aboutsummaryrefslogtreecommitdiff
path: root/vendor/lebe/src/lib.rs
diff options
context:
space:
mode:
authorValentin Popov <valentin@popov.link>2024-01-08 00:21:28 +0300
committerValentin Popov <valentin@popov.link>2024-01-08 00:21:28 +0300
commit1b6a04ca5504955c571d1c97504fb45ea0befee4 (patch)
tree7579f518b23313e8a9748a88ab6173d5e030b227 /vendor/lebe/src/lib.rs
parent5ecd8cf2cba827454317368b68571df0d13d7842 (diff)
downloadfparkan-1b6a04ca5504955c571d1c97504fb45ea0befee4.tar.xz
fparkan-1b6a04ca5504955c571d1c97504fb45ea0befee4.zip
Initial vendor packages
Signed-off-by: Valentin Popov <valentin@popov.link>
Diffstat (limited to 'vendor/lebe/src/lib.rs')
-rw-r--r--vendor/lebe/src/lib.rs578
1 files changed, 578 insertions, 0 deletions
diff --git a/vendor/lebe/src/lib.rs b/vendor/lebe/src/lib.rs
new file mode 100644
index 0000000..fbb4482
--- /dev/null
+++ b/vendor/lebe/src/lib.rs
@@ -0,0 +1,578 @@
+#![warn(
+ missing_docs, unused,
+ trivial_numeric_casts,
+ future_incompatible,
+ rust_2018_compatibility,
+ rust_2018_idioms,
+ clippy::all
+)]
+
+#![doc(html_root_url = "https://docs.rs/lebe/0.5.0")]
+
+//! Dead simple endianness conversions.
+//! The following operations are implemented on
+//! `u8`, `i8`, `u16`, `i16`, `u32`, `i32`, `u64`, `i64`, `u128`, `i128`, `f32`, `f64`:
+//!
+//!
+//! ### Read Numbers
+//! ```rust
+//! use lebe::prelude::*;
+//! let mut reader: &[u8] = &[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15];
+//!
+//! let number : u64 = reader.read_from_little_endian()?;
+//! let number = u64::read_from_big_endian(&mut reader)?;
+//! # Ok::<(), std::io::Error>(())
+//! ```
+//!
+//! ### Read Slices
+//! ```rust
+//! use std::io::Read;
+//! use lebe::prelude::*;
+//! let mut reader: &[u8] = &[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15];
+//!
+//! let mut numbers: &mut [u64] = &mut [0, 0];
+//! reader.read_from_little_endian_into(numbers)?;
+//! # Ok::<(), std::io::Error>(())
+//! ```
+//!
+//! ### Write Numbers
+//! ```rust
+//! use std::io::Read;
+//! use lebe::prelude::*;
+//! let mut writer: Vec<u8> = Vec::new();
+//!
+//! let number: u64 = 1237691;
+//! writer.write_as_big_endian(&number)?;
+//! # Ok::<(), std::io::Error>(())
+//! ```
+//!
+//! ### Write Slices
+//! ```rust
+//! use std::io::Write;
+//! use lebe::prelude::*;
+//! let mut writer: Vec<u8> = Vec::new();
+//!
+//! let numbers: &[u64] = &[1_u64, 234545_u64];
+//! writer.write_as_little_endian(numbers)?;
+//! # Ok::<(), std::io::Error>(())
+//! ```
+//!
+
+
+/// Exports some of the most common types.
+pub mod prelude {
+ pub use super::Endian;
+ pub use super::io::{ WriteEndian, ReadEndian, ReadPrimitive };
+}
+
+/// Represents values that can swap their bytes to reverse their endianness.
+///
+/// Supports converting values in-place using [`swap_bytes`] or [`convert_current_to_little_endian`]:
+/// Supports converting while transferring ownership using
+/// [`from_little_endian_into_current`] or [`from_current_into_little_endian`].
+///
+///
+/// For the types `u8`, `i8`, `&[u8]` and `&[i8]`, this trait will never transform any data,
+/// as they are just implemented for completeness.
+pub trait Endian {
+
+ /// Swaps all bytes in this value, inverting its endianness.
+ fn swap_bytes(&mut self);
+
+ /// On a little endian machine, this does nothing.
+ /// On a big endian machine, the bytes of this value are reversed.
+ #[inline] fn convert_current_to_little_endian(&mut self) {
+ #[cfg(target_endian = "big")] {
+ self.swap_bytes();
+ }
+ }
+
+ /// On a big endian machine, this does nothing.
+ /// On a little endian machine, the bytes of this value are reversed.
