Start work on better sehn base

This commit is contained in:
James Dyson 2019-01-20 14:26:37 +00:00
parent e49e427e5a
commit a202514b52
Signed by: avitex
GPG Key ID: 38C76CBF3749D62C
20 changed files with 837 additions and 0 deletions

View File

@ -1,5 +1,6 @@
[workspace]
members = [
"sehn",
"sehn-serde",
"sehn-std"
]

8
sehn/Cargo.toml Normal file
View File

@ -0,0 +1,8 @@
[package]
name = "sehn"
version = "0.1.0"
authors = ["James Dyson <theavitex@gmail.com>"]
edition = "2018"
[dependencies]
itoa = "0.4"

38
sehn/examples/basic.rs Normal file
View File

@ -0,0 +1,38 @@
extern crate sehn;
use std::collections::BTreeMap;
use sehn::*;
fn print_and_clear(s: BaseSerializer<DefaultConfig, &mut StringWriter>) {
let out = s.into_inner();
println!("{}", out.as_str().unwrap());
out.clear();
}
fn main() {
let mut w = StringWriter::new();
///////////////////////////////////////////////////////////////////////////
let mut s = default_base_serializer(&mut w);
let mut obj = BTreeMap::new();
obj.insert("hello", Tag::new("one", vec![1]));
obj.insert("world", Tag::new("two", vec![2]));
s.serialize_any(obj).unwrap();
print_and_clear(s);
///////////////////////////////////////////////////////////////////////////
let mut s = default_base_serializer(&mut w);
let lines = vec![
"hello",
"world"
];
s.serialize_multiline_text(lines).unwrap();
print_and_clear(s);
}

48
sehn/src/de/base/mod.rs Normal file
View File

@ -0,0 +1,48 @@
use super::{
Read,
Deserializer
};
pub struct BaseDeserializer<R> {
src: R,
peeked_byte: Option<u8>
}
impl<'de, R> BaseDeserializer<R>
where R: Read<'de>
{
pub fn new(src: R) -> Self {
Self {
src,
peeked_byte: None
}
}
fn peek_byte(&mut self) -> Result<u8, R::Error> {
match self.peeked_byte {
Some(byte) => Ok(byte),
None => {
let byte = self.src.read()?;
self.peeked_byte = Some(byte);
Ok(byte)
}
}
}
fn read_byte(&mut self) -> Result<u8, R::Error> {
match self.peeked_byte.take() {
Some(byte) => Ok(byte),
None => self.src.read()
}
}
}
impl<'de, R> Deserializer<'de> for &mut BaseDeserializer<R>
where R: Read<'de>
{
type Error = R::Error;
// fn deserialize_str(self) -> Result<&'de str, Self::Error> {
// }
}

20
sehn/src/de/error.rs Normal file
View File

@ -0,0 +1,20 @@
use core::fmt;
pub enum DeserializerError {
/// A unexpected error (should never get this).
Unexpected(&'static str)
}
impl fmt::Debug for DeserializerError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
use DeserializerError::*;
match self {
Unexpected(ref err) => write!(f, "unexpected error: {}", err)
}
}
}
pub trait Error: fmt::Debug + From<DeserializerError> {}
impl<T> Error for T where
T: fmt::Debug + From<DeserializerError> {}

52
sehn/src/de/mod.rs Normal file
View File

@ -0,0 +1,52 @@
mod base;
mod error;
pub use self::base::*;
pub use self::error::*;
pub trait Read<'a> {
/// The error that a read can return.
type Error: Error;
/// Read a single byte.
fn read(&mut self) -> Result<u8, Self::Error>;
/// Start mark.
fn start_mark(&mut self);
/// Zero-copy drain from mark.
fn drain_mark(&mut self) -> Result<&'a [u8], Self::Error>;
}
pub trait Deserialize: Sized {
fn deserialize<'de, D>(d: D) -> Result<Self, D::Error>
where D: Deserializer<'de>;
}
// pub trait Unit {
// fn from_name_value<T, E>(name: &str, value: T) -> Result<T, E>
// where
// T: Deserialize,
// E: Error;
// }
// pub enum MaybeUnit<'a, T> {
// Raw(T),
// Wrapped(&'a str, T)
// }
pub trait Deserializer<'de> {
type Error: Error;
//fn deserialize_real_u8(self) -> Result<u8, Self::Error>;
//fn deserialize_real_u16(self) -> Result<u16, Self::Error>;
//fn deserialize_real_u32(self) -> Result<u32, Self::Error>;
//fn deserialize_real_u64(self) -> Result<u64, Self::Error>;
//fn deserialize_real_i8(self) -> Result<i8, Self::Error>;
//fn deserialize_str(self) -> Result<&'de str, Self::Error>;
//fn deserialize_unit<U>(self) -> Result<U, Self::Error>
// where U: Unit;
//fn deserialize_maybe_unit(self) -> Result<MaybeUnit, Self::Error>;
}

