| 1 |
// Copyright Mozilla Foundation. See the COPYRIGHT |
| 2 |
// file at the top-level directory of this distribution. |
| 3 |
// |
| 4 |
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 5 |
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 6 |
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your |
| 7 |
// option. This file may not be copied, modified, or distributed |
| 8 |
// except according to those terms. |
| 9 |
|
| 10 |
//! This module provides structs that use lifetimes to couple bounds checking |
| 11 |
//! and space availability checking and detaching those from actual slice |
| 12 |
//! reading/writing. |
| 13 |
//! |
| 14 |
//! At present, the internals of the implementation are safe code, so the |
| 15 |
//! bound checks currently also happen on read/write. Once this code works, |
| 16 |
//! the plan is to replace the internals with unsafe code that omits the |
| 17 |
//! bound check at the read/write time. |
| 18 |
|
| 19 |
#[cfg(all( |
| 20 |
feature = "simd-accel", |
| 21 |
any( |
| 22 |
target_feature = "sse2", |
| 23 |
all(target_endian = "little", target_arch = "aarch64"), |
| 24 |
all(target_endian = "little", target_feature = "neon") |
| 25 |
) |
| 26 |
))] |
| 27 |
use crate::simd_funcs::*; |
| 28 |
|
| 29 |
#[cfg(all( |
| 30 |
feature = "simd-accel", |
| 31 |
any( |
| 32 |
target_feature = "sse2", |
| 33 |
all(target_endian = "little", target_arch = "aarch64"), |
| 34 |
all(target_endian = "little", target_feature = "neon") |
| 35 |
) |
| 36 |
))] |
| 37 |
use core::simd::u16x8; |
| 38 |
|
| 39 |
use super::DecoderResult; |
| 40 |
use super::EncoderResult; |
| 41 |
use crate::ascii::*; |
| 42 |
use crate::utf_8::convert_utf8_to_utf16_up_to_invalid; |
| 43 |
use crate::utf_8::utf8_valid_up_to; |
| 44 |
|
| 45 |
pub enum Space<T> { |
| 46 |
Available(T), |
| 47 |
Full(usize), |
| 48 |
} |
| 49 |
|
| 50 |
pub enum CopyAsciiResult<T, U> { |
| 51 |
Stop(T), |
| 52 |
GoOn(U), |
| 53 |
} |
| 54 |
|
| 55 |
pub enum NonAscii { |
| 56 |
BmpExclAscii(u16), |
| 57 |
Astral(char), |
| 58 |
} |
| 59 |
|
| 60 |
pub enum Unicode { |
| 61 |
Ascii(u8), |
| 62 |
NonAscii(NonAscii), |
| 63 |
} |
| 64 |
|
| 65 |
// Start UTF-16LE/BE fast path |
| 66 |
|
| 67 |
pub trait Endian { |
| 68 |
const OPPOSITE_ENDIAN: bool; |
| 69 |
} |
| 70 |
|
| 71 |
pub struct BigEndian; |
| 72 |
|
| 73 |
impl Endian for BigEndian { |
| 74 |
#[cfg(target_endian = "little")] |
| 75 |
const OPPOSITE_ENDIAN: bool = true; |
| 76 |
|
| 77 |
#[cfg(target_endian = "big")] |
| 78 |
const OPPOSITE_ENDIAN: bool = false; |
| 79 |
} |
| 80 |
|
| 81 |
pub struct LittleEndian; |
| 82 |
|
| 83 |
impl Endian for LittleEndian { |
| 84 |
#[cfg(target_endian = "little")] |
| 85 |
const OPPOSITE_ENDIAN: bool = false; |
| 86 |
|
| 87 |
#[cfg(target_endian = "big")] |
| 88 |
const OPPOSITE_ENDIAN: bool = true; |
| 89 |
} |
| 90 |
|
| 91 |
#[derive(Debug, Copy, Clone)] |
| 92 |
struct UnalignedU16Slice { |
| 93 |
// Safety invariant: ptr must be valid for reading 2*len bytes |
| 94 |
ptr: *const u8, |
| 95 |
len: usize, |
| 96 |
} |
| 97 |
|
| 98 |
impl UnalignedU16Slice { |
| 99 |
/// Safety: ptr must be valid for reading 2*len bytes |
| 100 |
#[inline(always)] |
| 101 |
pub unsafe fn new(ptr: *const u8, len: usize) -> UnalignedU16Slice { |
| 102 |
// Safety: field invariant passed up to caller here |
| 103 |
UnalignedU16Slice { ptr, len } |
| 104 |
} |
| 105 |
|
| 106 |
#[inline(always)] |
| 107 |
pub fn trim_last(&mut self) { |
| 108 |
assert!(self.len > 0); |
| 109 |
// Safety: invariant upheld here: a slice is still valid with a shorter len |
| 110 |
self.len -= 1; |
| 111 |
} |
| 112 |
|
| 113 |
#[inline(always)] |
| 114 |
pub fn at(&self, i: usize) -> u16 { |
| 115 |
use core::mem::MaybeUninit; |
| 116 |
|
| 117 |
assert!(i < self.len); |
| 118 |
unsafe { |
| 119 |
let mut u: MaybeUninit<u16> = MaybeUninit::uninit(); |
| 120 |
// Safety: i is at most len - 1, which works here |
| 121 |
::core::ptr::copy_nonoverlapping(self.ptr.add(i * 2), u.as_mut_ptr() as *mut u8, 2); |
| 122 |
// Safety: valid read above lets us do this |
| 123 |
u.assume_init() |
| 124 |
} |
| 125 |
} |
| 126 |
|
| 127 |
#[cfg(feature = "simd-accel")] |
| 128 |
#[inline(always)] |
| 129 |
pub fn simd_at(&self, i: usize) -> u16x8 { |
| 130 |
// Safety: i/len are on the scale of u16s, each one corresponds to 2 u8s |
| 131 |
assert!(i + SIMD_STRIDE_SIZE / 2 <= self.len); |
| 132 |
let byte_index = i * 2; |
| 133 |
// Safety: load16_unaligned needs SIMD_STRIDE_SIZE=16 u8 elements to read, |
| 134 |
// or 16/2 = 8 u16 elements to read. |
| 135 |
// We have checked that we have at least that many above. |
| 136 |
|
| 137 |
unsafe { to_u16_lanes(load16_unaligned(self.ptr.add(byte_index))) } |
| 138 |
} |
| 139 |
|
| 140 |
#[inline(always)] |
| 141 |
pub fn len(&self) -> usize { |
| 142 |
self.len |
| 143 |
} |
| 144 |
|
| 145 |
#[inline(always)] |
| 146 |
pub fn tail(&self, from: usize) -> UnalignedU16Slice { |
| 147 |
// XXX the return value should be restricted not to |
| 148 |
// outlive self. |
| 149 |
assert!(from <= self.len); |
| 150 |
// Safety: This upholds the same invariant: `from` is in bounds and we're returning a shorter slice |
| 151 |
unsafe { UnalignedU16Slice::new(self.ptr.add(from * 2), self.len - from) } |
| 152 |
} |
| 153 |
|
| 154 |
#[cfg(feature = "simd-accel")] |
| 155 |
#[inline(always)] |
| 156 |
pub fn copy_bmp_to<E: Endian>(&self, other: &mut [u16]) -> Option<(u16, usize)> { |
| 157 |
assert!(self.len <= other.len()); |
| 158 |
let mut offset = 0; |
| 159 |
// Safety: SIMD_STRIDE_SIZE is measured in bytes, whereas len is in u16s. We check we can |
| 160 |
// munch SIMD_STRIDE_SIZE / 2 u16s which means we can write SIMD_STRIDE_SIZE u8s |
| 161 |
if SIMD_STRIDE_SIZE / 2 <= self.len { |
| 162 |
let len_minus_stride = self.len - SIMD_STRIDE_SIZE / 2; |
| 163 |
loop { |
| 164 |
let mut simd = self.simd_at(offset); |
| 165 |
if E::OPPOSITE_ENDIAN { |
| 166 |
simd = simd_byte_swap(simd); |
| 167 |
} |
| 168 |
// Safety: we have enough space on the other side to write this |
| 169 |
unsafe { |
| 170 |
store8_unaligned(other.as_mut_ptr().add(offset), simd); |
| 171 |
} |
| 172 |
if contains_surrogates(simd) { |
| 173 |
break; |
| 174 |
} |
| 175 |
offset += SIMD_STRIDE_SIZE / 2; |
| 176 |
// Safety: This ensures we still have space for writing SIMD_STRIDE_SIZE u8s |
| 177 |
if offset > len_minus_stride { |
| 178 |
break; |
| 179 |
} |
| 180 |
} |
| 181 |
} |
| 182 |
while offset < self.len { |
| 183 |
let unit = swap_if_opposite_endian::<E>(self.at(offset)); |
| 184 |
other[offset] = unit; |
| 185 |
if super::in_range16(unit, 0xD800, 0xE000) { |
| 186 |
return Some((unit, offset)); |
| 187 |
} |
| 188 |
offset += 1; |
| 189 |
} |
| 190 |
None |
| 191 |
} |
| 192 |
|
| 193 |
#[cfg(not(feature = "simd-accel"))] |
| 194 |
#[inline(always)] |
| 195 |
fn copy_bmp_to<E: Endian>(&self, other: &mut [u16]) -> Option<(u16, usize)> { |
| 196 |
assert!(self.len <= other.len()); |
| 197 |
for (i, target) in other.iter_mut().enumerate().take(self.len) { |
| 198 |
let unit = swap_if_opposite_endian::<E>(self.at(i)); |
| 199 |
*target = unit; |
| 200 |
if super::in_range16(unit, 0xD800, 0xE000) { |
| 201 |
return Some((unit, i)); |
| 202 |
} |
| 203 |
} |
| 204 |
None |
| 205 |
} |
| 206 |
} |
| 207 |
|
| 208 |
#[inline(always)] |
| 209 |
fn copy_unaligned_basic_latin_to_ascii_alu<E: Endian>( |
| 210 |
src: UnalignedU16Slice, |
| 211 |
dst: &mut [u8], |
| 212 |
offset: usize, |
| 213 |
) -> CopyAsciiResult<usize, (u16, usize)> { |
| 214 |
let len = ::core::cmp::min(src.len(), dst.len()); |
| 215 |
let mut i = 0usize; |
| 216 |
loop { |
| 217 |
if i == len { |
| 218 |
return CopyAsciiResult::Stop(i + offset); |
| 219 |
} |
| 220 |
let unit = swap_if_opposite_endian::<E>(src.at(i)); |
| 221 |
if unit > 0x7F { |
| 222 |
return CopyAsciiResult::GoOn((unit, i + offset)); |
| 223 |
} |
| 224 |
dst[i] = unit as u8; |
| 225 |
i += 1; |
| 226 |
} |
| 227 |
} |
| 228 |
|
| 229 |
#[inline(always)] |
| 230 |
fn swap_if_opposite_endian<E: Endian>(unit: u16) -> u16 { |
| 231 |
if E::OPPOSITE_ENDIAN { |
| 232 |
unit.swap_bytes() |
| 233 |
} else { |
| 234 |
unit |
| 235 |
} |
| 236 |
} |
| 237 |
|
| 238 |
#[cfg(not(feature = "simd-accel"))] |
| 239 |
#[inline(always)] |
| 240 |
fn copy_unaligned_basic_latin_to_ascii<E: Endian>( |
| 241 |
src: UnalignedU16Slice, |
| 242 |
dst: &mut [u8], |
| 243 |
) -> CopyAsciiResult<usize, (u16, usize)> { |
| 244 |
copy_unaligned_basic_latin_to_ascii_alu::<E>(src, dst, 0) |
| 245 |
} |
| 246 |
|
| 247 |
#[cfg(feature = "simd-accel")] |
| 248 |
#[inline(always)] |
| 249 |
fn copy_unaligned_basic_latin_to_ascii<E: Endian>( |
| 250 |
src: UnalignedU16Slice, |
| 251 |
dst: &mut [u8], |
| 252 |
) -> CopyAsciiResult<usize, (u16, usize)> { |
| 253 |
let len = ::core::cmp::min(src.len(), dst.len()); |
| 254 |
let mut offset = 0; |
| 255 |
// Safety: This check ensures we are able to read/write at least SIMD_STRIDE_SIZE elements |
