| 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 |
use super::*; |
| 11 |
use crate::data::*; |
| 12 |
use crate::handles::*; |
| 13 |
use crate::variant::*; |
| 14 |
// Rust 1.14.0 requires the following despite the asterisk above. |
| 15 |
use super::in_inclusive_range; |
| 16 |
use super::in_inclusive_range16; |
| 17 |
|
| 18 |
pub struct ShiftJisDecoder { |
| 19 |
lead: Option<u8>, |
| 20 |
} |
| 21 |
|
| 22 |
impl ShiftJisDecoder { |
| 23 |
pub fn new() -> VariantDecoder { |
| 24 |
VariantDecoder::ShiftJis(ShiftJisDecoder { lead: None }) |
| 25 |
} |
| 26 |
|
| 27 |
pub fn in_neutral_state(&self) -> bool { |
| 28 |
self.lead.is_none() |
| 29 |
} |
| 30 |
|
| 31 |
fn plus_one_if_lead(&self, byte_length: usize) -> Option<usize> { |
| 32 |
byte_length.checked_add(match self.lead { |
| 33 |
None => 0, |
| 34 |
Some(_) => 1, |
| 35 |
}) |
| 36 |
} |
| 37 |
|
| 38 |
pub fn max_utf16_buffer_length(&self, byte_length: usize) -> Option<usize> { |
| 39 |
self.plus_one_if_lead(byte_length) |
| 40 |
} |
| 41 |
|
| 42 |
pub fn max_utf8_buffer_length_without_replacement(&self, byte_length: usize) -> Option<usize> { |
| 43 |
// worst case: 1 to 3 (half-width katakana) |
| 44 |
self.max_utf8_buffer_length(byte_length) |
| 45 |
} |
| 46 |
|
| 47 |
pub fn max_utf8_buffer_length(&self, byte_length: usize) -> Option<usize> { |
| 48 |
checked_mul(3, self.plus_one_if_lead(byte_length)) |
| 49 |
} |
| 50 |
|
| 51 |
ascii_compatible_two_byte_decoder_functions!( |
| 52 |
{ |
| 53 |
// If lead is between 0x81 and 0x9F, inclusive, |
| 54 |
// subtract offset 0x81. Else if lead is |
| 55 |
// between 0xE0 and 0xFC, inclusive, subtract |
| 56 |
// offset 0xC1. Else if lead is between |
| 57 |
// 0xA1 and 0xDF, inclusive, map to half-width |
| 58 |
// Katakana. Else if lead is 0x80, pass through. |
| 59 |
let mut non_ascii_minus_offset = |
| 60 |
non_ascii.wrapping_sub(0x81); |
| 61 |
if non_ascii_minus_offset > (0x9F - 0x81) { |
| 62 |
let non_ascii_minus_range_start = non_ascii.wrapping_sub(0xE0); |
| 63 |
if non_ascii_minus_range_start > (0xFC - 0xE0) { |
| 64 |
let non_ascii_minus_half_with_katakana_start = non_ascii.wrapping_sub(0xA1); |
| 65 |
if non_ascii_minus_half_with_katakana_start > (0xDF - 0xA1) { |
| 66 |
if non_ascii == 0x80 { |
| 67 |
handle.write_mid_bmp(0x80); |
| 68 |
// Not caring about optimizing subsequent non-ASCII |
| 69 |
continue 'outermost; |
| 70 |
} |
| 71 |
return (DecoderResult::Malformed(1, 0), |
| 72 |
source.consumed(), |
| 73 |
handle.written()); |
| 74 |
} |
| 75 |
handle.write_upper_bmp(0xFF61 + u16::from(non_ascii_minus_half_with_katakana_start)); |
| 76 |
// Not caring about optimizing subsequent non-ASCII |
| 77 |
continue 'outermost; |
| 78 |
} |
| 79 |
