1 unstable release
0.1.0 | Dec 7, 2020 |
---|
#2457 in Encoding
15KB
140 lines
The main goal of this crate is to simplify serialization of type's and structures to bytes.
Why?
If one works with binary formats/protocols a lot of time is spent implementing decoding and encoding types and structures of the format/protocol in order to further process the contained data.
For decoding parsers generators like nom are very helpful and easy the implmentation. This create tries to provide a lightweight encoding/output conbinator by just introducing 2 new traits which in turn then can make use of the iterator facilites to crate the desired output chain of bytes.
Why this extra step with the trait's
-
By introducing such a trait complex (compsites) structures in a lot of cases can be implemented by just encoding and chaining the childs in order
-
The fileds of a type still can be used for encoding but there is no hard dependency on their order nor their actual size
e.g. a protocol field size with the encoded size of 2 Bytes (u16), still can be represented e.g. as usize withn the structures/type which save quite some converting and casting.
-
There is no need of a type to provide a specific amount of memory in order to be serialized (the serialization of a type or a type could be 100% computational)
e.g.: assume this protocol type/structure (Packet)
+-----------------+-------------------+-----------------+ | field1 (1 Byte) | reserved (7 Byte) | filed2 (8 Byte) | +-----------------+-------------------+-----------------+
internally it could be represented and implemented like this
use tobytes::ByteView; use tobytes::ToBytes; struct Packet { field1: u8, field2: u64 } impl Packet { const RESERVED : u8 = 0x00; } impl ByteView for Packet { fn byte_at(&self, index: usize) -> Option<u8> { if index < ByteView::byte_size(self) { match index { 0 => self.field1.byte_at(index), 1..=7 => Some(Packet::RESERVED), 8..=15 => self.field2.byte_at(index -7), _ => None } } else { None } } fn byte_size(&self) -> usize { ByteView::byte_size(&self.field1) + 7usize + ByteView::byte_size(&self.field2) } } let field1 = 0xaau8; let field2 = 0xaabbccddeeff11u64.to_be(); let p = Packet {field1, field2}; let mut bytes = p.to_bytes(); assert_eq!(16usize, p.byte_size()); assert_eq!( vec![0xaa, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, 0xff, 0x11], bytes.collect::<Vec<u8>>() );
How? (Usage)
Example(s)
How to serialize integers of different endianess and size
use tobytes::ByteView;
use tobytes::ToBytes;
let uint16_be : u16 = 0x0A0Bu16.to_be();
let uint16_le : u16 = 0x0C0Du16.to_le();
let uint32_le : u32 = 0x01020304u32.to_le();
let uint16_be_bytes = uint16_be.to_bytes();
let uint16_le_bytes = uint16_le.to_bytes();
let uint32_le_bytes = uint32_le.to_bytes();
let mut bytes = uint16_be_bytes.chain(uint16_le_bytes.chain(uint32_le_bytes));
assert_eq!(vec![0x0A, 0x0B, 0x0D, 0x0C, 0x04, 0x03, 0x02, 0x01], bytes.collect::<Vec<u8>>())
How to serialize a custom type which contains different endinesses and types
TBD
How to serialize a custom type which contains types which also implent the ByteView trait
TBD
Todo's
- Implement ByteView for built in integer and float types
- Implement ByteView for slice like types
- Implement derive macro for ByteView