| 256 |
if SIMD_STRIDE_SIZE <= len { |
| 257 |
let len_minus_stride = len - SIMD_STRIDE_SIZE; |
| 258 |
loop { |
| 259 |
let mut first = src.simd_at(offset); |
| 260 |
let mut second = src.simd_at(offset + (SIMD_STRIDE_SIZE / 2)); |
| 261 |
if E::OPPOSITE_ENDIAN { |
| 262 |
first = simd_byte_swap(first); |
| 263 |
second = simd_byte_swap(second); |
| 264 |
} |
| 265 |
if !simd_is_basic_latin(first | second) { |
| 266 |
break; |
| 267 |
} |
| 268 |
let packed = simd_pack(first, second); |
| 269 |
// Safety: We are able to write SIMD_STRIDE_SIZE elements in this iteration |
| 270 |
unsafe { |
| 271 |
store16_unaligned(dst.as_mut_ptr().add(offset), packed); |
| 272 |
} |
| 273 |
offset += SIMD_STRIDE_SIZE; |
| 274 |
// Safety: This is `offset > len - SIMD_STRIDE_SIZE`, which ensures that we can write at least SIMD_STRIDE_SIZE elements |
| 275 |
// in the next iteration |
| 276 |
if offset > len_minus_stride { |
| 277 |
break; |
| 278 |
} |
| 279 |
} |
| 280 |
} |
| 281 |
copy_unaligned_basic_latin_to_ascii_alu::<E>(src.tail(offset), &mut dst[offset..], offset) |
| 282 |
} |
| 283 |
|
| 284 |
#[inline(always)] |
| 285 |
fn convert_unaligned_utf16_to_utf8<E: Endian>( |
| 286 |
src: UnalignedU16Slice, |
| 287 |
dst: &mut [u8], |
| 288 |
) -> (usize, usize, bool) { |
| 289 |
if dst.len() < 4 { |
| 290 |
return (0, 0, false); |
| 291 |
} |
| 292 |
let mut src_pos = 0usize; |
| 293 |
let mut dst_pos = 0usize; |
| 294 |
let src_len = src.len(); |
| 295 |
let dst_len_minus_three = dst.len() - 3; |
| 296 |
'outer: loop { |
| 297 |
let mut non_ascii = match copy_unaligned_basic_latin_to_ascii::<E>( |
| 298 |
src.tail(src_pos), |
| 299 |
&mut dst[dst_pos..], |
| 300 |
) { |
| 301 |
CopyAsciiResult::GoOn((unit, read_written)) => { |
| 302 |
src_pos += read_written; |
| 303 |
dst_pos += read_written; |
| 304 |
unit |
| 305 |
} |
| 306 |
CopyAsciiResult::Stop(read_written) => { |
| 307 |
return (src_pos + read_written, dst_pos + read_written, false); |
| 308 |
} |
| 309 |
}; |
| 310 |
if dst_pos >= dst_len_minus_three { |
| 311 |
break 'outer; |
| 312 |
} |
| 313 |
// We have enough destination space to commit to |
| 314 |
// having read `non_ascii`. |
| 315 |
src_pos += 1; |
| 316 |
'inner: loop { |
| 317 |
let non_ascii_minus_surrogate_start = non_ascii.wrapping_sub(0xD800); |
| 318 |
if non_ascii_minus_surrogate_start > (0xDFFF - 0xD800) { |
| 319 |
if non_ascii < 0x800 { |
| 320 |
dst[dst_pos] = ((non_ascii >> 6) | 0xC0) as u8; |
| 321 |
dst_pos += 1; |
| 322 |
dst[dst_pos] = ((non_ascii & 0x3F) | 0x80) as u8; |
| 323 |
dst_pos += 1; |
| 324 |
} else { |
| 325 |
dst[dst_pos] = ((non_ascii >> 12) | 0xE0) as u8; |
| 326 |
dst_pos += 1; |
| 327 |
dst[dst_pos] = (((non_ascii & 0xFC0) >> 6) | 0x80) as u8; |
| 328 |
dst_pos += 1; |
| 329 |
dst[dst_pos] = ((non_ascii & 0x3F) | 0x80) as u8; |
| 330 |
dst_pos += 1; |
| 331 |
} |
| 332 |
} else if non_ascii_minus_surrogate_start <= (0xDBFF - 0xD800) { |
| 333 |
// high surrogate |
| 334 |
if src_pos < src_len { |
| 335 |
let second = swap_if_opposite_endian::<E>(src.at(src_pos)); |
| 336 |
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00); |
| 337 |
if second_minus_low_surrogate_start <= (0xDFFF - 0xDC00) { |
| 338 |
// The next code unit is a low surrogate. Advance position. |
| 339 |
src_pos += 1; |
| 340 |
let point = (u32::from(non_ascii) << 10) + u32::from(second) |
| 341 |
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32); |
| 342 |
|
| 343 |
dst[dst_pos] = ((point >> 18) | 0xF0u32) as u8; |
| 344 |
dst_pos += 1; |
| 345 |
dst[dst_pos] = (((point & 0x3F000u32) >> 12) | 0x80u32) as u8; |
| 346 |
dst_pos += 1; |
| 347 |
dst[dst_pos] = (((point & 0xFC0u32) >> 6) | 0x80u32) as u8; |
| 348 |
dst_pos += 1; |
| 349 |
dst[dst_pos] = ((point & 0x3Fu32) | 0x80u32) as u8; |
| 350 |
dst_pos += 1; |
| 351 |
} else { |
| 352 |
// The next code unit is not a low surrogate. Don't advance |
| 353 |
// position and treat the high surrogate as unpaired. |
| 354 |
return (src_pos, dst_pos, true); |
| 355 |
} |
| 356 |
} else { |
| 357 |
// Unpaired surrogate at the end of buffer |
| 358 |
return (src_pos, dst_pos, true); |
| 359 |
} |
| 360 |
} else { |
| 361 |
// Unpaired low surrogate |
| 362 |
return (src_pos, dst_pos, true); |
| 363 |
} |
| 364 |
if dst_pos >= dst_len_minus_three || src_pos == src_len { |
| 365 |
break 'outer; |
| 366 |
} |
| 367 |
let unit = swap_if_opposite_endian::<E>(src.at(src_pos)); |
| 368 |
src_pos += 1; |
| 369 |
if unit > 0x7F { |
| 370 |
non_ascii = unit; |
| 371 |
continue 'inner; |
| 372 |
} |
| 373 |
dst[dst_pos] = unit as u8; |
| 374 |
dst_pos += 1; |
| 375 |
continue 'outer; |
| 376 |
} |
| 377 |
} |
| 378 |
(src_pos, dst_pos, false) |
| 379 |
} |
| 380 |
|
| 381 |
// Byte source |
| 382 |
|
| 383 |
pub struct ByteSource<'a> { |
| 384 |
slice: &'a [u8], |
| 385 |
pos: usize, |
| 386 |
} |
| 387 |
|
| 388 |
impl<'a> ByteSource<'a> { |
| 389 |
#[inline(always)] |
| 390 |
pub fn new(src: &[u8]) -> ByteSource { |
| 391 |
ByteSource { slice: src, pos: 0 } |
| 392 |
} |
| 393 |
#[inline(always)] |
| 394 |
pub fn check_available<'b>(&'b mut self) -> Space<ByteReadHandle<'b, 'a>> { |
| 395 |
if self.pos < self.slice.len() { |
| 396 |
Space::Available(ByteReadHandle::new(self)) |
| 397 |
} else { |
| 398 |
Space::Full(self.consumed()) |
| 399 |
} |
| 400 |
} |
| 401 |
#[inline(always)] |
| 402 |
fn read(&mut self) -> u8 { |
| 403 |
let ret = self.slice[self.pos]; |
| 404 |
self.pos += 1; |
| 405 |
ret |
| 406 |
} |
| 407 |
#[inline(always)] |
| 408 |
fn unread(&mut self) -> usize { |
| 409 |
self.pos -= 1; |
| 410 |
self.pos |
| 411 |
} |
| 412 |
#[inline(always)] |
| 413 |
pub fn consumed(&self) -> usize { |
| 414 |
self.pos |
| 415 |
} |
| 416 |
} |
| 417 |
|
| 418 |
pub struct ByteReadHandle<'a, 'b> |
| 419 |
where |
| 420 |
'b: 'a, |
| 421 |
{ |
| 422 |
source: &'a mut ByteSource<'b>, |
| 423 |
} |
| 424 |
|
| 425 |
impl<'a, 'b> ByteReadHandle<'a, 'b> |
| 426 |
where |
| 427 |
'b: 'a, |
| 428 |
{ |
| 429 |
#[inline(always)] |
| 430 |
fn new(src: &'a mut ByteSource<'b>) -> ByteReadHandle<'a, 'b> { |
| 431 |
ByteReadHandle { source: src } |
| 432 |
} |
| 433 |
#[inline(always)] |
| 434 |
pub fn read(self) -> (u8, ByteUnreadHandle<'a, 'b>) { |
| 435 |
let byte = self.source.read(); |
| 436 |
let handle = ByteUnreadHandle::new(self.source); |
| 437 |
(byte, handle) |
| 438 |
} |
| 439 |
#[inline(always)] |
| 440 |
pub fn consumed(&self) -> usize { |
| 441 |
self.source.consumed() |
| 442 |
} |
| 443 |
} |
| 444 |
|
| 445 |
pub struct ByteUnreadHandle<'a, 'b> |
| 446 |
where |
| 447 |
'b: 'a, |
| 448 |
{ |
| 449 |
source: &'a mut ByteSource<'b>, |
| 450 |
} |
| 451 |
|
| 452 |
impl<'a, 'b> ByteUnreadHandle<'a, 'b> |
| 453 |
where |
| 454 |
'b: 'a, |
| 455 |
{ |
| 456 |
#[inline(always)] |
| 457 |
fn new(src: &'a mut ByteSource<'b>) -> ByteUnreadHandle<'a, 'b> { |
| 458 |
ByteUnreadHandle { source: src } |
| 459 |
} |
| 460 |
#[inline(always)] |
| 461 |
pub fn unread(self) -> usize { |
| 462 |
self.source.unread() |
| 463 |
} |
| 464 |
#[inline(always)] |
| 465 |
pub fn consumed(&self) -> usize { |
| 466 |
self.source.consumed() |
| 467 |
} |
| 468 |
#[inline(always)] |
| 469 |
pub fn commit(self) -> &'a mut ByteSource<'b> { |
| 470 |
self.source |
| 471 |
} |
| 472 |
} |
| 473 |
|
| 474 |
// UTF-16 destination |
| 475 |
|
| 476 |
pub struct Utf16BmpHandle<'a, 'b> |
| 477 |
where |
| 478 |
'b: 'a, |
| 479 |
{ |
| 480 |
dest: &'a mut Utf16Destination<'b>, |
| 481 |
} |
| 482 |
|
| 483 |
impl<'a, 'b> Utf16BmpHandle<'a, 'b> |
| 484 |
where |
| 485 |
'b: 'a, |
| 486 |
{ |
| 487 |
#[inline(always)] |
| 488 |
fn new(dst: &'a mut Utf16Destination<'b>) -> Utf16BmpHandle<'a, 'b> { |
| 489 |
Utf16BmpHandle { dest: dst } |
| 490 |
} |
| 491 |
#[inline(always)] |
| 492 |
pub fn written(&self) -> usize { |
| 493 |
self.dest.written() |
| 494 |
} |
| 495 |
#[inline(always)] |
| 496 |
pub fn write_ascii(self, ascii: u8) -> &'a mut Utf16Destination<'b> { |
| 497 |
self.dest.write_ascii(ascii); |
| 498 |
self.dest |
| 499 |
} |
| 500 |
#[inline(always)] |
| 501 |
pub fn write_bmp(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 502 |
self.dest.write_bmp(bmp); |
| 503 |
self.dest |
| 504 |
} |
| 505 |
#[inline(always)] |
| 506 |
pub fn write_bmp_excl_ascii(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 507 |
self.dest.write_bmp_excl_ascii(bmp); |
| 508 |
self.dest |
| 509 |
} |
| 510 |
#[inline(always)] |
| 511 |
pub fn write_mid_bmp(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 512 |
self.dest.write_mid_bmp(bmp); |
| 513 |
self.dest |
| 514 |
} |
| 515 |
#[inline(always)] |
| 516 |
pub fn write_upper_bmp(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 517 |
self.dest.write_upper_bmp(bmp); |
| 518 |
self.dest |
| 519 |
} |
| 520 |
#[inline(always)] |
| 521 |
pub fn commit(self) -> &'a mut Utf16Destination<'b> { |
| 522 |
self.dest |
| 523 |
} |
| 524 |
} |
| 525 |
|
| 526 |
pub struct Utf16AstralHandle<'a, 'b> |