+ #[inline] fn convert_current_to_big_endian(&mut self) {
+ #[cfg(target_endian = "little")] {
+ self.swap_bytes();
+ }
+ }
+
+ /// On a little endian machine, this does nothing.
+ /// On a big endian machine, the bytes of this value are reversed.
+ #[inline] fn convert_little_endian_to_current(&mut self) {
+ #[cfg(target_endian = "big")] {
+ self.swap_bytes();
+ }
+ }
+
+ /// On a big endian machine, this does nothing.
+ /// On a little endian machine, the bytes of this value are reversed.
+ #[inline] fn convert_big_endian_to_current(&mut self) {
+ #[cfg(target_endian = "little")] {
+ self.swap_bytes();
+ }
+ }
+
+ /// On a little endian machine, this does nothing.
+ /// On a big endian machine, the bytes of this value are reversed.
+ #[inline] fn from_current_into_little_endian(mut self) -> Self where Self: Sized {
+ self.convert_current_to_little_endian();
+ self
+ }
+
+ /// On a big endian machine, this does nothing.
+ /// On a little endian machine, the bytes of this value are reversed.
+ #[inline] fn from_current_into_big_endian(mut self) -> Self where Self: Sized {
+ self.convert_current_to_big_endian();
+ self
+ }
+
+ /// On a little endian machine, this does nothing.
+ /// On a big endian machine, the bytes of this value are reversed.
+ #[inline] fn from_little_endian_into_current(mut self) -> Self where Self: Sized {
+ self.convert_little_endian_to_current();
+ self
+ }
+
+ /// On a big endian machine, this does nothing.
+ /// On a little endian machine, the bytes of this value are reversed.
+ #[inline] fn from_big_endian_into_current(mut self) -> Self where Self: Sized {
+ self.convert_big_endian_to_current();
+ self
+ }
+}
+
+
+// call a macro for each argument
+macro_rules! call_single_arg_macro_for_each {
+ ($macro: ident, $( $arguments: ident ),* ) => {
+ $( $macro! { $arguments } )*
+ };
+}
+
+// implement this interface for primitive signed and unsigned integers
+macro_rules! implement_simple_primitive_endian {
+ ($type: ident) => {
+ impl Endian for $type {
+ fn swap_bytes(&mut self) {
+ *self = $type::swap_bytes(*self);
+ }
+ }
+ };
+}
+
+
+call_single_arg_macro_for_each! {
+ implement_simple_primitive_endian,
+ u16, u32, u64, u128, i16, i32, i64, i128
+}
+
+// no-op implementations
+impl Endian for u8 { fn swap_bytes(&mut self) {} }
+impl Endian for i8 { fn swap_bytes(&mut self) {} }
+impl Endian for [u8] { fn swap_bytes(&mut self) {} }
+impl Endian for [i8] { fn swap_bytes(&mut self) {} }
+
+// implement this interface for primitive floats, because they do not have a `swap_bytes()` in `std`
+macro_rules! implement_float_primitive_by_bits {
+ ($type: ident) => {
+ impl Endian for $type {
+ fn swap_bytes(&mut self) {
+ *self = Self::from_bits(self.to_bits().swap_bytes());
+ }
+ }
+ };
+}
+
+
+implement_float_primitive_by_bits!(f32);
+implement_float_primitive_by_bits!(f64);
+
+macro_rules! implement_slice_by_element {
+ ($type: ident) => {
+ impl Endian for [$type] {
+ fn swap_bytes(&mut self) {
+ for number in self.iter_mut() { // TODO SIMD?
+ number.swap_bytes();
+ }
+ }
+ }
+ };
+}
+
+call_single_arg_macro_for_each! {
+ implement_slice_by_element,
+ u16, u32, u64, u128,
+ i16, i32, i64, i128,
+ f64, f32
+}
+
+/// Easily write primitives and slices of primitives to
+/// binary `std::io::Write` streams and easily read from binary `std::io::Read` streams.
+///
+/// Also contains the unsafe `bytes` module for reinterpreting values as byte slices and vice versa.
+pub mod io {
+ use super::Endian;
+ use std::io::{Read, Write, Result};
+
+ /// Reinterpret values as byte slices and byte slices as values unsafely.
+ pub mod bytes {
+ use std::io::{Read, Write, Result};
+
+ /// View this slice of values as a slice of bytes.
+ #[inline]
+ pub unsafe fn slice_as_bytes<T>(value: &[T]) -> &[u8] {
+ std::slice::from_raw_parts(
+ value.as_ptr() as *const u8,
+ value.len() * std::mem::size_of::<T>()
+ )
+ }
+
+ /// View this slice of values as a mutable slice of bytes.
