Crate scale_decode

parity-scale-codec provides a Decode trait which allows bytes to be scale decoded into types based on the shape of those types. This crate builds on this, and allows bytes to be decoded into types based on type information, rather than the shape of the target type. At a high level, this crate just aims to do the reverse of the scale-encode crate.

This crate exposes four traits:

• A visitor::Visitor trait which when implemented on some type, can be used in conjunction with visitor::decode_with_visitor to decode SCALE encoded bytes based on some type information into some arbitrary type.
• An IntoVisitor trait which can be used to obtain the visitor::Visitor implementation for some type.
• A DecodeAsType trait which is implemented for types which implement IntoVisitor, and provides a high level interface for decoding SCALE encoded bytes into some type with the help of a type ID and a scale_type_resolver::TypeResolver.
• A DecodeAsFields trait which when implemented on some type, describes how SCALE encoded bytes can be decoded into it with the help of an iterator of Fields and a type registry describing the shape of the encoded bytes. This is generally only implemented for tuples and structs, since we need a set of fields to map to the provided slices.

Implementations for many built-in types are also provided for each trait, and the DecodeAsType macro can be used to generate the relevant impls on new struct and enum types such that they get a DecodeAsType impl.

The DecodeAsType and DecodeAsFields traits are basically the mirror of scale-encode’s EncodeAsType and EncodeAsFields traits in terms of their interface.

Motivation

By de-coupling the shape of a type from how bytes are decoded into it, we make it much more likely that the decoding will succeed, and are no longer reliant on types having a precise layout in order to be decoded into correctly. Some examples of this follow.

use codec::Encode;
use scale_decode::DecodeAsType;
use scale_info::{PortableRegistry, TypeInfo};
use std::fmt::Debug;

// We normally expect to have type information to hand, but for our examples
// we construct type info from any type that implements `TypeInfo`.
fn get_type_info<T: TypeInfo + 'static>() -> (u32, PortableRegistry) {
    let m = scale_info::MetaType::new::<T>();
    let mut types = scale_info::Registry::new();
    let ty = types.register_type(&m);
    let portable_registry: PortableRegistry = types.into();
    (ty.id, portable_registry)
}

// Encode the left value statically.
// Decode those bytes into the right type via `DecodeAsType`.
// Assert that the decoded bytes are identical to the right value.
fn assert_decodes_to<A, B>(a: A, b: B)
where
    A: Encode + TypeInfo + 'static,
    B: DecodeAsType + PartialEq + Debug,
{
    let (type_id, types) = get_type_info::<A>();
    let a_bytes = a.encode();
    let new_b = B::decode_as_type(&mut &*a_bytes, type_id, &types).unwrap();
    assert_eq!(b, new_b);
}

// Start simple; a u8 can DecodeAsType into a u64 and vice versa. Numbers will all
// try to convert into the desired output size, failing if this isn't possible:
assert_decodes_to(123u64, 123u8);
assert_decodes_to(123u8, 123u64);

// Compact decoding is also handled "under the hood" by DecodeAsType, so no "compact"
// annotations are needed on values.
assert_decodes_to(codec::Compact(123u64), 123u64);

// Enum variants are lined up by variant name, so no explicit "index" annotation are
// needed either; DecodeAsType will take care of it.
#[derive(Encode, TypeInfo)]
enum Foo {
    #[codec(index = 10)]
    Something(u64),
}
#[derive(DecodeAsType, PartialEq, Debug)]
enum FooTarget {
    Something(u128),
}
assert_decodes_to(Foo::Something(123), FooTarget::Something(123));

// DecodeAsType will skip annotated fields and not look for them in the encoded bytes.
// #[codec(skip)] and #[decode_as_type(skip)] both work.
#[derive(Encode, TypeInfo)]
struct Bar {
    a: bool,
}
#[derive(DecodeAsType, PartialEq, Debug)]
struct BarTarget {
    a: bool,
    #[decode_as_type(skip)]
    b: String,
}
assert_decodes_to(
    Bar { a: true },
    BarTarget { a: true, b: String::new() },
);

// DecodeAsType impls will generally skip through any newtype wrappers.
#[derive(DecodeAsType, Encode, TypeInfo, PartialEq, Debug)]
struct Wrapper {
    value: u64
}
assert_decodes_to(
    (Wrapper { value: 123 },),
    123u64
);
assert_decodes_to(
    123u64,
    (123,)
);

// Things like arrays and sequences are generally interchangeable despite the
// encoding format being slightly different:
assert_decodes_to([1u8,2,3,4,5], vec![1u64,2,3,4,5]);
assert_decodes_to(vec![1u64,2,3,4,5], [1u8,2,3,4,5]);

If this high level interface isn’t suitable, you can implement your own visitor::Visitor’s. These can support zero-copy decoding (unlike the higher level DecodeAsType interface), and generally the Visitor construction and execution is zero alloc, allowing for efficient type based decoding.

Re-exports

• pub use crate::error::Error;
• pub use visitor::Visitor;

Modules

• error
  An error that is emitted whenever some decoding fails.
• ext
  Re-exports of external crates.
• visitor
  The Visitor trait and associated types.

Structs

• Field
  Information about a composite field.

Traits

• DecodeAsFields
  This is similar to DecodeAsType, except that it’s instead implemented for types that can be given a list of fields denoting the type being decoded from and attempt to do this decoding. This is generally implemented just for tuple and struct types, and is automatically implemented via the DecodeAsType macro.
• DecodeAsType
  This trait is implemented for any type T where T implements IntoVisitor and the errors returned from this Visitor can be converted into Error. It’s essentially a convenience wrapper around visitor::decode_with_visitor that mirrors scale-encode’s EncodeAsType.
• FieldIter
  An iterator over a set of fields.
• IntoVisitor
  This trait can be implemented on any type that has an associated Visitor responsible for decoding SCALE encoded bytes to it whose error type is Error. Anything that implements this trait gets a DecodeAsType implementation for free.
• TypeResolver
  This trait can be implemented for any type that is capable of describing how some type (identified by a TypeResolver::TypeId) is represented in terms of SCALE encoded bytes.

Derive Macros

• DecodeAsType
  The DecodeAsType derive macro can be used to implement DecodeAsType on structs and enums whose fields all implement DecodeAsType. Under the hood, the macro generates scale_decode::visitor::Visitor and scale_decode::IntoVisitor implementations for each type (as well as an associated Visitor struct), which in turn means that the type will automatically implement scale_decode::DecodeAsType.