| 527 |
where |
| 528 |
'b: 'a, |
| 529 |
{ |
| 530 |
dest: &'a mut Utf16Destination<'b>, |
| 531 |
} |
| 532 |
|
| 533 |
impl<'a, 'b> Utf16AstralHandle<'a, 'b> |
| 534 |
where |
| 535 |
'b: 'a, |
| 536 |
{ |
| 537 |
#[inline(always)] |
| 538 |
fn new(dst: &'a mut Utf16Destination<'b>) -> Utf16AstralHandle<'a, 'b> { |
| 539 |
Utf16AstralHandle { dest: dst } |
| 540 |
} |
| 541 |
#[inline(always)] |
| 542 |
pub fn written(&self) -> usize { |
| 543 |
self.dest.written() |
| 544 |
} |
| 545 |
#[inline(always)] |
| 546 |
pub fn write_ascii(self, ascii: u8) -> &'a mut Utf16Destination<'b> { |
| 547 |
self.dest.write_ascii(ascii); |
| 548 |
self.dest |
| 549 |
} |
| 550 |
#[inline(always)] |
| 551 |
pub fn write_bmp(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 552 |
self.dest.write_bmp(bmp); |
| 553 |
self.dest |
| 554 |
} |
| 555 |
#[inline(always)] |
| 556 |
pub fn write_bmp_excl_ascii(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 557 |
self.dest.write_bmp_excl_ascii(bmp); |
| 558 |
self.dest |
| 559 |
} |
| 560 |
#[inline(always)] |
| 561 |
pub fn write_upper_bmp(self, bmp: u16) -> &'a mut Utf16Destination<'b> { |
| 562 |
self.dest.write_upper_bmp(bmp); |
| 563 |
self.dest |
| 564 |
} |
| 565 |
#[inline(always)] |
| 566 |
pub fn write_astral(self, astral: u32) -> &'a mut Utf16Destination<'b> { |
| 567 |
self.dest.write_astral(astral); |
| 568 |
self.dest |
| 569 |
} |
| 570 |
#[inline(always)] |
| 571 |
pub fn write_surrogate_pair(self, high: u16, low: u16) -> &'a mut Utf16Destination<'b> { |
| 572 |
self.dest.write_surrogate_pair(high, low); |
| 573 |
self.dest |
| 574 |
} |
| 575 |
#[inline(always)] |
| 576 |
pub fn write_big5_combination( |
| 577 |
self, |
| 578 |
combined: u16, |
| 579 |
combining: u16, |
| 580 |
) -> &'a mut Utf16Destination<'b> { |
| 581 |
self.dest.write_big5_combination(combined, combining); |
| 582 |
self.dest |
| 583 |
} |
| 584 |
#[inline(always)] |
| 585 |
pub fn commit(self) -> &'a mut Utf16Destination<'b> { |
| 586 |
self.dest |
| 587 |
} |
| 588 |
} |
| 589 |
|
| 590 |
pub struct Utf16Destination<'a> { |
| 591 |
slice: &'a mut [u16], |
| 592 |
pos: usize, |
| 593 |
} |
| 594 |
|
| 595 |
impl<'a> Utf16Destination<'a> { |
| 596 |
#[inline(always)] |
| 597 |
pub fn new(dst: &mut [u16]) -> Utf16Destination { |
| 598 |
Utf16Destination { slice: dst, pos: 0 } |
| 599 |
} |
| 600 |
#[inline(always)] |
| 601 |
pub fn check_space_bmp<'b>(&'b mut self) -> Space<Utf16BmpHandle<'b, 'a>> { |
| 602 |
if self.pos < self.slice.len() { |
| 603 |
Space::Available(Utf16BmpHandle::new(self)) |
| 604 |
} else { |
| 605 |
Space::Full(self.written()) |
| 606 |
} |
| 607 |
} |
| 608 |
#[inline(always)] |
| 609 |
pub fn check_space_astral<'b>(&'b mut self) -> Space<Utf16AstralHandle<'b, 'a>> { |
| 610 |
if self.pos + 1 < self.slice.len() { |
| 611 |
Space::Available(Utf16AstralHandle::new(self)) |
| 612 |
} else { |
| 613 |
Space::Full(self.written()) |
| 614 |
} |
| 615 |
} |
| 616 |
#[inline(always)] |
| 617 |
pub fn written(&self) -> usize { |
| 618 |
self.pos |
| 619 |
} |
| 620 |
#[inline(always)] |
| 621 |
fn write_code_unit(&mut self, u: u16) { |
| 622 |
unsafe { |
| 623 |
// OK, because we checked before handing out a handle. |
| 624 |
*(self.slice.get_unchecked_mut(self.pos)) = u; |
| 625 |
} |
| 626 |
self.pos += 1; |
| 627 |
} |
| 628 |
#[inline(always)] |
| 629 |
fn write_ascii(&mut self, ascii: u8) { |
| 630 |
debug_assert!(ascii < 0x80); |
| 631 |
self.write_code_unit(u16::from(ascii)); |
| 632 |
} |
| 633 |
#[inline(always)] |
| 634 |
fn write_bmp(&mut self, bmp: u16) { |
| 635 |
self.write_code_unit(bmp); |
| 636 |
} |
| 637 |
#[inline(always)] |
| 638 |
fn write_bmp_excl_ascii(&mut self, bmp: u16) { |
| 639 |
debug_assert!(bmp >= 0x80); |
| 640 |
self.write_code_unit(bmp); |
| 641 |
} |
| 642 |
#[inline(always)] |
| 643 |
fn write_mid_bmp(&mut self, bmp: u16) { |
| 644 |
debug_assert!(bmp >= 0x80); // XXX |
| 645 |
self.write_code_unit(bmp); |
| 646 |
} |
| 647 |
#[inline(always)] |
| 648 |
fn write_upper_bmp(&mut self, bmp: u16) { |
| 649 |
debug_assert!(bmp >= 0x80); |
| 650 |
self.write_code_unit(bmp); |
| 651 |
} |
| 652 |
#[inline(always)] |
| 653 |
fn write_astral(&mut self, astral: u32) { |
| 654 |
debug_assert!(astral > 0xFFFF); |
| 655 |
debug_assert!(astral <= 0x10_FFFF); |
| 656 |
self.write_code_unit((0xD7C0 + (astral >> 10)) as u16); |
| 657 |
self.write_code_unit((0xDC00 + (astral & 0x3FF)) as u16); |
| 658 |
} |
| 659 |
#[inline(always)] |
| 660 |
fn write_surrogate_pair(&mut self, high: u16, low: u16) { |
| 661 |
self.write_code_unit(high); |
| 662 |
self.write_code_unit(low); |
| 663 |
} |
| 664 |
#[inline(always)] |
| 665 |
fn write_big5_combination(&mut self, combined: u16, combining: u16) { |
| 666 |
self.write_bmp_excl_ascii(combined); |
| 667 |
self.write_bmp_excl_ascii(combining); |
| 668 |
} |
| 669 |
// Safety-usable invariant: CopyAsciiResult::GoOn will only contain bytes >=0x80 |
| 670 |
#[inline(always)] |
| 671 |
pub fn copy_ascii_from_check_space_bmp<'b>( |
| 672 |
&'b mut self, |
| 673 |
source: &mut ByteSource, |
| 674 |
) -> CopyAsciiResult<(DecoderResult, usize, usize), (u8, Utf16BmpHandle<'b, 'a>)> { |
| 675 |
let non_ascii_ret = { |
| 676 |
let src_remaining = &source.slice[source.pos..]; |
| 677 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 678 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 679 |
(DecoderResult::OutputFull, dst_remaining.len()) |
| 680 |
} else { |
| 681 |
(DecoderResult::InputEmpty, src_remaining.len()) |
| 682 |
}; |
| 683 |
// Safety: This function is documented as needing valid pointers for src/dest and len, which |
| 684 |
// is true since we've passed the minumum length of the two |
| 685 |
match unsafe { |
| 686 |
ascii_to_basic_latin(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 687 |
} { |
| 688 |
None => { |
| 689 |
source.pos += length; |
| 690 |
self.pos += length; |
| 691 |
return CopyAsciiResult::Stop((pending, source.pos, self.pos)); |
| 692 |
} |
| 693 |
// Safety: the function is documented as returning bytes >=0x80 in the Some |
| 694 |
Some((non_ascii, consumed)) => { |
| 695 |
source.pos += consumed; |
| 696 |
self.pos += consumed; |
| 697 |
source.pos += 1; // +1 for non_ascii |
| 698 |
// Safety: non-ascii bubbled out here |
| 699 |
non_ascii |
| 700 |
} |
| 701 |
} |
| 702 |
}; |
| 703 |
// Safety: non-ascii returned here |
| 704 |
CopyAsciiResult::GoOn((non_ascii_ret, Utf16BmpHandle::new(self))) |
| 705 |
} |
| 706 |
// Safety-usable invariant: CopyAsciiResult::GoOn will only contain bytes >=0x80 |
| 707 |
#[inline(always)] |
| 708 |
pub fn copy_ascii_from_check_space_astral<'b>( |
| 709 |
&'b mut self, |
| 710 |
source: &mut ByteSource, |
| 711 |
) -> CopyAsciiResult<(DecoderResult, usize, usize), (u8, Utf16AstralHandle<'b, 'a>)> { |
| 712 |
let non_ascii_ret = { |
| 713 |
let dst_len = self.slice.len(); |
| 714 |
let src_remaining = &source.slice[source.pos..]; |
| 715 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 716 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 717 |
(DecoderResult::OutputFull, dst_remaining.len()) |
| 718 |
} else { |
| 719 |
(DecoderResult::InputEmpty, src_remaining.len()) |
| 720 |
}; |
| 721 |
// Safety: This function is documented as needing valid pointers for src/dest and len, which |
| 722 |
// is true since we've passed the minumum length of the two |
| 723 |
match unsafe { |
| 724 |
ascii_to_basic_latin(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 725 |
} { |
| 726 |
None => { |
| 727 |
source.pos += length; |
| 728 |
self.pos += length; |
| 729 |
return CopyAsciiResult::Stop((pending, source.pos, self.pos)); |
| 730 |
} |
| 731 |
// Safety: the function is documented as returning bytes >=0x80 in the Some |
| 732 |
Some((non_ascii, consumed)) => { |
| 733 |
source.pos += consumed; |
| 734 |
self.pos += consumed; |
| 735 |
if self.pos + 1 < dst_len { |
| 736 |
source.pos += 1; // +1 for non_ascii |
| 737 |
// Safety: non-ascii bubbled out here |
| 738 |
non_ascii |
| 739 |
} else { |
| 740 |
return CopyAsciiResult::Stop(( |
| 741 |
DecoderResult::OutputFull, |
| 742 |
source.pos, |
| 743 |
self.pos, |
| 744 |
)); |
| 745 |
} |
| 746 |
} |
| 747 |
} |
| 748 |
}; |
| 749 |
// Safety: non-ascii returned here |
| 750 |
CopyAsciiResult::GoOn((non_ascii_ret, Utf16AstralHandle::new(self))) |
| 751 |
} |
| 752 |
#[inline(always)] |
| 753 |
pub fn copy_utf8_up_to_invalid_from(&mut self, source: &mut ByteSource) { |
| 754 |
let src_remaining = &source.slice[source.pos..]; |
| 755 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 756 |
let (read, written) = convert_utf8_to_utf16_up_to_invalid(src_remaining, dst_remaining); |
| 757 |
source.pos += read; |
| 758 |
self.pos += written; |
| 759 |
} |
| 760 |
#[inline(always)] |
| 761 |
pub fn copy_utf16_from<E: Endian>( |
| 762 |
&mut self, |
| 763 |
source: &mut ByteSource, |
| 764 |
) -> Option<(usize, usize)> { |
| 765 |