+ #[inline]
+ pub unsafe fn slice_as_bytes_mut<T>(value: &mut [T]) -> &mut [u8] {
+ std::slice::from_raw_parts_mut(
+ value.as_mut_ptr() as *mut u8,
+ value.len() * std::mem::size_of::<T>()
+ )
+ }
+
+ /// View this reference as a slice of bytes.
+ #[inline]
+ pub unsafe fn value_as_bytes<T: Sized>(value: &T) -> &[u8] {
+ std::slice::from_raw_parts(
+ value as *const T as *const u8,
+ std::mem::size_of::<T>()
+ )
+ }
+
+ /// View this reference as a mutable slice of bytes.
+ #[inline]
+ pub unsafe fn value_as_bytes_mut<T: Sized>(value: &mut T) ->&mut [u8] {
+ std::slice::from_raw_parts_mut(
+ value as *mut T as *mut u8,
+ std::mem::size_of::<T>()
+ )
+ }
+
+ /// View this slice as a mutable slice of bytes and write it.
+ #[inline]
+ pub unsafe fn write_slice<T>(write: &mut impl Write, value: &[T]) -> Result<()> {
+ write.write_all(slice_as_bytes(value))
+ }
+
+ /// Read a slice of bytes into the specified slice.
+ #[inline]
+ pub unsafe fn read_slice<T>(read: &mut impl Read, value: &mut [T]) -> Result<()> {
+ read.read_exact(slice_as_bytes_mut(value))
+ }
+
+ /// View this reference as a mutable slice of bytes and write it.
+ #[inline]
+ pub unsafe fn write_value<T: Sized>(write: &mut impl Write, value: &T) -> Result<()> {
+ write.write_all(value_as_bytes(value))
+ }
+
+ /// Read a slice of bytes into the specified reference.
+ #[inline]
+ pub unsafe fn read_value<T: Sized>(read: &mut impl Read, value: &mut T) -> Result<()> {
+ read.read_exact(value_as_bytes_mut(value))
+ }
+ }
+
+ /// A `std::io::Write` output stream which supports writing any primitive values as bytes.
+ /// Will encode the values to be either little endian or big endian, as desired.
+ ///
+ /// This extension trait is implemented for all `Write` types.
+ /// Add `use lebe::io::WriteEndian;` to your code
+ /// to automatically unlock this functionality for all types that implement `Write`.
+ pub trait WriteEndian<T: ?Sized> {
+
+ /// Write the byte value of the specified reference, converting it to little endianness
+ fn write_as_little_endian(&mut self, value: &T) -> Result<()>;
+
+ /// Write the byte value of the specified reference, converting it to big endianness
+ fn write_as_big_endian(&mut self, value: &T) -> Result<()>;
+
+ /// Write the byte value of the specified reference, not converting it
+ fn write_as_native_endian(&mut self, value: &T) -> Result<()> {
+ #[cfg(target_endian = "little")] { self.write_as_little_endian(value) }
+ #[cfg(target_endian = "big")] { self.write_as_big_endian(value) }
+ }
+ }
+
+ /// A `std::io::Read` input stream which supports reading any primitive values from bytes.
+ /// Will decode the values from either little endian or big endian, as desired.
+ ///
+ /// This extension trait is implemented for all `Read` types.
+ /// Add `use lebe::io::ReadEndian;` to your code
+ /// to automatically unlock this functionality for all types that implement `Read`.
+ pub trait ReadEndian<T: ?Sized> {
+
+ /// Read into the supplied reference. Acts the same as `std::io::Read::read_exact`.
+ fn read_from_little_endian_into(&mut self, value: &mut T) -> Result<()>;
+
+ /// Read into the supplied reference. Acts the same as `std::io::Read::read_exact`.
+ fn read_from_big_endian_into(&mut self, value: &mut T) -> Result<()>;
+
+ /// Read into the supplied reference. Acts the same as `std::io::Read::read_exact`.