non_ascii_minus_offset = non_ascii - 0xC1; |
| 80 |
} |
| 81 |
non_ascii_minus_offset |
| 82 |
}, |
| 83 |
{ |
| 84 |
// If trail is between 0x40 and 0x7E, inclusive, |
| 85 |
// subtract offset 0x40. Else if trail is |
| 86 |
// between 0x80 and 0xFC, inclusive, subtract |
| 87 |
// offset 0x41. |
| 88 |
// Fast-track Hiragana (60% according to Lunde) |
| 89 |
// and Katakana (10% acconding to Lunde). |
| 90 |
// Hiragana doesn't cross 0x7F, but Katakana does. |
| 91 |
// We can check for Hiragana before normalizing |
| 92 |
// trail. |
| 93 |
let trail_minus_hiragana = byte.wrapping_sub(0x9F); |
| 94 |
if lead_minus_offset == 0x01 && trail_minus_hiragana < 0x53 { |
| 95 |
// Hiragana |
| 96 |
handle.write_upper_bmp(0x3041 + u16::from(trail_minus_hiragana)) |
| 97 |
} else { |
| 98 |
let mut trail_minus_offset = |
| 99 |
byte.wrapping_sub(0x40); |
| 100 |
if trail_minus_offset > (0x7E - 0x40) { |
| 101 |
let trail_minus_range_start = |
| 102 |
byte.wrapping_sub(0x80); |
| 103 |
if trail_minus_range_start > (0xFC - 0x80) { |
| 104 |
if byte < 0x80 { |
| 105 |
return (DecoderResult::Malformed(1, 0), |
| 106 |
unread_handle_trail.unread(), |
| 107 |
handle.written()); |
| 108 |
} |
| 109 |
return (DecoderResult::Malformed(2, 0), |
| 110 |
unread_handle_trail.consumed(), |
| 111 |
handle.written()); |
| 112 |
} |
| 113 |
trail_minus_offset = byte - 0x41; |
| 114 |
} |
| 115 |
if lead_minus_offset == 0x02 && |
| 116 |
trail_minus_offset < 0x56 { |
| 117 |
// Katakana |
| 118 |
handle.write_upper_bmp(0x30A1 + u16::from(trail_minus_offset)) |
| 119 |
} else { |
| 120 |
let pointer = lead_minus_offset as usize * |
| 121 |
188usize + |
| 122 |
trail_minus_offset as usize; |
| 123 |
let level1_pointer = pointer.wrapping_sub(1410); |
| 124 |
if level1_pointer < JIS0208_LEVEL1_KANJI.len() { |
| 125 |
handle.write_upper_bmp(JIS0208_LEVEL1_KANJI[level1_pointer]) |
| 126 |
} else { |
| 127 |
let level2_pointer = pointer.wrapping_sub(4418); |
| 128 |
if level2_pointer < |
| 129 |
JIS0208_LEVEL2_AND_ADDITIONAL_KANJI.len() { |
| 130 |
handle.write_upper_bmp(JIS0208_LEVEL2_AND_ADDITIONAL_KANJI[level2_pointer]) |
| 131 |
} else { |
| 132 |
let upper_ibm_pointer = pointer.wrapping_sub(10744); |
| 133 |
if upper_ibm_pointer < IBM_KANJI.len() { |
| 134 |
handle.write_upper_bmp(IBM_KANJI[upper_ibm_pointer]) |
| 135 |
} else { |
| 136 |
let lower_ibm_pointer = pointer.wrapping_sub(8272); |
| 137 |
if lower_ibm_pointer < IBM_KANJI.len() { |
| 138 |
handle.write_upper_bmp(IBM_KANJI[lower_ibm_pointer]) |
| 139 |
} else if in_inclusive_range(pointer, 8836, 10715) { |
| 140 |
handle.write_upper_bmp((0xE000 - 8836 + pointer) as u16) |
| 141 |
} else if let Some(bmp) = jis0208_symbol_decode(pointer) { |
| 142 |
handle.write_bmp_excl_ascii(bmp) |