let src_remaining = &source.slice[source.pos..]; |
| 766 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 767 |
|
| 768 |
let mut src_unaligned = unsafe { |
| 769 |
UnalignedU16Slice::new( |
| 770 |
src_remaining.as_ptr(), |
| 771 |
::core::cmp::min(src_remaining.len() / 2, dst_remaining.len()), |
| 772 |
) |
| 773 |
}; |
| 774 |
if src_unaligned.len() == 0 { |
| 775 |
return None; |
| 776 |
} |
| 777 |
let last_unit = swap_if_opposite_endian::<E>(src_unaligned.at(src_unaligned.len() - 1)); |
| 778 |
if super::in_range16(last_unit, 0xD800, 0xDC00) { |
| 779 |
// Last code unit is a high surrogate. It might |
| 780 |
// legitimately form a pair later, so let's not |
| 781 |
// include it. |
| 782 |
src_unaligned.trim_last(); |
| 783 |
} |
| 784 |
let mut offset = 0usize; |
| 785 |
loop { |
| 786 |
if let Some((surrogate, bmp_len)) = { |
| 787 |
let src_left = src_unaligned.tail(offset); |
| 788 |
let dst_left = &mut dst_remaining[offset..src_unaligned.len()]; |
| 789 |
src_left.copy_bmp_to::<E>(dst_left) |
| 790 |
} { |
| 791 |
offset += bmp_len; // surrogate has not been consumed yet |
| 792 |
let second_pos = offset + 1; |
| 793 |
if surrogate > 0xDBFF || second_pos == src_unaligned.len() { |
| 794 |
// Unpaired surrogate |
| 795 |
source.pos += second_pos * 2; |
| 796 |
self.pos += offset; |
| 797 |
return Some((source.pos, self.pos)); |
| 798 |
} |
| 799 |
let second = swap_if_opposite_endian::<E>(src_unaligned.at(second_pos)); |
| 800 |
if !super::in_range16(second, 0xDC00, 0xE000) { |
| 801 |
// Unpaired surrogate |
| 802 |
source.pos += second_pos * 2; |
| 803 |
self.pos += offset; |
| 804 |
return Some((source.pos, self.pos)); |
| 805 |
} |
| 806 |
// `surrogate` was already speculatively written |
| 807 |
dst_remaining[second_pos] = second; |
| 808 |
offset += 2; |
| 809 |
continue; |
| 810 |
} else { |
| 811 |
source.pos += src_unaligned.len() * 2; |
| 812 |
self.pos += src_unaligned.len(); |
| 813 |
return None; |
| 814 |
} |
| 815 |
} |
| 816 |
} |
| 817 |
} |
| 818 |
|
| 819 |
// UTF-8 destination |
| 820 |
|
| 821 |
pub struct Utf8BmpHandle<'a, 'b> |
| 822 |
where |
| 823 |
'b: 'a, |
| 824 |
{ |
| 825 |
dest: &'a mut Utf8Destination<'b>, |
| 826 |
} |
| 827 |
|
| 828 |
impl<'a, 'b> Utf8BmpHandle<'a, 'b> |
| 829 |
where |
| 830 |
'b: 'a, |
| 831 |
{ |
| 832 |
#[inline(always)] |
| 833 |
fn new(dst: &'a mut Utf8Destination<'b>) -> Utf8BmpHandle<'a, 'b> { |
| 834 |
Utf8BmpHandle { dest: dst } |
| 835 |
} |
| 836 |
#[inline(always)] |
| 837 |
pub fn written(&self) -> usize { |
| 838 |
self.dest.written() |
| 839 |
} |
| 840 |
#[inline(always)] |
| 841 |
pub fn write_ascii(self, ascii: u8) -> &'a mut Utf8Destination<'b> { |
| 842 |
self.dest.write_ascii(ascii); |
| 843 |
self.dest |
| 844 |
} |
| 845 |
#[inline(always)] |
| 846 |
pub fn write_bmp(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 847 |
self.dest.write_bmp(bmp); |
| 848 |
self.dest |
| 849 |
} |
| 850 |
#[inline(always)] |
| 851 |
pub fn write_bmp_excl_ascii(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 852 |
self.dest.write_bmp_excl_ascii(bmp); |
| 853 |
self.dest |
| 854 |
} |
| 855 |
#[inline(always)] |
| 856 |
pub fn write_mid_bmp(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 857 |
self.dest.write_mid_bmp(bmp); |
| 858 |
self.dest |
| 859 |
} |
| 860 |
#[inline(always)] |
| 861 |
pub fn write_upper_bmp(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 862 |
self.dest.write_upper_bmp(bmp); |
| 863 |
self.dest |
| 864 |
} |
| 865 |
#[inline(always)] |
| 866 |
pub fn commit(self) -> &'a mut Utf8Destination<'b> { |
| 867 |
self.dest |
| 868 |
} |
| 869 |
} |
| 870 |
|
| 871 |
pub struct Utf8AstralHandle<'a, 'b> |
| 872 |
where |
| 873 |
'b: 'a, |
| 874 |
{ |
| 875 |
dest: &'a mut Utf8Destination<'b>, |
| 876 |
} |
| 877 |
|
| 878 |
impl<'a, 'b> Utf8AstralHandle<'a, 'b> |
| 879 |
where |
| 880 |
'b: 'a, |
| 881 |
{ |
| 882 |
#[inline(always)] |
| 883 |
fn new(dst: &'a mut Utf8Destination<'b>) -> Utf8AstralHandle<'a, 'b> { |
| 884 |
Utf8AstralHandle { dest: dst } |
| 885 |
} |
| 886 |
#[inline(always)] |
| 887 |
pub fn written(&self) -> usize { |
| 888 |
self.dest.written() |
| 889 |
} |
| 890 |
#[inline(always)] |
| 891 |
pub fn write_ascii(self, ascii: u8) -> &'a mut Utf8Destination<'b> { |
| 892 |
self.dest.write_ascii(ascii); |
| 893 |
self.dest |
| 894 |
} |
| 895 |
#[inline(always)] |
| 896 |
pub fn write_bmp(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 897 |
self.dest.write_bmp(bmp); |
| 898 |
self.dest |
| 899 |
} |
| 900 |
#[inline(always)] |
| 901 |
pub fn write_bmp_excl_ascii(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 902 |
self.dest.write_bmp_excl_ascii(bmp); |
| 903 |
self.dest |
| 904 |
} |
| 905 |
#[inline(always)] |
| 906 |
pub fn write_upper_bmp(self, bmp: u16) -> &'a mut Utf8Destination<'b> { |
| 907 |
self.dest.write_upper_bmp(bmp); |
| 908 |
self.dest |
| 909 |
} |
| 910 |
#[inline(always)] |
| 911 |
pub fn write_astral(self, astral: u32) -> &'a mut Utf8Destination<'b> { |
| 912 |
self.dest.write_astral(astral); |
| 913 |
self.dest |
| 914 |
} |
| 915 |
#[inline(always)] |
| 916 |
pub fn write_surrogate_pair(self, high: u16, low: u16) -> &'a mut Utf8Destination<'b> { |
| 917 |
self.dest.write_surrogate_pair(high, low); |
| 918 |
self.dest |
| 919 |
} |
| 920 |
#[inline(always)] |
| 921 |
pub fn write_big5_combination( |
| 922 |
self, |
| 923 |
combined: u16, |
| 924 |
combining: u16, |
| 925 |
) -> &'a mut Utf8Destination<'b> { |
| 926 |
self.dest.write_big5_combination(combined, combining); |
| 927 |
self.dest |
| 928 |
} |
| 929 |
#[inline(always)] |
| 930 |
pub fn commit(self) -> &'a mut Utf8Destination<'b> { |
| 931 |
self.dest |
| 932 |
} |
| 933 |
} |
| 934 |
|
| 935 |
pub struct Utf8Destination<'a> { |
| 936 |
slice: &'a mut [u8], |
| 937 |
pos: usize, |
| 938 |
} |
| 939 |
|
| 940 |
impl<'a> Utf8Destination<'a> { |
| 941 |
#[inline(always)] |
| 942 |
pub fn new(dst: &mut [u8]) -> Utf8Destination { |
| 943 |
Utf8Destination { slice: dst, pos: 0 } |
| 944 |
} |
| 945 |
#[inline(always)] |
| 946 |
pub fn check_space_bmp<'b>(&'b mut self) -> Space<Utf8BmpHandle<'b, 'a>> { |
| 947 |
if self.pos + 2 < self.slice.len() { |
| 948 |
Space::Available(Utf8BmpHandle::new(self)) |
| 949 |
} else { |
| 950 |
Space::Full(self.written()) |
| 951 |
} |
| 952 |
} |
| 953 |
#[inline(always)] |
| 954 |
pub fn check_space_astral<'b>(&'b mut self) -> Space<Utf8AstralHandle<'b, 'a>> { |
| 955 |
if self.pos + 3 < self.slice.len() { |
| 956 |
Space::Available(Utf8AstralHandle::new(self)) |
| 957 |
} else { |
| 958 |
Space::Full(self.written()) |
| 959 |
} |
| 960 |
} |
| 961 |
#[inline(always)] |
| 962 |
pub fn written(&self) -> usize { |
| 963 |
self.pos |
| 964 |
} |
| 965 |
#[inline(always)] |
| 966 |
fn write_code_unit(&mut self, u: u8) { |
| 967 |
unsafe { |
| 968 |
// OK, because we checked before handing out a handle. |
| 969 |
*(self.slice.get_unchecked_mut(self.pos)) = u; |
| 970 |
} |
| 971 |
self.pos += 1; |
| 972 |
} |
| 973 |
#[inline(always)] |
| 974 |
fn write_ascii(&mut self, ascii: u8) { |
| 975 |
debug_assert!(ascii < 0x80); |
| 976 |
self.write_code_unit(ascii); |
| 977 |
} |
| 978 |
#[inline(always)] |
| 979 |
fn write_bmp(&mut self, bmp: u16) { |
| 980 |
if bmp < 0x80u16 { |
| 981 |
self.write_ascii(bmp as u8); |
| 982 |
} else if bmp < 0x800u16 { |
| 983 |
self.write_mid_bmp(bmp); |
| 984 |
} else { |
| 985 |
self.write_upper_bmp(bmp); |
| 986 |
} |
| 987 |
} |
| 988 |
#[inline(always)] |
| 989 |
fn write_mid_bmp(&mut self, mid_bmp: u16) { |
| 990 |
debug_assert!(mid_bmp >= 0x80); |
| 991 |
debug_assert!(mid_bmp < 0x800); |
| 992 |
self.write_code_unit(((mid_bmp >> 6) | 0xC0) as u8); |
| 993 |
self.write_code_unit(((mid_bmp & 0x3F) | 0x80) as u8); |
| 994 |
} |
| 995 |
#[inline(always)] |
| 996 |
fn write_upper_bmp(&mut self, upper_bmp: u16) { |
| 997 |
debug_assert!(upper_bmp >= 0x800); |
| 998 |
self.write_code_unit(((upper_bmp >> 12) | 0xE0) as u8); |
| 999 |
self.write_code_unit((((upper_bmp & 0xFC0) >> 6) | 0x80) as u8); |
| 1000 |
self.write_code_unit(((upper_bmp & 0x3F) | 0x80) as u8); |
| 1001 |
} |
| 1002 |
#[inline(always)] |
| 1003 |
fn write_bmp_excl_ascii(&mut self, bmp: u16) { |
| 1004 |
if bmp < 0x800u16 { |
| 1005 |
self.write_mid_bmp(bmp); |
| 1006 |
} else { |
| 1007 |
self.write_upper_bmp(bmp); |
| 1008 |
} |
| 1009 |
} |
| 1010 |
#[inline(always)] |
| 1011 |
fn write_astral(&mut self, astral: u32) { |
| 1012 |
debug_assert!(astral > 0xFFFF); |
| 1013 |
debug_assert!(astral <= 0x10_FFFF); |
| 1014 |
self.write_code_unit(((astral >> 18) | 0xF0) as u8); |
| 1015 |
self.write_code_unit((((astral & 0x3F000) >> 12) | 0x80) as u8); |
| 1016 |
self.write_code_unit((((astral & 0xFC0) >> 6) | 0x80) as u8); |
| 1017 |
self.write_code_unit(((astral & 0x3F) | 0x80) as u8); |
| 1018 |
} |
| 1019 |
#[inline(always)] |
| 1020 |
pub fn write_surrogate_pair(&mut self, high: u16, low: u16) { |
| 1021 |
self.write_astral( |
| 1022 |
(u32::from(high) << 10) + u32::from(low) |