24
sehn/src/grammar.rs Normal file
View File

@ -0,0 +1,24 @@
// Braces
pub const GRAMMAR_BRACE_OPEN: u8 = b'{';
pub const GRAMMAR_BRACE_CLOSE: u8 = b'}';
// Parentheses
pub const GRAMMAR_PAREN_OPEN: u8 = b'(';
pub const GRAMMAR_PAREN_CLOSE: u8 = b')';
// Brackets
pub const GRAMMAR_BRKET_OPEN: u8 = b'[';
pub const GRAMMAR_BRKET_CLOSE: u8 = b']';
// Whitespace
pub const GRAMMAR_LF: u8 = b'\n';
pub const GRAMMAR_CR: u8 = b'\r';
pub const GRAMMAR_TAB: u8 = b'\t';
pub const GRAMMAR_SPACE: u8 = b' ';
// Other
pub const GRAMMAR_BTICK: u8 = b'`';
pub const GRAMMAR_QUOTE: u8 = b'"';
pub const GRAMMAR_COMMA: u8 = b',';
pub const GRAMMAR_COLON: u8 = b':';
pub const GRAMMAR_BSLSH: u8 = b'\\';

15
sehn/src/lib.rs Normal file
View File

@ -0,0 +1,15 @@
pub mod se;
pub mod de;
pub mod value;
pub mod grammar;
pub mod utils;
// Serialization/Deserialization
pub use self::de::*;
pub use self::se::*;
// Supporting values
pub use self::value::*;
// Utilities
pub use self::utils::*;

