut hunk) = hunk_iter.next() {
        hunk.read_hunk_in(&mut temp_buf, &mut out_buf)?;
    }
}
```

A similar API exists for metadata in `Chd::metadata`.


### Verifying Hunk Checksums
By default, chd-rs does not verify the checksums of decompressed hunks for performance. The feature `verify_block_crc` should be enabled 
to verify hunk checksums.

```toml
[dependencies]
chd = { version = "0.2", features = ["verify_block_crc"] }
```

### Supported Codecs
chd-rs supports the following compression codecs, with wider coverage than libchdr. For implementation details,
see the [`chd::compression`](https://github.com/SnowflakePowered/chd-rs/tree/master/chd-rs/src/compression) module.

#### V1-4 Codecs
⚠️*V1-4 support has not been as rigorously tested as V5 support.* ⚠️
* None (`CHDCOMPRESSION_NONE`)
* Zlib (`CHDCOMPRESSION_ZLIB`)
* Zlib+ (`CHDCOMPRESSION_ZLIB`)

#### V5 Codecs
* None (`CHD_CODEC_NONE`)
* LZMA (`CHD_CODEC_LZMA`)
* Deflate (`CHD_CODEC_ZLIB`)
* FLAC (`CHD_CODEC_FLAC`)
* Huffman (`CHD_CODEC_HUFF`)
* Zstandard (`CHD_CODEC_ZSTD`)
* CD LZMA (`CHD_CODEC_CD_LZMA`)
* CD Deflate (`CHD_CODEC_CD_ZLIB`)
* CD FLAC (`CHD_CODEC_CD_FLAC`)
* CD Zstandard (`CHD_CODEC_CD_ZSTD`)
* AV Huffman (`CHD_CODEC_AVHUFF`)

#### Codecs and Huffman API 
By default, the codecs and static Huffman implementations are not exposed as part of the public API, 
but can be enabled with the `codec_api` and `huffman_api` features respectively. These APIs are subject
to change but should be considered mostly stable. 

In particular the type signature for [`HuffmanDecoder`](https://github.com/SnowflakePowered/chd-rs/blob/e03e093021f1705d46fe6aaa8b32593489e55467/chd-rs/src/huffman.rs#L110)
is subject to change once [`generic_const_exprs`](https://github.com/rust-lang/rust/issues/76560) is stabilized.

## `rchdman` command line tool
As a proof of concept, chd-rs implements an *extremely* basic reimplementation of chdman for read-only purposes. The following functions are available with rchdman.

* `info` Displays information about a CHD.
* `verify` Verify the integrity of a CHD. Metadata integrity is not verified.
* `extractraw` Extract the raw file from a CHD input file.
* `dumpmeta` Dump metadata from the CHD to stdout or to a file.

The results from rchdman should be identical from chdman. rchdman is intended to be basic and does not implement multithreading or other functions, so in general it is slower than chdman. There are
no plans to implement write-operations into rchdman.

## Performance
By default, chd-rs uses pure Rust codecs but if maximum performance is needed, `max_perf` can be enabled. This enables the zlib-ng backend of [flate2](https://crates.io/crates/flate2)
as well as using experimental APIs in a custom [lzma-rs fork](https://github.com/SnowflakePowered/lzma-rs/tree/feature-perf-experiments) for some improvements in
performance at the expense of higher memory usage. Combined with `codegen-units=1` and [Profile Guided Optimization](https://github.com/vadimcn/cargo-pgo), chd-rs is within 1% of libchdr performance.

Without `max_perf`, chd-rs is already within 15% of libchdr without needing to link with C libraries like zlib-ng.

## `libchdr` API
⚠️*The C API has not been heavily tested. Use at your own risk.* ⚠️

chd-rs provides a C API compatible with [chd.h](https://github.com/rtissera/libchdr/blob/6eeb6abc4adc094d489c8ba8cafdcff9ff61251b/include/libchdr/chd.h). 
ABI compatibility is detailed below but is untested when compiling as a dynamic library. See [/chd-rs-capi](https://github.com/SnowflakePowered/chd-rs/tree/master/chd-rs-capi) for more details.