| 1023 |
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32), |
| 1024 |
); |
| 1025 |
} |
| 1026 |
#[inline(always)] |
| 1027 |
fn write_big5_combination(&mut self, combined: u16, combining: u16) { |
| 1028 |
self.write_mid_bmp(combined); |
| 1029 |
self.write_mid_bmp(combining); |
| 1030 |
} |
| 1031 |
#[inline(always)] |
| 1032 |
pub fn copy_ascii_from_check_space_bmp<'b>( |
| 1033 |
&'b mut self, |
| 1034 |
source: &mut ByteSource, |
| 1035 |
) -> CopyAsciiResult<(DecoderResult, usize, usize), (u8, Utf8BmpHandle<'b, 'a>)> { |
| 1036 |
let non_ascii_ret = { |
| 1037 |
let dst_len = self.slice.len(); |
| 1038 |
let src_remaining = &source.slice[source.pos..]; |
| 1039 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 1040 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1041 |
(DecoderResult::OutputFull, dst_remaining.len()) |
| 1042 |
} else { |
| 1043 |
(DecoderResult::InputEmpty, src_remaining.len()) |
| 1044 |
}; |
| 1045 |
match unsafe { |
| 1046 |
ascii_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1047 |
} { |
| 1048 |
None => { |
| 1049 |
source.pos += length; |
| 1050 |
self.pos += length; |
| 1051 |
return CopyAsciiResult::Stop((pending, source.pos, self.pos)); |
| 1052 |
} |
| 1053 |
Some((non_ascii, consumed)) => { |
| 1054 |
source.pos += consumed; |
| 1055 |
self.pos += consumed; |
| 1056 |
if self.pos + 2 < dst_len { |
| 1057 |
source.pos += 1; // +1 for non_ascii |
| 1058 |
non_ascii |
| 1059 |
} else { |
| 1060 |
return CopyAsciiResult::Stop(( |
| 1061 |
DecoderResult::OutputFull, |
| 1062 |
source.pos, |
| 1063 |
self.pos, |
| 1064 |
)); |
| 1065 |
} |
| 1066 |
} |
| 1067 |
} |
| 1068 |
}; |
| 1069 |
CopyAsciiResult::GoOn((non_ascii_ret, Utf8BmpHandle::new(self))) |
| 1070 |
} |
| 1071 |
#[inline(always)] |
| 1072 |
pub fn copy_ascii_from_check_space_astral<'b>( |
| 1073 |
&'b mut self, |
| 1074 |
source: &mut ByteSource, |
| 1075 |
) -> CopyAsciiResult<(DecoderResult, usize, usize), (u8, Utf8AstralHandle<'b, 'a>)> { |
| 1076 |
let non_ascii_ret = { |
| 1077 |
let dst_len = self.slice.len(); |
| 1078 |
let src_remaining = &source.slice[source.pos..]; |
| 1079 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 1080 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1081 |
(DecoderResult::OutputFull, dst_remaining.len()) |
| 1082 |
} else { |
| 1083 |
(DecoderResult::InputEmpty, src_remaining.len()) |
| 1084 |
}; |
| 1085 |
match unsafe { |
| 1086 |
ascii_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1087 |
} { |
| 1088 |
None => { |
| 1089 |
source.pos += length; |
| 1090 |
self.pos += length; |
| 1091 |
return CopyAsciiResult::Stop((pending, source.pos, self.pos)); |
| 1092 |
} |
| 1093 |
Some((non_ascii, consumed)) => { |
| 1094 |
source.pos += consumed; |
| 1095 |
self.pos += consumed; |
| 1096 |
if self.pos + 3 < dst_len { |
| 1097 |
source.pos += 1; // +1 for non_ascii |
| 1098 |
non_ascii |
| 1099 |
} else { |
| 1100 |
return CopyAsciiResult::Stop(( |
| 1101 |
DecoderResult::OutputFull, |
| 1102 |
source.pos, |
| 1103 |
self.pos, |
| 1104 |
)); |
| 1105 |
} |
| 1106 |
} |
| 1107 |
} |
| 1108 |
}; |
| 1109 |
CopyAsciiResult::GoOn((non_ascii_ret, Utf8AstralHandle::new(self))) |
| 1110 |
} |
| 1111 |
#[inline(always)] |
| 1112 |
pub fn copy_utf8_up_to_invalid_from(&mut self, source: &mut ByteSource) { |
| 1113 |
let src_remaining = &source.slice[source.pos..]; |
| 1114 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 1115 |
let min_len = ::core::cmp::min(src_remaining.len(), dst_remaining.len()); |
| 1116 |
// Validate first, then memcpy to let memcpy do its thing even for |
| 1117 |
// non-ASCII. (And potentially do something better than SSE2 for ASCII.) |
| 1118 |
let valid_len = utf8_valid_up_to(&src_remaining[..min_len]); |
| 1119 |
(&mut dst_remaining[..valid_len]).copy_from_slice(&src_remaining[..valid_len]); |
| 1120 |
source.pos += valid_len; |
| 1121 |
self.pos += valid_len; |
| 1122 |
} |
| 1123 |
#[inline(always)] |
| 1124 |
pub fn copy_utf16_from<E: Endian>( |
| 1125 |
&mut self, |
| 1126 |
source: &mut ByteSource, |
| 1127 |
) -> Option<(usize, usize)> { |
| 1128 |
let src_remaining = &source.slice[source.pos..]; |
| 1129 |
let dst_remaining = &mut self.slice[self.pos..]; |
| 1130 |
|
| 1131 |
let mut src_unaligned = |
| 1132 |
unsafe { UnalignedU16Slice::new(src_remaining.as_ptr(), src_remaining.len() / 2) }; |
| 1133 |
if src_unaligned.len() == 0 { |
| 1134 |
return None; |
| 1135 |
} |
| 1136 |
let mut last_unit = src_unaligned.at(src_unaligned.len() - 1); |
| 1137 |
if E::OPPOSITE_ENDIAN { |
| 1138 |
last_unit = last_unit.swap_bytes(); |
| 1139 |
} |
| 1140 |
if super::in_range16(last_unit, 0xD800, 0xDC00) { |
| 1141 |
// Last code unit is a high surrogate. It might |
| 1142 |
// legitimately form a pair later, so let's not |
| 1143 |
// include it. |
| 1144 |
src_unaligned.trim_last(); |
| 1145 |
} |
| 1146 |
let (read, written, had_error) = |
| 1147 |
convert_unaligned_utf16_to_utf8::<E>(src_unaligned, dst_remaining); |
| 1148 |
source.pos += read * 2; |
| 1149 |
self.pos += written; |
| 1150 |
if had_error { |
| 1151 |
Some((source.pos, self.pos)) |
| 1152 |
} else { |
| 1153 |
None |
| 1154 |
} |
| 1155 |
} |
| 1156 |
} |
| 1157 |
|
| 1158 |
// UTF-16 source |
| 1159 |
|
| 1160 |
pub struct Utf16Source<'a> { |
| 1161 |
slice: &'a [u16], |
| 1162 |
pos: usize, |
| 1163 |
old_pos: usize, |
| 1164 |
} |
| 1165 |
|
| 1166 |
impl<'a> Utf16Source<'a> { |
| 1167 |
#[inline(always)] |
| 1168 |
pub fn new(src: &[u16]) -> Utf16Source { |
| 1169 |
Utf16Source { |
| 1170 |
slice: src, |
| 1171 |
pos: 0, |
| 1172 |
old_pos: 0, |
| 1173 |
} |
| 1174 |
} |
| 1175 |
#[inline(always)] |
| 1176 |
pub fn check_available<'b>(&'b mut self) -> Space<Utf16ReadHandle<'b, 'a>> { |
| 1177 |
if self.pos < self.slice.len() { |
| 1178 |
Space::Available(Utf16ReadHandle::new(self)) |
| 1179 |
} else { |
| 1180 |
Space::Full(self.consumed()) |
| 1181 |
} |
| 1182 |
} |
| 1183 |
#[cfg_attr(feature = "cargo-clippy", allow(collapsible_if))] |
| 1184 |
#[inline(always)] |
| 1185 |
fn read(&mut self) -> char { |
| 1186 |
self.old_pos = self.pos; |
| 1187 |
let unit = self.slice[self.pos]; |
| 1188 |
self.pos += 1; |
| 1189 |
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800); |
| 1190 |
if unit_minus_surrogate_start > (0xDFFF - 0xD800) { |
| 1191 |
return unsafe { ::core::char::from_u32_unchecked(u32::from(unit)) }; |
| 1192 |
} |
| 1193 |
if unit_minus_surrogate_start <= (0xDBFF - 0xD800) { |
| 1194 |
// high surrogate |
| 1195 |
if self.pos < self.slice.len() { |
| 1196 |
let second = self.slice[self.pos]; |
| 1197 |
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00); |
| 1198 |
if second_minus_low_surrogate_start <= (0xDFFF - 0xDC00) { |
| 1199 |
// The next code unit is a low surrogate. Advance position. |
| 1200 |
self.pos += 1; |
| 1201 |
return unsafe { |
| 1202 |
::core::char::from_u32_unchecked( |
| 1203 |
(u32::from(unit) << 10) + u32::from(second) |
| 1204 |
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32), |
| 1205 |
) |
| 1206 |
}; |
| 1207 |
} |
| 1208 |
// The next code unit is not a low surrogate. Don't advance |
| 1209 |
// position and treat the high surrogate as unpaired. |
| 1210 |
// fall through |
| 1211 |
} |
| 1212 |
// Unpaired surrogate at the end of buffer, fall through |
| 1213 |
} |
| 1214 |
// Unpaired low surrogate |
| 1215 |
'\u{FFFD}' |
| 1216 |
} |
| 1217 |
#[cfg_attr(feature = "cargo-clippy", allow(collapsible_if))] |
| 1218 |
#[inline(always)] |
| 1219 |
fn read_enum(&mut self) -> Unicode { |
| 1220 |
self.old_pos = self.pos; |
| 1221 |
let unit = self.slice[self.pos]; |
| 1222 |
self.pos += 1; |
| 1223 |
if unit < 0x80 { |
| 1224 |
return Unicode::Ascii(unit as u8); |
| 1225 |
} |
| 1226 |
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800); |
| 1227 |
if unit_minus_surrogate_start > (0xDFFF - 0xD800) { |
| 1228 |
return Unicode::NonAscii(NonAscii::BmpExclAscii(unit)); |
| 1229 |
} |
| 1230 |
if unit_minus_surrogate_start <= (0xDBFF - 0xD800) { |
| 1231 |
// high surrogate |
| 1232 |
if self.pos < self.slice.len() { |
| 1233 |
let second = self.slice[self.pos]; |
| 1234 |
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00); |
| 1235 |
if second_minus_low_surrogate_start <= (0xDFFF - 0xDC00) { |
| 1236 |
// The next code unit is a low surrogate. Advance position. |
| 1237 |
self.pos += 1; |
| 1238 |
return Unicode::NonAscii(NonAscii::Astral(unsafe { |
| 1239 |
::core::char::from_u32_unchecked( |
| 1240 |
(u32::from(unit) << 10) + u32::from(second) |
| 1241 |
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32), |
| 1242 |
) |
| 1243 |
})); |
| 1244 |
} |
| 1245 |
// The next code unit is not a low surrogate. Don't advance |
| 1246 |
// position and treat the high surrogate as unpaired. |
| 1247 |
// fall through |
| 1248 |
} |
| 1249 |
// Unpaired surrogate at the end of buffer, fall through |
| 1250 |
} |
| 1251 |
// Unpaired low surrogate |
| 1252 |