192
sehn/src/se/base/mod.rs Normal file
View File

@ -0,0 +1,192 @@
mod real;
use core::marker::PhantomData;
use super::{
Write,
Serialize,
Serializer
};
use self::real::{
RealSerializer,
PrimitiveRealSerializer,
FastRealSerializer
};
use crate::grammar::*;
pub trait Config {
type RealSerializer: RealSerializer;
}
pub struct DefaultConfig {}
impl Config for DefaultConfig {
type RealSerializer = FastRealSerializer;
}
pub struct BaseSerializer<C: Config, W: Write> {
out: W,
cfg: PhantomData<C>
}
impl<C, W> BaseSerializer<C, W>
where
C: Config,
W: Write
{
pub fn new(out: W) -> Self {
Self { out, cfg: PhantomData }
}
pub fn into_inner(self) -> W {
self.out
}
#[inline]
pub fn serialize_real<R>(&mut self, r: R) -> Result<(), W::Error>
where
C::RealSerializer: PrimitiveRealSerializer<R>
{
C::RealSerializer::serialize_real::<W>(&mut self.out, r)
}
#[inline]
fn write_delim(&mut self, delim: u8) -> Result<(), W::Error> {
self.out.write(&[delim])
}
#[inline]
fn write_utf8(&mut self, s: &str) -> Result<(), W::Error> {
self.out.write(s.as_bytes())
}
}
impl<C: Config, W: Write> Serializer for &mut BaseSerializer<C, W> {
type Error = W::Error;
#[inline]
fn serialize_real_u8(self, real: u8) -> Result<(), Self::Error> {
self.serialize_real::<u8>(real)
}
#[inline]
fn serialize_real_u16(self, real: u16) -> Result<(), Self::Error> {
self.serialize_real::<u16>(real)
}
#[inline]
fn serialize_real_u32(self, real: u32) -> Result<(), Self::Error> {
self.serialize_real::<u32>(real)
}
#[inline]
fn serialize_real_u64(self, real: u64) -> Result<(), Self::Error> {
self.serialize_real::<u64>(real)
}
#[inline]
fn serialize_real_i8(self, real: i8) -> Result<(), Self::Error> {
self.serialize_real::<i8>(real)
}
#[inline]
fn serialize_real_i16(self, real: i16) -> Result<(), Self::Error> {
self.serialize_real::<i16>(real)
}
#[inline]
fn serialize_real_i32(self, real: i32) -> Result<(), Self::Error> {
self.serialize_real::<i32>(real)
}
#[inline]
fn serialize_real_i64(self, real: i64) -> Result<(), Self::Error> {
self.serialize_real::<i64>(real)
}
#[inline]
fn serialize_real_f32(self, real: f32) -> Result<(), Self::Error> {
self.serialize_real::<f32>(real)
}
#[inline]
fn serialize_real_f64(self, real: f64) -> Result<(), Self::Error> {
self.serialize_real::<f64>(real)
}
#[inline]
fn serialize_unit<'a, V>(self, name: &'a str, value: V) -> Result<(), Self::Error>
where V: Serialize
{
self.serialize_atom(name)?;
self.write_delim(GRAMMAR_PAREN_OPEN)?;
self.serialize_any(value)?;
self.write_delim(GRAMMAR_PAREN_CLOSE)
}
#[inline]
fn serialize_atom<'a>(self, atom: &'a str) -> Result<(), Self::Error> {
self.write_utf8(atom)
}
#[inline]
fn serialize_text<'a>(self, text: &'a str) -> Result<(), Self::Error> {
self.write_delim(GRAMMAR_QUOTE)?;
self.write_utf8(text)?;
self.write_delim(GRAMMAR_QUOTE)
}
#[inline]
fn serialize_multiline_text<'a, M>(self, lines: M) -> Result<(), Self::Error>
where M: IntoIterator<Item = &'a str>
{
self.write_delim(GRAMMAR_BTICK)?;
for line in lines.into_iter() {
self.serialize_text(line)?;
}
self.write_delim(GRAMMAR_BTICK)
}
#[inline]
fn serialize_dict<K, V, D>(self, dict: D) -> Result<(), Self::Error>
where
K: Serialize,
V: Serialize,
D: IntoIterator<Item = (K, V)>
{
self.write_delim(GRAMMAR_BRACE_OPEN)?;
let mut first = true;
for (k, v) in dict.into_iter() {
if first {
first = false;
} else {
self.write_delim(GRAMMAR_COMMA)?;
}
self.serialize_any(&k)?;
self.write_delim(GRAMMAR_COLON)?;
self.serialize_any(&v)?;
}
self.write_delim(GRAMMAR_BRACE_CLOSE)
}
#[inline]
fn serialize_list<T, L>(self, list: L) -> Result<(), Self::Error>
where
T: Serialize,
L: IntoIterator<Item = T>
{
self.write_delim(GRAMMAR_BRKET_OPEN)?;
let mut first = true;
for v in list.into_iter() {
if first {
first = false;
} else {
self.write_delim(GRAMMAR_COMMA)?;
}
self.serialize_any(&v)?;
}
self.write_delim(GRAMMAR_BRKET_CLOSE)
}
}

View File

@ -0,0 +1,54 @@
use core::fmt;
// Should be greater than MAX_SIG_DIGITS = 17
const DEC_BUF_SIZE: usize = 32;
pub struct DecBuf {
buf: [u8; DEC_BUF_SIZE],
len: usize
}
impl DecBuf {
pub fn new() -> Self {
Self {
buf: unsafe { core::mem::uninitialized() },
len: 0
}
}
pub fn as_bytes(&self) -> &[u8] {
&self.buf[0..self.len]
}
fn extend_mut_slice(&mut self, add_len: usize) -> Result<&mut [u8], fmt::Error> {
let cur_len = self.len;
let new_len = cur_len + add_len;
if new_len > self.buf.len() {
Err(fmt::Error)
} else {
self.len = new_len;
Ok(&mut self.buf[cur_len..new_len])
}
}
}
impl fmt::Write for DecBuf {
#[inline]
fn write_str(&mut self, s: &str) -> fmt::Result {
if s.is_empty() {
Ok(())
} else {
let new_bytes = s.as_bytes();
self.extend_mut_slice(new_bytes.len()).map(|b| {
b.copy_from_slice(new_bytes);
})
}
}
#[inline]
fn write_char(&mut self, c: char) -> fmt::Result {
self.extend_mut_slice(c.len_utf8()).map(|b| {
c.encode_utf8(b);
})
}
}