+ fn read_from_native_endian_into(&mut self, value: &mut T) -> Result<()> {
+ #[cfg(target_endian = "little")] { self.read_from_little_endian_into(value) }
+ #[cfg(target_endian = "big")] { self.read_from_big_endian_into(value) }
+ }
+
+ /// Read the byte value of the inferred type
+ #[inline]
+ fn read_from_little_endian(&mut self) -> Result<T> where T: Sized + Default {
+ let mut value = T::default();
+ self.read_from_little_endian_into(&mut value)?;
+ Ok(value)
+ }
+
+ /// Read the byte value of the inferred type
+ #[inline]
+ fn read_from_big_endian(&mut self) -> Result<T> where T: Sized + Default {
+ let mut value = T::default();
+ self.read_from_big_endian_into(&mut value)?;
+ Ok(value)
+ }
+
+ /// Read the byte value of the inferred type
+ #[inline]
+ fn read_from_native_endian(&mut self) -> Result<T> where T: Sized + Default {
+ #[cfg(target_endian = "little")] { self.read_from_little_endian() }
+ #[cfg(target_endian = "big")] { self.read_from_big_endian() }
+ }
+ }
+
+ // implement primitive for all types that are implemented by `Read`
+ impl<R: Read + ReadEndian<P>, P: Default> ReadPrimitive<R> for P {}
+
+
+ /// Offers a prettier versions of reading a primitive number.
+ ///
+ /// The default way of reading a value is:
+ /// ```rust
+ /// # use std::io::Read;
+ /// # use lebe::prelude::*;
+ /// # let mut reader : &[u8] = &[2, 1];
+ ///
+ /// let number: u16 = reader.read_from_little_endian()?;
+ /// println!("{}", number);
+ /// # Ok::<(), std::io::Error>(())
+ ///
+ /// ```
+ ///
+ /// This trait enables you to use expressions:
+ /// ```rust
+ /// # use std::io::Read;
+ /// # use lebe::prelude::*;
+ /// # let mut reader : &[u8] = &[2, 1];
+ ///
+ /// println!("{}", u16::read_from_little_endian(&mut reader)?);
+ /// # Ok::<(), std::io::Error>(())
+ /// ```
+ /// .
+ ///
+ pub trait ReadPrimitive<R: Read + ReadEndian<Self>> : Sized + Default {
+ /// Read this value from the supplied reader. Same as `ReadEndian::read_from_little_endian()`.
+ fn read_from_little_endian(read: &mut R) -> Result<Self> {
+ read.read_from_little_endian()
+ }
+
+ /// Read this value from the supplied reader. Same as `ReadEndian::read_from_big_endian()`.
+ fn read_from_big_endian(read: &mut R) -> Result<Self> {
+ read.read_from_big_endian()
+ }
+
+ /// Read this value from the supplied reader. Same as `ReadEndian::read_from_native_endian()`.
+ fn read_from_native_endian(read: &mut R) -> Result<Self> {
+ read.read_from_native_endian()
+ }
+ }
+
+ macro_rules! implement_simple_primitive_write {
+ ($type: ident) => {
+ impl<W: Write> WriteEndian<$type> for W {
+ fn write_as_little_endian(&mut self, value: &$type) -> Result<()> {
+ unsafe { bytes::write_value(self, &value.from_current_into_little_endian()) }
+ }
+
+ fn write_as_big_endian(&mut self, value: &$type) -> Result<()> {
+ unsafe { bytes::write_value(self, &value.from_current_into_big_endian()) }
+ }
+ }
+
+ impl<R: Read> ReadEndian<$type> for R {
+ #[inline]
+ fn read_from_little_endian_into(&mut self, value: &mut $type) -> Result<()> {
+ unsafe { bytes::read_value(self, value)?; }
+ value.convert_little_endian_to_current();
+ Ok(())
+ }
+
+ #[inline]
+ fn read_from_big_endian_into(&mut self, value: &mut $type) -> Result<()> {
+ unsafe { bytes::read_value(self, value)?; }
+ value.convert_big_endian_to_current();
+ Ok(())
+ }
+ }
+ };
+ }
+
+ call_single_arg_macro_for_each! {
+ implement_simple_primitive_write,
+ u8, u16, u32, u64, u128,
+ i8, i16, i32, i64, i128,
+ f32, f64
+ }
+
+
+ macro_rules! implement_slice_io {
+ ($type: ident) => {
+ impl<W: Write> WriteEndian<[$type]> for W {
+ fn write_as_little_endian(&mut self, value: &[$type]) -> Result<()> {
+ #[cfg(target_endian = "big")] {
+ for number in value { // TODO SIMD!
+ self.write_as_little_endian(number)?;
+ }
+ }
+
+ // else write whole slice
+ #[cfg(target_endian = "little")]
+ unsafe { bytes::write_slice(self, value)?; }
+
+ Ok(())
+ }
+
+ fn write_as_big_endian(&mut self, value: &[$type]) -> Result<()> {
+ #[cfg(target_endian = "little")] {
+ for number in value { // TODO SIMD!