Unicode::NonAscii(NonAscii::BmpExclAscii(0xFFFDu16)) |
| 1253 |
} |
| 1254 |
#[inline(always)] |
| 1255 |
fn unread(&mut self) -> usize { |
| 1256 |
self.pos = self.old_pos; |
| 1257 |
self.pos |
| 1258 |
} |
| 1259 |
#[inline(always)] |
| 1260 |
pub fn consumed(&self) -> usize { |
| 1261 |
self.pos |
| 1262 |
} |
| 1263 |
#[inline(always)] |
| 1264 |
pub fn copy_ascii_to_check_space_two<'b>( |
| 1265 |
&mut self, |
| 1266 |
dest: &'b mut ByteDestination<'a>, |
| 1267 |
) -> CopyAsciiResult<(EncoderResult, usize, usize), (NonAscii, ByteTwoHandle<'b, 'a>)> { |
| 1268 |
let non_ascii_ret = { |
| 1269 |
let dst_len = dest.slice.len(); |
| 1270 |
let src_remaining = &self.slice[self.pos..]; |
| 1271 |
let dst_remaining = &mut dest.slice[dest.pos..]; |
| 1272 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1273 |
(EncoderResult::OutputFull, dst_remaining.len()) |
| 1274 |
} else { |
| 1275 |
(EncoderResult::InputEmpty, src_remaining.len()) |
| 1276 |
}; |
| 1277 |
match unsafe { |
| 1278 |
basic_latin_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1279 |
} { |
| 1280 |
None => { |
| 1281 |
self.pos += length; |
| 1282 |
dest.pos += length; |
| 1283 |
return CopyAsciiResult::Stop((pending, self.pos, dest.pos)); |
| 1284 |
} |
| 1285 |
Some((non_ascii, consumed)) => { |
| 1286 |
self.pos += consumed; |
| 1287 |
dest.pos += consumed; |
| 1288 |
if dest.pos + 1 < dst_len { |
| 1289 |
self.pos += 1; // commit to reading `non_ascii` |
| 1290 |
let unit = non_ascii; |
| 1291 |
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800); |
| 1292 |
if unit_minus_surrogate_start > (0xDFFF - 0xD800) { |
| 1293 |
NonAscii::BmpExclAscii(unit) |
| 1294 |
} else if unit_minus_surrogate_start <= (0xDBFF - 0xD800) { |
| 1295 |
// high surrogate |
| 1296 |
if self.pos < self.slice.len() { |
| 1297 |
let second = self.slice[self.pos]; |
| 1298 |
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00); |
| 1299 |
if second_minus_low_surrogate_start <= (0xDFFF - 0xDC00) { |
| 1300 |
// The next code unit is a low surrogate. Advance position. |
| 1301 |
self.pos += 1; |
| 1302 |
NonAscii::Astral(unsafe { |
| 1303 |
::core::char::from_u32_unchecked( |
| 1304 |
(u32::from(unit) << 10) + u32::from(second) |
| 1305 |
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32), |
| 1306 |
) |
| 1307 |
}) |
| 1308 |
} else { |
| 1309 |
// The next code unit is not a low surrogate. Don't advance |
| 1310 |
// position and treat the high surrogate as unpaired. |
| 1311 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1312 |
} |
| 1313 |
} else { |
| 1314 |
// Unpaired surrogate at the end of the buffer. |
| 1315 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1316 |
} |
| 1317 |
} else { |
| 1318 |
// Unpaired low surrogate |
| 1319 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1320 |
} |
| 1321 |
} else { |
| 1322 |
return CopyAsciiResult::Stop(( |
| 1323 |
EncoderResult::OutputFull, |
| 1324 |
self.pos, |
| 1325 |
dest.pos, |
| 1326 |
)); |
| 1327 |
} |
| 1328 |
} |
| 1329 |
} |
| 1330 |
}; |
| 1331 |
CopyAsciiResult::GoOn((non_ascii_ret, ByteTwoHandle::new(dest))) |
| 1332 |
} |
| 1333 |
#[inline(always)] |
| 1334 |
pub fn copy_ascii_to_check_space_four<'b>( |
| 1335 |
&mut self, |
| 1336 |
dest: &'b mut ByteDestination<'a>, |
| 1337 |
) -> CopyAsciiResult<(EncoderResult, usize, usize), (NonAscii, ByteFourHandle<'b, 'a>)> { |
| 1338 |
let non_ascii_ret = { |
| 1339 |
let dst_len = dest.slice.len(); |
| 1340 |
let src_remaining = &self.slice[self.pos..]; |
| 1341 |
let dst_remaining = &mut dest.slice[dest.pos..]; |
| 1342 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1343 |
(EncoderResult::OutputFull, dst_remaining.len()) |
| 1344 |
} else { |
| 1345 |
(EncoderResult::InputEmpty, src_remaining.len()) |
| 1346 |
}; |
| 1347 |
match unsafe { |
| 1348 |
basic_latin_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1349 |
} { |
| 1350 |
None => { |
| 1351 |
self.pos += length; |
| 1352 |
dest.pos += length; |
| 1353 |
return CopyAsciiResult::Stop((pending, self.pos, dest.pos)); |
| 1354 |
} |
| 1355 |
Some((non_ascii, consumed)) => { |
| 1356 |
self.pos += consumed; |
| 1357 |
dest.pos += consumed; |
| 1358 |
if dest.pos + 3 < dst_len { |
| 1359 |
self.pos += 1; // commit to reading `non_ascii` |
| 1360 |
let unit = non_ascii; |
| 1361 |
let unit_minus_surrogate_start = unit.wrapping_sub(0xD800); |
| 1362 |
if unit_minus_surrogate_start > (0xDFFF - 0xD800) { |
| 1363 |
NonAscii::BmpExclAscii(unit) |
| 1364 |
} else if unit_minus_surrogate_start <= (0xDBFF - 0xD800) { |
| 1365 |
// high surrogate |
| 1366 |
if self.pos == self.slice.len() { |
| 1367 |
// Unpaired surrogate at the end of the buffer. |
| 1368 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1369 |
} else { |
| 1370 |
let second = self.slice[self.pos]; |
| 1371 |
let second_minus_low_surrogate_start = second.wrapping_sub(0xDC00); |
| 1372 |
if second_minus_low_surrogate_start <= (0xDFFF - 0xDC00) { |
| 1373 |
// The next code unit is a low surrogate. Advance position. |
| 1374 |
self.pos += 1; |
| 1375 |
NonAscii::Astral(unsafe { |
| 1376 |
::core::char::from_u32_unchecked( |
| 1377 |
(u32::from(unit) << 10) + u32::from(second) |
| 1378 |
- (((0xD800u32 << 10) - 0x1_0000u32) + 0xDC00u32), |
| 1379 |
) |
| 1380 |
}) |
| 1381 |
} else { |
| 1382 |
// The next code unit is not a low surrogate. Don't advance |
| 1383 |
// position and treat the high surrogate as unpaired. |
| 1384 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1385 |
} |
| 1386 |
} |
| 1387 |
} else { |
| 1388 |
// Unpaired low surrogate |
| 1389 |
NonAscii::BmpExclAscii(0xFFFDu16) |
| 1390 |
} |
| 1391 |
} else { |
| 1392 |
return CopyAsciiResult::Stop(( |
| 1393 |
EncoderResult::OutputFull, |
| 1394 |
self.pos, |
| 1395 |
dest.pos, |
| 1396 |
)); |
| 1397 |
} |
| 1398 |
} |
| 1399 |
} |
| 1400 |
}; |
| 1401 |
CopyAsciiResult::GoOn((non_ascii_ret, ByteFourHandle::new(dest))) |
| 1402 |
} |
| 1403 |
} |
| 1404 |
|
| 1405 |
pub struct Utf16ReadHandle<'a, 'b> |
| 1406 |
where |
| 1407 |
'b: 'a, |
| 1408 |
{ |
| 1409 |
source: &'a mut Utf16Source<'b>, |
| 1410 |
} |
| 1411 |
|
| 1412 |
impl<'a, 'b> Utf16ReadHandle<'a, 'b> |
| 1413 |
where |
| 1414 |
'b: 'a, |
| 1415 |
{ |
| 1416 |
#[inline(always)] |
| 1417 |
fn new(src: &'a mut Utf16Source<'b>) -> Utf16ReadHandle<'a, 'b> { |
| 1418 |
Utf16ReadHandle { source: src } |
| 1419 |
} |
| 1420 |
#[inline(always)] |
| 1421 |
pub fn read(self) -> (char, Utf16UnreadHandle<'a, 'b>) { |
| 1422 |
let character = self.source.read(); |
| 1423 |
let handle = Utf16UnreadHandle::new(self.source); |
| 1424 |
(character, handle) |
| 1425 |
} |
| 1426 |
#[inline(always)] |
| 1427 |
pub fn read_enum(self) -> (Unicode, Utf16UnreadHandle<'a, 'b>) { |
| 1428 |
let character = self.source.read_enum(); |
| 1429 |
let handle = Utf16UnreadHandle::new(self.source); |
| 1430 |
(character, handle) |
| 1431 |
} |
| 1432 |
#[inline(always)] |
| 1433 |
pub fn consumed(&self) -> usize { |
| 1434 |
self.source.consumed() |
| 1435 |
} |
| 1436 |
} |
| 1437 |
|
| 1438 |
pub struct Utf16UnreadHandle<'a, 'b> |
| 1439 |
where |
| 1440 |
'b: 'a, |
| 1441 |
{ |
| 1442 |
source: &'a mut Utf16Source<'b>, |
| 1443 |
} |
| 1444 |
|
| 1445 |
impl<'a, 'b> Utf16UnreadHandle<'a, 'b> |
| 1446 |
where |
| 1447 |
'b: 'a, |
| 1448 |
{ |
| 1449 |
#[inline(always)] |
| 1450 |
fn new(src: &'a mut Utf16Source<'b>) -> Utf16UnreadHandle<'a, 'b> { |
| 1451 |
Utf16UnreadHandle { source: src } |
| 1452 |
} |
| 1453 |
#[inline(always)] |
| 1454 |
pub fn unread(self) -> usize { |
| 1455 |
self.source.unread() |
| 1456 |
} |
| 1457 |
#[inline(always)] |
| 1458 |
pub fn consumed(&self) -> usize { |
| 1459 |
self.source.consumed() |
| 1460 |
} |
| 1461 |
#[inline(always)] |
| 1462 |
pub fn commit(self) -> &'a mut Utf16Source<'b> { |
| 1463 |
self.source |
| 1464 |
} |
| 1465 |
} |
| 1466 |
|
| 1467 |
// UTF-8 source |
| 1468 |
|
| 1469 |
pub struct Utf8Source<'a> { |
| 1470 |
slice: &'a [u8], |
| 1471 |
pos: usize, |
| 1472 |
old_pos: usize, |
| 1473 |
} |
| 1474 |
|
| 1475 |
impl<'a> Utf8Source<'a> { |
| 1476 |
#[inline(always)] |
| 1477 |
pub fn new(src: &str) -> Utf8Source { |
| 1478 |
Utf8Source { |
| 1479 |
slice: src.as_bytes(), |
| 1480 |
pos: 0, |
| 1481 |
old_pos: 0, |
| 1482 |
} |
| 1483 |
} |
| 1484 |
#[inline(always)] |
| 1485 |
pub fn check_available<'b>(&'b mut self) -> Space<Utf8ReadHandle<'b, 'a>> { |
| 1486 |
if self.pos < self.slice.len() { |
| 1487 |
Space::Available(Utf8ReadHandle::new(self)) |
| 1488 |
} else { |
| 1489 |
Space::Full(self.consumed()) |
| 1490 |
} |
| 1491 |
} |
| 1492 |
#[inline(always)] |
| 1493 |
fn read(&mut self) -> char { |
| 1494 |
self.old_pos = self.pos; |
| 1495 |
let unit = self.slice[self.pos]; |
| 1496 |
if unit < 0x80 { |
| 1497 |
self.pos += 1; |
| 1498 |
return char::from(unit); |
| 1499 |
} |
| 1500 |
if unit < 0xE0 { |
| 1501 |
let point = |
| 1502 |
((u32::from(unit) & 0x1F) << 6) | (u32::from(self.slice[self.pos + 1]) & 0x3F); |