View File

@ -0,0 +1,47 @@
use core::fmt::Write as FmtWrite;
use crate::se::{Write, SerializerError};
use super::{
dec_buf::DecBuf,
RealSerializer,
PrimitiveRealSerializer
};
macro_rules! impl_write_real_int_for_fast_serializer {
($($prim:ident)*) => {
$(impl PrimitiveRealSerializer<$prim> for FastRealSerializer {
fn serialize_real<W: Write>(out: &mut W, r: $prim) -> Result<(), W::Error> {
let mut buf = itoa::Buffer::new();
let real_str = buf.format(r);
out.write(real_str.as_bytes())
}
})*
}
}
macro_rules! impl_write_real_float_for_fast_serializer {
($($prim:ident)*) => {
$(impl PrimitiveRealSerializer<$prim> for FastRealSerializer {
fn serialize_real<W: Write>(out: &mut W, r: $prim) -> Result<(), W::Error> {
let mut buf = DecBuf::new();
write!(buf, "{}", r).map_err(|_|
SerializerError::Unexpected("failed to format real")
)?;
out.write(buf.as_bytes())
}
})*
}
}
pub struct FastRealSerializer;
impl_write_real_int_for_fast_serializer!(
u8 u16 u32 u64
i8 i16 i32 i64
);
impl_write_real_float_for_fast_serializer!(
f32
f64
);
impl RealSerializer for FastRealSerializer {}

View File

@ -0,0 +1,24 @@
mod fast;
mod dec_buf;
pub use self::fast::*;
use crate::se::Write;
pub trait PrimitiveRealSerializer<T> {
fn serialize_real<W>(w: &mut W, i: T) -> Result<(), W::Error>
where
W: Write;
}
pub trait RealSerializer:
PrimitiveRealSerializer<i8> +
PrimitiveRealSerializer<i16> +
PrimitiveRealSerializer<i32> +
PrimitiveRealSerializer<i64> +
PrimitiveRealSerializer<u8> +
PrimitiveRealSerializer<u16> +
PrimitiveRealSerializer<u32> +
PrimitiveRealSerializer<u64> +
PrimitiveRealSerializer<f32> +
PrimitiveRealSerializer<f64> {}

20
sehn/src/se/error.rs Normal file
View File

@ -0,0 +1,20 @@
use core::fmt;
pub enum SerializerError {
/// A unexpected error (should never get this).
Unexpected(&'static str)
}
impl fmt::Debug for SerializerError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
use SerializerError::*;
match self {
Unexpected(ref err) => write!(f, "unexpected error: {}", err)
}
}
}
pub trait Error: fmt::Debug + From<SerializerError> {}
impl<T> Error for T where
T: fmt::Debug + From<SerializerError> {}

