gecko-dev/third_party/rust/scroll

Scroll - cast some magic

         _______________
    ()==(              (@==()
         '______________'|
           |             |
           |   ἀρετή     |
         __)_____________|
    ()==(               (@==()
         '--------------'

Documentation

https://docs.rs/scroll

Usage

Add to your Cargo.toml

[dependencies]
scroll = "0.10"

Overview

Scroll implements several traits for read/writing generic containers (byte buffers are currently implemented by default). Most familiar will likely be the Pread trait, which at its basic takes an immutable reference to self, an immutable offset to read at, (and a parsing context, more on that later), and then returns the deserialized value.

Because self is immutable, all reads can be performed in parallel and hence are trivially parallelizable.

A simple example demonstrates its flexibility:

use scroll::{ctx, Pread, LE};

fn parse() -> Result<(), scroll::Error> {
    let bytes: [u8; 4] = [0xde, 0xad, 0xbe, 0xef];

    // reads a u32 out of `b` with the endianness of the host machine, at offset 0, turbofish-style
    let number: u32 = bytes.pread::<u32>(0)?;
    // ...or a byte, with type ascription on the binding.
    let byte: u8 = bytes.pread(0)?;

    //If the type is known another way by the compiler, say reading into a struct field, we can omit the turbofish, and type ascription altogether!

    // If we want, we can explicitly add a endianness to read with by calling `pread_with`.
    // The following reads a u32 out of `b` with Big Endian byte order, at offset 0
    let be_number: u32 = bytes.pread_with(0, scroll::BE)?;
    // or a u16 - specify the type either on the variable or with the beloved turbofish
    let be_number2 = bytes.pread_with::<u16>(2, scroll::BE)?;

    // Scroll has core friendly errors (no allocation). This will have the type `scroll::Error::BadOffset` because it tried to read beyond the bound
    let byte: scroll::Result<i64> = bytes.pread(0);

    // Scroll is extensible: as long as the type implements `TryWithCtx`, then you can read your type out of the byte array!

    // We can parse out custom datatypes, or types with lifetimes
    // if they implement the conversion trait `TryFromCtx`; here we parse a C-style \0 delimited &str (safely)
    let hello: &[u8] = b"hello_world\0more words";
    let hello_world: &str = hello.pread(0)?;
    assert_eq!("hello_world", hello_world);

    // ... and this parses the string if its space separated!
    use scroll::ctx::*;
    let spaces: &[u8] = b"hello world some junk";
    let world: &str = spaces.pread_with(6, StrCtx::Delimiter(SPACE))?;
    assert_eq!("world", world);
    Ok(())
}

fn main() {
  parse().unwrap();
}

Deriving Pread and Pwrite

Scroll implements a custom derive that can provide Pread and Pwrite implementations for your structs.

#[macro_use]
extern crate scroll_derive;

use scroll::{Pread, Pwrite, BE};

#[derive(Pread, Pwrite)]
struct Data {
    one: u32,
    two: u16,
    three: u8,
}

fn parse() -> Result<(), scroll::Error> {
    let bytes: [u8; 7] = [0xde, 0xad, 0xbe, 0xef, 0xfa, 0xce, 0xff];
    // Read a single `Data` at offset zero in big-endian byte order.
    let data: Data = bytes.pread_with(0, BE)?;
    assert_eq!(data.one, 0xdeadbeef);
    assert_eq!(data.two, 0xface);
    assert_eq!(data.three, 0xff);

    // Write it back to a buffer
    let mut out: [u8; 7] = [0; 7];
    out.pwrite_with(data, 0, BE)?;
    assert_eq!(bytes, out);
    Ok(())
}

fn main() {
  parse().unwrap();
}

This feature is not enabled by default, you must enable the derive feature in Cargo.toml to use it:

[dependencies]
scroll = { version = "0.10", features = ["derive"] }

std::io API

Scroll can also read/write simple types from a std::io::Read or std::io::Write implementor. The built-in numeric types are taken care of for you. If you want to read a custom type, you need to implement the FromCtx (how to parse) and SizeWith (how big the parsed thing will be) traits. You must compile with default features. For example:

use std::io::Cursor;
use scroll::IOread;

fn parse_io() -> Result<(), scroll::Error> {
    let bytes_ = [0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, 0xef,0xbe,0x00,0x00,];
    let mut bytes = Cursor::new(bytes_);

    // this will bump the cursor's Seek
    let foo = bytes.ioread::<usize>()?;
    // ..ditto
    let bar = bytes.ioread::<u32>()?;
    Ok(())
}

fn main() {
  parse_io().unwrap();
}

Similarly, we can write to anything that implements std::io::Write quite naturally:

use scroll::{IOwrite, LE, BE};
use std::io::{Write, Cursor};

fn write_io() -> Result<(), scroll::Error> {
    let mut bytes = [0x0u8; 10];
    let mut cursor = Cursor::new(&mut bytes[..]);
    cursor.write_all(b"hello")?;
    cursor.iowrite_with(0xdeadbeef as u32, BE)?;
    assert_eq!(cursor.into_inner(), [0x68, 0x65, 0x6c, 0x6c, 0x6f, 0xde, 0xad, 0xbe, 0xef, 0x0]);
    Ok(())
}

fn main() {
  write_io().unwrap();
}

Advanced Uses

Scroll is designed to be highly configurable - it allows you to implement various context (Ctx) sensitive traits, which then grants the implementor automatic uses of the Pread and/or Pwrite traits.

For example, suppose we have a datatype and we want to specify how to parse or serialize this datatype out of some arbitrary byte buffer. In order to do this, we need to provide a TryFromCtx impl for our datatype.

In particular, if we do this for the [u8] target, using the convention (usize, YourCtx), you will automatically get access to calling pread_with::<YourDatatype> on arrays of bytes.

use scroll::{ctx, Pread, BE, Endian};

struct Data<'a> {
  name: &'a str,
  id: u32,
}

// note the lifetime specified here
impl<'a> ctx::TryFromCtx<'a, Endian> for Data<'a> {
  type Error = scroll::Error;
  // and the lifetime annotation on `&'a [u8]` here
  fn try_from_ctx (src: &'a [u8], endian: Endian)
    -> Result<(Self, usize), Self::Error> {
    let offset = &mut 0;
    let name = src.gread::<&str>(offset)?;
    let id = src.gread_with(offset, endian)?;
    Ok((Data { name: name, id: id }, *offset))
  }
}

fn parse_data() -> Result<(), scroll::Error> {
    let bytes = b"UserName\x00\x01\x02\x03\x04";
    let data = bytes.pread_with::<Data>(0, BE)?;
    assert_eq!(data.id, 0x01020304);
    assert_eq!(data.name.to_string(), "UserName".to_string());
    Ok(())
}

fn main() {
  parse_data().unwrap();
}

Please see the official documentation, or a simple example for more.

Contributing

Any ideas, thoughts, or contributions are welcome!