| 1503 |
self.pos += 2; |
| 1504 |
return unsafe { ::core::char::from_u32_unchecked(point) }; |
| 1505 |
} |
| 1506 |
if unit < 0xF0 { |
| 1507 |
let point = ((u32::from(unit) & 0xF) << 12) |
| 1508 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 6) |
| 1509 |
| (u32::from(self.slice[self.pos + 2]) & 0x3F); |
| 1510 |
self.pos += 3; |
| 1511 |
return unsafe { ::core::char::from_u32_unchecked(point) }; |
| 1512 |
} |
| 1513 |
let point = ((u32::from(unit) & 0x7) << 18) |
| 1514 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 12) |
| 1515 |
| ((u32::from(self.slice[self.pos + 2]) & 0x3F) << 6) |
| 1516 |
| (u32::from(self.slice[self.pos + 3]) & 0x3F); |
| 1517 |
self.pos += 4; |
| 1518 |
unsafe { ::core::char::from_u32_unchecked(point) } |
| 1519 |
} |
| 1520 |
#[inline(always)] |
| 1521 |
fn read_enum(&mut self) -> Unicode { |
| 1522 |
self.old_pos = self.pos; |
| 1523 |
let unit = self.slice[self.pos]; |
| 1524 |
if unit < 0x80 { |
| 1525 |
self.pos += 1; |
| 1526 |
return Unicode::Ascii(unit); |
| 1527 |
} |
| 1528 |
if unit < 0xE0 { |
| 1529 |
let point = |
| 1530 |
((u16::from(unit) & 0x1F) << 6) | (u16::from(self.slice[self.pos + 1]) & 0x3F); |
| 1531 |
self.pos += 2; |
| 1532 |
return Unicode::NonAscii(NonAscii::BmpExclAscii(point)); |
| 1533 |
} |
| 1534 |
if unit < 0xF0 { |
| 1535 |
let point = ((u16::from(unit) & 0xF) << 12) |
| 1536 |
| ((u16::from(self.slice[self.pos + 1]) & 0x3F) << 6) |
| 1537 |
| (u16::from(self.slice[self.pos + 2]) & 0x3F); |
| 1538 |
self.pos += 3; |
| 1539 |
return Unicode::NonAscii(NonAscii::BmpExclAscii(point)); |
| 1540 |
} |
| 1541 |
let point = ((u32::from(unit) & 0x7) << 18) |
| 1542 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 12) |
| 1543 |
| ((u32::from(self.slice[self.pos + 2]) & 0x3F) << 6) |
| 1544 |
| (u32::from(self.slice[self.pos + 3]) & 0x3F); |
| 1545 |
self.pos += 4; |
| 1546 |
Unicode::NonAscii(NonAscii::Astral(unsafe { |
| 1547 |
::core::char::from_u32_unchecked(point) |
| 1548 |
})) |
| 1549 |
} |
| 1550 |
#[inline(always)] |
| 1551 |
fn unread(&mut self) -> usize { |
| 1552 |
self.pos = self.old_pos; |
| 1553 |
self.pos |
| 1554 |
} |
| 1555 |
#[inline(always)] |
| 1556 |
pub fn consumed(&self) -> usize { |
| 1557 |
self.pos |
| 1558 |
} |
| 1559 |
#[inline(always)] |
| 1560 |
pub fn copy_ascii_to_check_space_one<'b>( |
| 1561 |
&mut self, |
| 1562 |
dest: &'b mut ByteDestination<'a>, |
| 1563 |
) -> CopyAsciiResult<(EncoderResult, usize, usize), (NonAscii, ByteOneHandle<'b, 'a>)> { |
| 1564 |
let non_ascii_ret = { |
| 1565 |
let src_remaining = &self.slice[self.pos..]; |
| 1566 |
let dst_remaining = &mut dest.slice[dest.pos..]; |
| 1567 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1568 |
(EncoderResult::OutputFull, dst_remaining.len()) |
| 1569 |
} else { |
| 1570 |
(EncoderResult::InputEmpty, src_remaining.len()) |
| 1571 |
}; |
| 1572 |
match unsafe { |
| 1573 |
ascii_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1574 |
} { |
| 1575 |
None => { |
| 1576 |
self.pos += length; |
| 1577 |
dest.pos += length; |
| 1578 |
return CopyAsciiResult::Stop((pending, self.pos, dest.pos)); |
| 1579 |
} |
| 1580 |
Some((non_ascii, consumed)) => { |
| 1581 |
self.pos += consumed; |
| 1582 |
dest.pos += consumed; |
| 1583 |
// We don't need to check space in destination, because |
| 1584 |
// `ascii_to_ascii()` already did. |
| 1585 |
if non_ascii < 0xE0 { |
| 1586 |
let point = ((u16::from(non_ascii) & 0x1F) << 6) |
| 1587 |
| (u16::from(self.slice[self.pos + 1]) & 0x3F); |
| 1588 |
self.pos += 2; |
| 1589 |
NonAscii::BmpExclAscii(point) |
| 1590 |
} else if non_ascii < 0xF0 { |
| 1591 |
let point = ((u16::from(non_ascii) & 0xF) << 12) |
| 1592 |
| ((u16::from(self.slice[self.pos + 1]) & 0x3F) << 6) |
| 1593 |
| (u16::from(self.slice[self.pos + 2]) & 0x3F); |
| 1594 |
self.pos += 3; |
| 1595 |
NonAscii::BmpExclAscii(point) |
| 1596 |
} else { |
| 1597 |
let point = ((u32::from(non_ascii) & 0x7) << 18) |
| 1598 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 12) |
| 1599 |
| ((u32::from(self.slice[self.pos + 2]) & 0x3F) << 6) |
| 1600 |
| (u32::from(self.slice[self.pos + 3]) & 0x3F); |
| 1601 |
self.pos += 4; |
| 1602 |
NonAscii::Astral(unsafe { ::core::char::from_u32_unchecked(point) }) |
| 1603 |
} |
| 1604 |
} |
| 1605 |
} |
| 1606 |
}; |
| 1607 |
CopyAsciiResult::GoOn((non_ascii_ret, ByteOneHandle::new(dest))) |
| 1608 |
} |
| 1609 |
#[inline(always)] |
| 1610 |
pub fn copy_ascii_to_check_space_two<'b>( |
| 1611 |
&mut self, |
| 1612 |
dest: &'b mut ByteDestination<'a>, |
| 1613 |
) -> CopyAsciiResult<(EncoderResult, usize, usize), (NonAscii, ByteTwoHandle<'b, 'a>)> { |
| 1614 |
let non_ascii_ret = { |
| 1615 |
let dst_len = dest.slice.len(); |
| 1616 |
let src_remaining = &self.slice[self.pos..]; |
| 1617 |
let dst_remaining = &mut dest.slice[dest.pos..]; |
| 1618 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1619 |
(EncoderResult::OutputFull, dst_remaining.len()) |
| 1620 |
} else { |
| 1621 |
(EncoderResult::InputEmpty, src_remaining.len()) |
| 1622 |
}; |
| 1623 |
match unsafe { |
| 1624 |
ascii_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1625 |
} { |
| 1626 |
None => { |
| 1627 |
self.pos += length; |
| 1628 |
dest.pos += length; |
| 1629 |
return CopyAsciiResult::Stop((pending, self.pos, dest.pos)); |
| 1630 |
} |
| 1631 |
Some((non_ascii, consumed)) => { |
| 1632 |
self.pos += consumed; |
| 1633 |
dest.pos += consumed; |
| 1634 |
if dest.pos + 1 < dst_len { |
| 1635 |
if non_ascii < 0xE0 { |
| 1636 |
let point = ((u16::from(non_ascii) & 0x1F) << 6) |
| 1637 |
| (u16::from(self.slice[self.pos + 1]) & 0x3F); |
| 1638 |
self.pos += 2; |
| 1639 |
NonAscii::BmpExclAscii(point) |
| 1640 |
} else if non_ascii < 0xF0 { |
| 1641 |
let point = ((u16::from(non_ascii) & 0xF) << 12) |
| 1642 |
| ((u16::from(self.slice[self.pos + 1]) & 0x3F) << 6) |
| 1643 |
| (u16::from(self.slice[self.pos + 2]) & 0x3F); |
| 1644 |
self.pos += 3; |
| 1645 |
NonAscii::BmpExclAscii(point) |
| 1646 |
} else { |
| 1647 |
let point = ((u32::from(non_ascii) & 0x7) << 18) |
| 1648 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 12) |
| 1649 |
| ((u32::from(self.slice[self.pos + 2]) & 0x3F) << 6) |
| 1650 |
| (u32::from(self.slice[self.pos + 3]) & 0x3F); |
| 1651 |
self.pos += 4; |
| 1652 |
NonAscii::Astral(unsafe { ::core::char::from_u32_unchecked(point) }) |
| 1653 |
} |
| 1654 |
} else { |
| 1655 |
return CopyAsciiResult::Stop(( |
| 1656 |
EncoderResult::OutputFull, |
| 1657 |
self.pos, |
| 1658 |
dest.pos, |
| 1659 |
)); |
| 1660 |
} |
| 1661 |
} |
| 1662 |
} |
| 1663 |
}; |
| 1664 |
CopyAsciiResult::GoOn((non_ascii_ret, ByteTwoHandle::new(dest))) |
| 1665 |
} |
| 1666 |
#[inline(always)] |
| 1667 |
pub fn copy_ascii_to_check_space_four<'b>( |
| 1668 |
&mut self, |
| 1669 |
dest: &'b mut ByteDestination<'a>, |
| 1670 |
) -> CopyAsciiResult<(EncoderResult, usize, usize), (NonAscii, ByteFourHandle<'b, 'a>)> { |
| 1671 |
let non_ascii_ret = { |
| 1672 |
let dst_len = dest.slice.len(); |
| 1673 |
let src_remaining = &self.slice[self.pos..]; |
| 1674 |
let dst_remaining = &mut dest.slice[dest.pos..]; |
| 1675 |
let (pending, length) = if dst_remaining.len() < src_remaining.len() { |
| 1676 |
(EncoderResult::OutputFull, dst_remaining.len()) |
| 1677 |
} else { |
| 1678 |
(EncoderResult::InputEmpty, src_remaining.len()) |
| 1679 |
}; |
| 1680 |
match unsafe { |
| 1681 |
ascii_to_ascii(src_remaining.as_ptr(), dst_remaining.as_mut_ptr(), length) |
| 1682 |
} { |
| 1683 |
None => { |
| 1684 |
self.pos += length; |
| 1685 |
dest.pos += length; |
| 1686 |
return CopyAsciiResult::Stop((pending, self.pos, dest.pos)); |
| 1687 |
} |
| 1688 |
Some((non_ascii, consumed)) => { |
| 1689 |
self.pos += consumed; |
| 1690 |
dest.pos += consumed; |
| 1691 |
if dest.pos + 3 < dst_len { |
| 1692 |
if non_ascii < 0xE0 { |
| 1693 |
let point = ((u16::from(non_ascii) & 0x1F) << 6) |
| 1694 |
| (u16::from(self.slice[self.pos + 1]) & 0x3F); |
| 1695 |
self.pos += 2; |
| 1696 |
NonAscii::BmpExclAscii(point) |
| 1697 |
} else if non_ascii < 0xF0 { |
| 1698 |
let point = ((u16::from(non_ascii) & 0xF) << 12) |
| 1699 |
| ((u16::from(self.slice[self.pos + 1]) & 0x3F) << 6) |
| 1700 |
| (u16::from(self.slice[self.pos + 2]) & 0x3F); |
| 1701 |
self.pos += 3; |
| 1702 |
NonAscii::BmpExclAscii(point) |
| 1703 |
} else { |
| 1704 |
let point = ((u32::from(non_ascii) & 0x7) << 18) |
| 1705 |
| ((u32::from(self.slice[self.pos + 1]) & 0x3F) << 12) |
| 1706 |
| ((u32::from(self.slice[self.pos + 2]) & 0x3F) << 6) |
| 1707 |
| (u32::from(self.slice[self.pos + 3]) & 0x3F); |
| 1708 |
self.pos += 4; |
| 1709 |
NonAscii::Astral(unsafe { ::core::char::from_u32_unchecked(point) }) |
| 1710 |
} |
| 1711 |
} else { |
| 1712 |
return CopyAsciiResult::Stop(( |
| 1713 |
EncoderResult::OutputFull, |
| 1714 |
self.pos, |
| 1715 |
dest.pos, |
| 1716 |
)); |
| 1717 |
} |
| 1718 |
} |
| 1719 |
} |
| 1720 |
}; |
| 1721 |