144
sehn/src/se/impls.rs Normal file
View File

@ -0,0 +1,144 @@
use std::hash::{Hash, BuildHasher};
use std::collections::HashMap;
use std::collections::BTreeMap;
use super::{Serialize, Serializer};
////////////////////////////////////////////////////////////////////////////////
macro_rules! real_impl {
($prim:ty, $func:ident) => {
impl Serialize for $prim {
fn serialize<S: Serializer>(&self, s: S) -> Result<(), S::Error> {
s.$func(*self)
}
}
}
}
real_impl!(u8, serialize_real_u8);
real_impl!(u16, serialize_real_u16);
real_impl!(u32, serialize_real_u32);
real_impl!(u64, serialize_real_u64);
real_impl!(i8, serialize_real_i8);
real_impl!(i16, serialize_real_i16);
real_impl!(i32, serialize_real_i32);
real_impl!(i64, serialize_real_i64);
real_impl!(f32, serialize_real_f32);
real_impl!(f64, serialize_real_f64);
////////////////////////////////////////////////////////////////////////////////
impl Serialize for str {
fn serialize<S: Serializer>(&self, s: S) -> Result<(), S::Error> {
s.serialize_text(self)
}
}
////////////////////////////////////////////////////////////////////////////////
//#[cfg(any(feature = "std", feature = "alloc"))]
macro_rules! seq_impl {
($ty:ident < T $(: $tbound1:ident $(+ $tbound2:ident)*)* $(, $typaram:ident : $bound:ident)* >) => {
impl<T $(, $typaram)*> Serialize for $ty<T $(, $typaram)*>
where
T: Serialize $(+ $tbound1 $(+ $tbound2)*)*,
$($typaram: $bound,)*
{
#[inline]
fn serialize<S>(&self, serializer: S) -> Result<(), S::Error>
where
S: Serializer,
{
serializer.serialize_list(self)
}
}
}
}
//#[cfg(any(feature = "std", feature = "alloc"))]
seq_impl!(Vec<T>);
// #[cfg(any(feature = "std", feature = "alloc"))]
// seq_impl!(BinaryHeap<T: Ord>);
// #[cfg(any(feature = "std", feature = "alloc"))]
// seq_impl!(BTreeSet<T: Ord>);
// #[cfg(feature = "std")]
// seq_impl!(HashSet<T: Eq + Hash, H: BuildHasher>);
// #[cfg(any(feature = "std", feature = "alloc"))]
// seq_impl!(LinkedList<T>);
// #[cfg(any(feature = "std", feature = "alloc"))]
// seq_impl!(VecDeque<T>);
////////////////////////////////////////////////////////////////////////////////
impl<T> Serialize for [T]
where
T: Serialize,
{
#[inline]
fn serialize<S>(&self, serializer: S) -> Result<(), S::Error>
where
S: Serializer,
{
serializer.serialize_list(self)
}
}
////////////////////////////////////////////////////////////////////////////////
//#[cfg(any(feature = "std", feature = "alloc"))]
macro_rules! dict_impl {
($ty:ident < K $(: $kbound1:ident $(+ $kbound2:ident)*)*, V $(, $typaram:ident : $bound:ident)* >) => {
impl<K, V $(, $typaram)*> Serialize for $ty<K, V $(, $typaram)*>
where
K: Serialize $(+ $kbound1 $(+ $kbound2)*)*,
V: Serialize,
$($typaram: $bound,)*
{
#[inline]
fn serialize<S>(&self, serializer: S) -> Result<(), S::Error>
where
S: Serializer,
{
serializer.serialize_dict(self)
}
}
}
}
//#[cfg(any(feature = "std", feature = "alloc"))]
dict_impl!(BTreeMap<K: Ord, V>);
//#[cfg(feature = "std")]
dict_impl!(HashMap<K: Eq + Hash, V, H: BuildHasher>);
////////////////////////////////////////////////////////////////////////////////
macro_rules! deref_impl {
(
$(#[doc = $doc:tt])*
<$($desc:tt)+
) => {
$(#[doc = $doc])*
impl <$($desc)+ {
#[inline]
fn serialize<S>(&self, serializer: S) -> Result<(), S::Error>
where
S: Serializer,
{
(**self).serialize(serializer)
}
}
};
}
deref_impl!(<'a, T: ?Sized> Serialize for &'a T where T: Serialize);
deref_impl!(<'a, T: ?Sized> Serialize for &'a mut T where T: Serialize);
//#[cfg(any(feature = "std", feature = "alloc"))]
deref_impl!(<T: ?Sized> Serialize for Box<T> where T: Serialize);

