Crate nachricht[−][src]
Expand description
All encoding functions take &self
and a writer and return the amount of written bytes. All decoding functions take
a buffer and return Self
and the number of consumed bytes.
A note on usize
nachricht
internally uses 64 bit unsigned integers to signify field lengths. Rust however uses the
architecture-dependent usize
for slice indexing. This means that on architectures where usize
is smaller than
u64
(32 bit i386 for instance), some valid nachricht
messages can not be decoded since there would be no way to
efficiently index the containers. A DecodeError::Length
will be raised in these instances. Likewise, on
architectures where usize
is larger than u64
, some valid Rust datastructures can not be encoded since there is
no way to represent them in the wire format. A EncodeError::Length
will be raised in these instances.
Examples
use nachricht::*;
use std::borrow::Cow;
let mut buf = Vec::new();
let field = Field { name: Some(Cow::Borrowed("key")), value: Value::Str(Cow::Borrowed("value")) };
Encoder::encode(&field, &mut buf);
assert_eq!(buf, [
0xc3, // Key of length 3
0x6b, // 'k'
0x65, // 'e'
0x79, // 'y'
0x85, // Str of length 5
0x76, // 'v'
0x61, // 'a'
0x6c, // 'l'
0x75, // 'u',
0x65, // 'e'
]);
let decoded = Decoder::decode(&buf).unwrap();
assert_eq!(field, decoded.0);
assert_eq!(10, decoded.1);
Structs
Used to decode nachricht
fields. This uses an internal symbol table to allow the decoding of encountered
references.
Used to encode nachricht
fields. This uses an internal symbol table to allow referencing keys and symbols which
get repeated.
When encoding struct-like data structures, name
should be the identifier of the current field. When encoding
sequence-like data, name
can be omitted. Note that the type of name
is fixed at &str
, which means maps with
arbitrary key types need to be encoded as sequences.
Enums
The sign of an integer. Not that the encoder accepts negative zero but transparently translates it to positive zero. Likewise, decoders will accept the wire format for negative zero (which can only be achieved by purposefully chosing an inefficient encoding) but return positive zero, so that testing the output doesn’t need to concern itself with another special case.
The possible values according to the nachricht
data model.