CopyAsciiResult::GoOn((non_ascii_ret, ByteFourHandle::new(dest))) |
| 1722 |
} |
| 1723 |
} |
| 1724 |
|
| 1725 |
pub struct Utf8ReadHandle<'a, 'b> |
| 1726 |
where |
| 1727 |
'b: 'a, |
| 1728 |
{ |
| 1729 |
source: &'a mut Utf8Source<'b>, |
| 1730 |
} |
| 1731 |
|
| 1732 |
impl<'a, 'b> Utf8ReadHandle<'a, 'b> |
| 1733 |
where |
| 1734 |
'b: 'a, |
| 1735 |
{ |
| 1736 |
#[inline(always)] |
| 1737 |
fn new(src: &'a mut Utf8Source<'b>) -> Utf8ReadHandle<'a, 'b> { |
| 1738 |
Utf8ReadHandle { source: src } |
| 1739 |
} |
| 1740 |
#[inline(always)] |
| 1741 |
pub fn read(self) -> (char, Utf8UnreadHandle<'a, 'b>) { |
| 1742 |
let character = self.source.read(); |
| 1743 |
let handle = Utf8UnreadHandle::new(self.source); |
| 1744 |
(character, handle) |
| 1745 |
} |
| 1746 |
#[inline(always)] |
| 1747 |
pub fn read_enum(self) -> (Unicode, Utf8UnreadHandle<'a, 'b>) { |
| 1748 |
let character = self.source.read_enum(); |
| 1749 |
let handle = Utf8UnreadHandle::new(self.source); |
| 1750 |
(character, handle) |
| 1751 |
} |
| 1752 |
#[inline(always)] |
| 1753 |
pub fn consumed(&self) -> usize { |
| 1754 |
self.source.consumed() |
| 1755 |
} |
| 1756 |
} |
| 1757 |
|
| 1758 |
pub struct Utf8UnreadHandle<'a, 'b> |
| 1759 |
where |
| 1760 |
'b: 'a, |
| 1761 |
{ |
| 1762 |
source: &'a mut Utf8Source<'b>, |
| 1763 |
} |
| 1764 |
|
| 1765 |
impl<'a, 'b> Utf8UnreadHandle<'a, 'b> |
| 1766 |
where |
| 1767 |
'b: 'a, |
| 1768 |
{ |
| 1769 |
#[inline(always)] |
| 1770 |
fn new(src: &'a mut Utf8Source<'b>) -> Utf8UnreadHandle<'a, 'b> { |
| 1771 |
Utf8UnreadHandle { source: src } |
| 1772 |
} |
| 1773 |
#[inline(always)] |
| 1774 |
pub fn unread(self) -> usize { |
| 1775 |
self.source.unread() |
| 1776 |
} |
| 1777 |
#[inline(always)] |
| 1778 |
pub fn consumed(&self) -> usize { |
| 1779 |
self.source.consumed() |
| 1780 |
} |
| 1781 |
#[inline(always)] |
| 1782 |
pub fn commit(self) -> &'a mut Utf8Source<'b> { |
| 1783 |
self.source |
| 1784 |
} |
| 1785 |
} |
| 1786 |
|
| 1787 |
// Byte destination |
| 1788 |
|
| 1789 |
pub struct ByteOneHandle<'a, 'b> |
| 1790 |
where |
| 1791 |
'b: 'a, |
| 1792 |
{ |
| 1793 |
dest: &'a mut ByteDestination<'b>, |
| 1794 |
} |
| 1795 |
|
| 1796 |
impl<'a, 'b> ByteOneHandle<'a, 'b> |
| 1797 |
where |
| 1798 |
'b: 'a, |
| 1799 |
{ |
| 1800 |
#[inline(always)] |
| 1801 |
fn new(dst: &'a mut ByteDestination<'b>) -> ByteOneHandle<'a, 'b> { |
| 1802 |
ByteOneHandle { dest: dst } |
| 1803 |
} |
| 1804 |
#[inline(always)] |
| 1805 |
pub fn written(&self) -> usize { |
| 1806 |
self.dest.written() |
| 1807 |
} |
| 1808 |
#[inline(always)] |
| 1809 |
pub fn write_one(self, first: u8) -> &'a mut ByteDestination<'b> { |
| 1810 |
self.dest.write_one(first); |
| 1811 |
self.dest |
| 1812 |
} |
| 1813 |
} |
| 1814 |
|
| 1815 |
pub struct ByteTwoHandle<'a, 'b> |
| 1816 |
where |
| 1817 |
'b: 'a, |
| 1818 |
{ |
| 1819 |
dest: &'a mut ByteDestination<'b>, |
| 1820 |
} |
| 1821 |
|
| 1822 |
impl<'a, 'b> ByteTwoHandle<'a, 'b> |
| 1823 |
where |
| 1824 |
'b: 'a, |
| 1825 |
{ |
| 1826 |
#[inline(always)] |
| 1827 |
fn new(dst: &'a mut ByteDestination<'b>) -> ByteTwoHandle<'a, 'b> { |
| 1828 |
ByteTwoHandle { dest: dst } |
| 1829 |
} |
| 1830 |
#[inline(always)] |
| 1831 |
pub fn written(&self) -> usize { |
| 1832 |
self.dest.written() |
| 1833 |
} |
| 1834 |
#[inline(always)] |
| 1835 |
pub fn write_one(self, first: u8) -> &'a mut ByteDestination<'b> { |
| 1836 |
self.dest.write_one(first); |
| 1837 |
self.dest |
| 1838 |
} |
| 1839 |
#[inline(always)] |
| 1840 |
pub fn write_two(self, first: u8, second: u8) -> &'a mut ByteDestination<'b> { |
| 1841 |
self.dest.write_two(first, second); |
| 1842 |
self.dest |
| 1843 |
} |
| 1844 |
} |
| 1845 |
|
| 1846 |
pub struct ByteThreeHandle<'a, 'b> |
| 1847 |
where |
| 1848 |
'b: 'a, |
| 1849 |
{ |
| 1850 |
dest: &'a mut ByteDestination<'b>, |
| 1851 |
} |
| 1852 |
|
| 1853 |
impl<'a, 'b> ByteThreeHandle<'a, 'b> |
| 1854 |
where |
| 1855 |
'b: 'a, |
| 1856 |
{ |
| 1857 |
#[inline(always)] |
| 1858 |
fn new(dst: &'a mut ByteDestination<'b>) -> ByteThreeHandle<'a, 'b> { |
| 1859 |
ByteThreeHandle { dest: dst } |
| 1860 |
} |
| 1861 |
#[inline(always)] |
| 1862 |
pub fn written(&self) -> usize { |
| 1863 |
self.dest.written() |
| 1864 |
} |
| 1865 |
#[inline(always)] |
| 1866 |
pub fn write_one(self, first: u8) -> &'a mut ByteDestination<'b> { |
| 1867 |
self.dest.write_one(first); |
| 1868 |
self.dest |
| 1869 |
} |
| 1870 |
#[inline(always)] |
| 1871 |
pub fn write_two(self, first: u8, second: u8) -> &'a mut ByteDestination<'b> { |
| 1872 |
self.dest.write_two(first, second); |
| 1873 |
self.dest |
| 1874 |
} |
| 1875 |
#[inline(always)] |
| 1876 |
pub fn write_three(self, first: u8, second: u8, third: u8) -> &'a mut ByteDestination<'b> { |
| 1877 |
self.dest.write_three(first, second, third); |
| 1878 |
self.dest |
| 1879 |
} |
| 1880 |
#[inline(always)] |
| 1881 |
pub fn write_three_return_written(self, first: u8, second: u8, third: u8) -> usize { |
| 1882 |
self.dest.write_three(first, second, third); |
| 1883 |
self.dest.written() |
| 1884 |
} |
| 1885 |
} |
| 1886 |
|
| 1887 |
pub struct ByteFourHandle<'a, 'b> |
| 1888 |
where |
| 1889 |
'b: 'a, |
| 1890 |
{ |
| 1891 |
dest: &'a mut ByteDestination<'b>, |
| 1892 |
} |
| 1893 |
|
| 1894 |
impl<'a, 'b> ByteFourHandle<'a, 'b> |
| 1895 |
where |
| 1896 |
'b: 'a, |
| 1897 |
{ |
| 1898 |
#[inline(always)] |
| 1899 |
fn new(dst: &'a mut ByteDestination<'b>) -> ByteFourHandle<'a, 'b> { |
| 1900 |
ByteFourHandle { dest: dst } |
| 1901 |
} |
| 1902 |
#[inline(always)] |
| 1903 |
pub fn written(&self) -> usize { |
| 1904 |
self.dest.written() |
| 1905 |
} |
| 1906 |
#[inline(always)] |
| 1907 |
pub fn write_one(self, first: u8) -> &'a mut ByteDestination<'b> { |
| 1908 |
self.dest.write_one(first); |
| 1909 |
self.dest |
| 1910 |
} |
| 1911 |
#[inline(always)] |
| 1912 |
pub fn write_two(self, first: u8, second: u8) -> &'a mut ByteDestination<'b> { |
| 1913 |
self.dest.write_two(first, second); |
| 1914 |
self.dest |
| 1915 |
} |
| 1916 |
#[inline(always)] |
| 1917 |
pub fn write_four( |
| 1918 |
self, |
| 1919 |
first: u8, |
| 1920 |
second: u8, |
| 1921 |
third: u8, |
| 1922 |
fourth: u8, |
| 1923 |
) -> &'a mut ByteDestination<'b> { |
| 1924 |
self.dest.write_four(first, second, third, fourth); |
| 1925 |
self.dest |
| 1926 |
} |
| 1927 |
} |
| 1928 |
|
| 1929 |
pub struct ByteDestination<'a> { |
| 1930 |
slice: &'a mut [u8], |
| 1931 |
pos: usize, |
| 1932 |
} |
| 1933 |
|
| 1934 |
impl<'a> ByteDestination<'a> { |
| 1935 |
#[inline(always)] |
| 1936 |
pub fn new(dst: &mut [u8]) -> ByteDestination { |
| 1937 |
ByteDestination { slice: dst, pos: 0 } |
| 1938 |
} |
| 1939 |
#[inline(always)] |
| 1940 |
pub fn check_space_one<'b>(&'b mut self) -> Space<ByteOneHandle<'b, 'a>> { |
| 1941 |
if self.pos < self.slice.len() { |
| 1942 |
Space::Available(ByteOneHandle::new(self)) |
| 1943 |
} else { |
| 1944 |
Space::Full(self.written()) |
| 1945 |
} |
| 1946 |
} |
| 1947 |
#[inline(always)] |
| 1948 |
pub fn check_space_two<'b>(&'b mut self) -> Space<ByteTwoHandle<'b, 'a>> { |
| 1949 |
if self.pos + 1 < self.slice.len() { |
| 1950 |
Space::Available(ByteTwoHandle::new(self)) |
| 1951 |
} else { |
| 1952 |
Space::Full(self.written()) |
| 1953 |
} |
| 1954 |
} |
| 1955 |
#[inline(always)] |
| 1956 |
pub fn check_space_three<'b>(&'b mut self) -> Space<ByteThreeHandle<'b, 'a>> { |
| 1957 |
if self.pos + 2 < self.slice.len() { |
| 1958 |
Space::Available(ByteThreeHandle::new(self)) |
| 1959 |
} else { |
| 1960 |
Space::Full(self.written()) |
| 1961 |
} |
| 1962 |
} |
| 1963 |
#[inline(always)] |
| 1964 |
pub fn check_space_four<'b>(&'b mut self) -> Space<ByteFourHandle<'b, 'a>> { |
| 1965 |
if self.pos + 3 < self.slice.len() { |
| 1966 |
Space::Available(ByteFourHandle::new(self)) |
| 1967 |
} else { |
| 1968 |
Space::Full(self.written()) |
| 1969 |
} |
| 1970 |
} |
| 1971 |
#[inline(always)] |
| 1972 |
pub fn written(&self) -> usize { |
| 1973 |
self.pos |
| 1974 |
} |
| 1975 |
#[inline(always)] |
| 1976 |
fn write_one(&mut self, first: u8) { |
| 1977 |
self.slice[self.pos] = first; |
| 1978 |
self.pos += 1; |
| 1979 |
} |
| 1980 |
#[inline(always)] |
| 1981 |
fn write_two(&mut self, first: u8, second: u8) { |
| 1982 |
self.slice[self.pos] = first; |
| 1983 |
self.slice[self.pos + 1] = second; |
| 1984 |
self.pos += 2; |
| 1985 |
} |
| 1986 |
#[inline(always)] |
| 1987 |
fn write_three(&mut self, first: u8, second: u8, third: u8) { |
| 1988 |
self.slice[self.pos] = first; |
| 1989 |
self.slice[self.pos + 1] = second; |
| 1990 |
self.slice[self.pos + 2] = third; |
| 1991 |
self.pos += 3; |
| 1992 |
} |
| 1993 |
#[inline(always)] |
| 1994 |
fn write_four(&mut self, first: u8, second: u8, third: u8, fourth: u8) { |
| 1995 |
self.slice[self.pos] = first; |
| 1996 |
self.slice[self.pos + 1] = second; |
| 1997 |
self.slice[self.pos + 2] = third; |
| 1998 |
self.slice[self.pos + 3] = fourth; |
| 1999 |
self.pos += 4; |
| 2000 |
} |
| 2001 |
} |