+ self.write_as_big_endian(number)?;
+ }
+ }
+
+ // else write whole slice
+ #[cfg(target_endian = "big")]
+ unsafe { bytes::write_slice(self, value)?; }
+
+ Ok(())
+ }
+ }
+
+ impl<R: Read> ReadEndian<[$type]> for R {
+ fn read_from_little_endian_into(&mut self, value: &mut [$type]) -> Result<()> {
+ unsafe { bytes::read_slice(self, value)? };
+ value.convert_little_endian_to_current();
+ Ok(())
+ }
+
+ fn read_from_big_endian_into(&mut self, value: &mut [$type]) -> Result<()> {
+ unsafe { bytes::read_slice(self, value)? };
+ value.convert_big_endian_to_current();
+ Ok(())
+ }
+ }
+ };
+ }
+
+ call_single_arg_macro_for_each! {
+ implement_slice_io,
+ u8, u16, u32, u64, u128,
+ i8, i16, i32, i64, i128,
+ f64, f32
+ }
+
+
+
+ // TODO: SIMD
+ /*impl<R: Read> ReadEndian<[f32]> for R {
+ fn read_from_little_endian_into(&mut self, value: &mut [f32]) -> Result<()> {
+ unsafe { bytes::read_slice(self, value)? };
+ value.convert_little_endian_to_current();
+ Ok(())
+ }
+
+ fn read_from_big_endian_into(&mut self, value: &mut [f32]) -> Result<()> {
+ unsafe { bytes::read_slice(self, value)? };
+ value.convert_big_endian_to_current();
+ Ok(())
+ }
+ }
+
+ impl<W: Write> WriteEndian<[f32]> for W {
+ fn write_as_big_endian(&mut self, value: &[f32]) -> Result<()> {
+ if cfg!(target_endian = "little") {
+
+ // FIX ME this SIMD optimization makes no difference ... why? like, ZERO difference, not even worse
+// #[cfg(feature = "simd")]
+ #[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+ unsafe {
+ if is_x86_feature_detected!("avx2") {
+ write_bytes_avx(self, value);
+ return Ok(());
+ }
+ }
+
+ // otherwise (no avx2 available)
+// for number in value {
+// self.write_as_little_endian(number);
+// }
+//
+// return Ok(());
+ unimplemented!();
+
+ #[target_feature(enable = "avx2")]
+ #[cfg(any(target_arch = "x86", target_arch = "x86_64"))]
+ unsafe fn write_bytes_avx(write: &mut impl Write, slice: &[f32]) -> Result<()> {
+ #[cfg(target_arch = "x86")] use std::arch::x86 as mm;
+ #[cfg(target_arch = "x86_64")] use std::arch::x86_64 as mm;
+
+ let bytes: &[u8] = crate::io::bytes::slice_as_bytes(slice);
+ let mut chunks = bytes.chunks_exact(32);
+
+ let indices = mm::_mm256_set_epi8(
+ 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,
+ 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
+// 3,2,1,0, 7,6,5,4, 11,10,9,8, 15,14,13,12,
+// 3,2,1,0, 7,6,5,4, 11,10,9,8, 15,14,13,12
+ );
+
+ for chunk in &mut chunks {
+ let data = mm::_mm256_loadu_si256(chunk.as_ptr() as _);
+ let result = mm::_mm256_shuffle_epi8(data, indices);
+ let mut out = [0_u8; 32];
+ mm::_mm256_storeu_si256(out.as_mut_ptr() as _, result);
+ write.write_all(&out)?;
+ }
+
+ let remainder = chunks.remainder();
+
+ { // copy remainder into larger slice, with zeroes at the end
+ let mut last_chunk = [0_u8; 32];
+ last_chunk[0..remainder.len()].copy_from_slice(remainder);
+ let data = mm::_mm256_loadu_si256(last_chunk.as_ptr() as _);
+ let result = mm::_mm256_shuffle_epi8(data, indices);
+ mm::_mm256_storeu_si256(last_chunk.as_mut_ptr() as _, result);
+ write.write_all(&last_chunk[0..remainder.len()])?;
+ }
+
+ Ok(())
+ }
+ }
+
+ else {
+ unsafe { bytes::write_slice(self, value)?; }
+ Ok(())
+ }
+ }
+
+ fn write_as_little_endian(&mut self, value: &[f32]) -> Result<()> {
+ for number in value {
+ self.write_as_little_endian(number)?;
+ }
+
+ Ok(())
+ }
+ }*/
+}
+