Crates.io | ethercrab-wire-derive |
lib.rs | ethercrab-wire-derive |
version | 0.2.0 |
source | src |
created_at | 2024-01-11 11:42:51.022141 |
updated_at | 2024-07-28 09:49:14.438632 |
description | Derive macros for EtherCrab |
homepage | |
repository | https://github.com/ethercrab-rs/ethercrab |
max_upload_size | |
id | 1096209 |
size | 58,633 |
Derive attributes for ethercrab-wire
.
This crate is in its early stages and may contain bugs or publish breaking changes at any time.
It is in use by ethercrab
and is well exercised there, but please use with caution in your
own code.
These derives support both structs with bit- and (multi)byte-sized fields for structs, as well as enums with optional catch-all variant.
#[wire(bits = N)]
OR #[wire(bytes = N)]
The size of this struct when packed on the wire. These attributes may not be present at the same time.
#[wire(bits = N)]
OR #[wire(bytes = N)]
How many bytes this field consumes on the wire. These attributes may not be present at the same time.
#[wire(pre_skip = N)]
OR #[wire(pre_skip_bytes = N)]
Skip one or more whole bytes before or after this field in the packed representation.
#[wire(post_skip = N)]
OR #[wire(post_skip_bytes = N)]
How many bits or bytes to skip in the raw data after this field.
These attributes are only applicable to fields that are less than 8 bits wide.
Enums must have a #[repr()]
attribute, as well as implement the Copy
trait.
Enum discriminants may not contain fields.
#[wire(alternatives = [])]
A discriminant with this attribute will be parsed successfully if either its direct value or any of the listed alternatives are found in the input data.
The discriminant value is used when packing to the wire.
#[wire(catch_all)]
Apply this once to a discriminant with a single unnamed field the same type as the enum's
#[repr()]
to catch any unrecognised values.
#[derive(ethercrab_wire::EtherCrabWireReadWrite)]
#[wire(bytes = 4)]
struct Mixed {
#[wire(bits = 1)]
one_bit: u8,
#[wire(bits = 2)]
two_bits: u8,
// Fields that are 8 bits or larger must be byte aligned, so we skip the two remaining bits
// of the previous byte with `post_skip`.
#[wire(bits = 3, post_skip = 2)]
three_bits: u8,
/// Whole `u8`
#[wire(bytes = 1)]
one_byte: u8,
/// Whole `u16`
#[wire(bytes = 2)]
one_word: u16,
}
#[derive(Copy, Clone, ethercrab_wire::EtherCrabWireReadWrite)]
#[repr(u8)]
enum OneByte {
Foo = 0x01,
#[wire(alternatives = [ 3, 4, 5, 6 ])]
Bar = 0x02,
Baz = 0x07,
Quux = 0xab,
#[wire(catch_all)]
Unknown(u8),
}
// Normal discriminant
assert_eq!(OneByte::unpack_from_slice(&[0x07]), Ok(OneByte::Baz));
// Alternative value for `Bar`
assert_eq!(OneByte::unpack_from_slice(&[0x05]), Ok(OneByte::Bar));
// Catch all
assert_eq!(OneByte::unpack_from_slice(&[0xaa]), Ok(OneByte::Unknown(0xaa)));
Struct fields of 1 byte or more MUST be byte-aligned. For example, the following struct will be
rejected due to bar
being 5 bits "early":
#[derive(ethercrab_wire::EtherCrabWireReadWrite)]
#[wire(bytes = 2)]
struct Broken {
#[wire(bits = 3)]
foo: u8,
// There are 5 bits here unaccounted for
#[wire(bytes = 1)]
bar: u8,
}
This can easily be fixed by using the pre_skip
or post_skip
attributes to realign the next
field to 8 bits (or skip whole bytes of the input data):
#[derive(ethercrab_wire::EtherCrabWireReadWrite)]
#[wire(bytes = 2)]
struct Fixed {
#[wire(bits = 3, post_skip = 5)]
foo: u8,
#[wire(bytes = 1)]
bar: u8,
}
A field in the middle of a byte can be written as such, maintaining 8 bit alignment:
#[derive(ethercrab_wire::EtherCrabWireReadWrite)]
#[wire(bytes = 1)]
struct Middle {
#[wire(pre_skip = 2, bits = 3, post_skip = 3)]
foo: u8,
}
Licensed under either of
at your option.