65
sehn/src/se/mod.rs Normal file
View File

@ -0,0 +1,65 @@
mod base;
mod error;
mod impls;
pub use self::base::*;
pub use self::error::*;
pub trait Write {
/// The core error that may bubble from
/// attempting to write.
type Error: Error;
/// The write function used with serialization.
fn write(&mut self, buf: &[u8]) -> Result<(), Self::Error>;
}
pub trait Serialize {
fn serialize<S>(&self, s: S) -> Result<(), S::Error>
where
S: Serializer;
}
pub trait Serializer: Sized {
type Error: Error;
fn serialize_any<T>(self, any: T) -> Result<(), Self::Error>
where
T: Serialize
{
any.serialize(self)
}
fn serialize_real_u8(self, real: u8) -> Result<(), Self::Error>;
fn serialize_real_u16(self, real: u16) -> Result<(), Self::Error>;
fn serialize_real_u32(self, real: u32) -> Result<(), Self::Error>;
fn serialize_real_u64(self, real: u64) -> Result<(), Self::Error>;
fn serialize_real_i8(self, real: i8) -> Result<(), Self::Error>;
fn serialize_real_i16(self, real: i16) -> Result<(), Self::Error>;
fn serialize_real_i32(self, real: i32) -> Result<(), Self::Error>;
fn serialize_real_i64(self, real: i64) -> Result<(), Self::Error>;
fn serialize_real_f32(self, real: f32) -> Result<(), Self::Error>;
fn serialize_real_f64(self, real: f64) -> Result<(), Self::Error>;
fn serialize_atom<'a>(self, atom: &'a str) -> Result<(), Self::Error>;
fn serialize_text<'a>(self, text: &'a str) -> Result<(), Self::Error>;
fn serialize_multiline_text<'a, M>(self, lines: M) -> Result<(), Self::Error>
where M: IntoIterator<Item = &'a str>;
fn serialize_unit<'a, V>(self, name: &'a str, value: V) -> Result<(), Self::Error>
where V: Serialize;
fn serialize_dict<K, V, D>(self, dict: D) -> Result<(), Self::Error>
where
K: Serialize,
V: Serialize,
D: IntoIterator<Item = (K, V)>;
fn serialize_list<V, L>(self, list: L) -> Result<(), Self::Error>
where
V: Serialize,
L: IntoIterator<Item = V>;
}

5
sehn/src/utils/mod.rs Normal file
View File

@ -0,0 +1,5 @@
mod to_string;
mod string_writer;
pub use self::to_string::*;
pub use self::string_writer::*;

View File

@ -0,0 +1,37 @@
use std::str::Utf8Error;
use std::string::FromUtf8Error;
use crate::se::{Write, SerializerError};
pub struct StringWriter {
buf: Vec<u8>
}
impl StringWriter {
pub fn new() -> Self {
Self {
buf: Vec::new()
}
}
pub fn clear(&mut self) {
self.buf.clear();
}
pub fn as_str(&self) -> Result<&str, Utf8Error> {
std::str::from_utf8(self.buf.as_ref())
}
pub fn to_string(self) -> Result<String, FromUtf8Error> {
String::from_utf8(self.buf)
}
}
impl Write for &mut StringWriter {
type Error = SerializerError;
fn write(&mut self, bytes: &[u8]) -> Result<(), Self::Error> {
self.buf.extend_from_slice(bytes);
Ok(())
}
}

View File

@ -0,0 +1,17 @@
use crate::se::*;
use super::StringWriter;
pub fn default_base_serializer<W>(w: W) -> BaseSerializer<DefaultConfig, W>
where W: Write
{
BaseSerializer::new(w)
}
pub fn to_string<T>(value: T) -> Result<String, SerializerError>
where T: Serialize
{
let mut w = StringWriter::new();
let mut s = default_base_serializer(&mut w);
value.serialize(&mut s)?;
Ok(w.to_string().unwrap())
}

3
sehn/src/value/mod.rs Normal file
View File

@ -0,0 +1,3 @@
mod tag;
pub use self::tag::*;

23
sehn/src/value/tag.rs Normal file
View File

@ -0,0 +1,23 @@
use crate::se::{Serializer, Serialize};
pub struct Tag<'a, T> {
name: &'a str,
value: T
}
impl<'a, T> Tag<'a, T> {
pub fn new(name: &'a str, value: T) -> Self {
Self { name, value }
}
}
impl<'a, T> Serialize for Tag<'a, T>
where
T: Serialize
{
fn serialize<S>(&self, s: S) -> Result<(), S::Error>
where S: Serializer
{
s.serialize_unit(self.name, &self.value)
}
}