| 143 |
} else if let Some(bmp) = jis0208_range_decode(pointer) { |
| 144 |
handle.write_bmp_excl_ascii(bmp) |
| 145 |
} else { |
| 146 |
if byte < 0x80 { |
| 147 |
return (DecoderResult::Malformed(1, 0), |
| 148 |
unread_handle_trail.unread(), |
| 149 |
handle.written()); |
| 150 |
} |
| 151 |
return (DecoderResult::Malformed(2, 0), |
| 152 |
unread_handle_trail.consumed(), |
| 153 |
handle.written()); |
| 154 |
} |
| 155 |
} |
| 156 |
} |
| 157 |
} |
| 158 |
} |
| 159 |
} |
| 160 |
}, |
| 161 |
self, |
| 162 |
non_ascii, |
| 163 |
byte, |
| 164 |
lead_minus_offset, |
| 165 |
unread_handle_trail, |
| 166 |
source, |
| 167 |
handle, |
| 168 |
'outermost, |
| 169 |
copy_ascii_from_check_space_bmp, |
| 170 |
check_space_bmp, |
| 171 |
false); |
| 172 |
} |
| 173 |
|
| 174 |
#[cfg(feature = "fast-kanji-encode")] |
| 175 |
#[inline(always)] |
| 176 |
fn encode_kanji(bmp: u16) -> Option<(u8, u8)> { |
| 177 |
jis0208_kanji_shift_jis_encode(bmp) |
| 178 |
} |
| 179 |
|
| 180 |
#[cfg(not(feature = "fast-kanji-encode"))] |
| 181 |
#[inline(always)] |
| 182 |
fn encode_kanji(bmp: u16) -> Option<(u8, u8)> { |
| 183 |
if let Some((lead, trail)) = jis0208_level1_kanji_shift_jis_encode(bmp) { |
| 184 |
return Some((lead, trail)); |
| 185 |
} |
| 186 |
let pointer = if 0x4EDD == bmp { |
| 187 |
// Ideograph on the symbol row! |
| 188 |
23 |
| 189 |
} else if let Some(pos) = jis0208_level2_and_additional_kanji_encode(bmp) { |
| 190 |
4418 + pos |
| 191 |
} else if let Some(pos) = position(&IBM_KANJI[..], bmp) { |
| 192 |
10744 + pos |
| 193 |
} else { |
| 194 |
return None; |
| 195 |
}; |
| 196 |
let lead = pointer / 188; |
| 197 |
let lead_offset = if lead < 0x1F { 0x81usize } else { 0xC1usize }; |
| 198 |
let trail = pointer % 188; |
| 199 |
let trail_offset = if trail < 0x3F { 0x40usize } else { 0x41usize }; |
| 200 |
Some(((lead + lead_offset) as u8, (trail + trail_offset) as u8)) |
| 201 |
} |
| 202 |
|
| 203 |
pub struct ShiftJisEncoder; |
| 204 |
|
| 205 |
impl ShiftJisEncoder { |
| 206 |
pub fn new(encoding: &'static Encoding) -> Encoder { |
| 207 |
Encoder::new(encoding, VariantEncoder::ShiftJis(ShiftJisEncoder)) |
| 208 |
} |
| 209 |
|
| 210 |
pub fn max_buffer_length_from_utf16_without_replacement( |
| 211 |
&self, |
| 212 |
u16_length: usize, |
| 213 |
) -> Option<usize> { |
| 214 |
u16_length.checked_mul(2) |
| 215 |
} |
| 216 |
|
| 217 |
pub fn max_buffer_length_from_utf8_without_replacement( |
| 218 |
&self, |
| 219 |
byte_length: usize, |
| 220 |
) -> Option<usize> { |
| 221 |
byte_length.checked_add(1) |
| 222 |
} |
| 223 |
|
| 224 |
ascii_compatible_bmp_encoder_functions!( |
| 225 |
{ |
| 226 |
// Lunde says 60% Hiragana, 30% Kanji, 10% Katakana |
| 227 |
let bmp_minus_hiragana = bmp.wrapping_sub(0x3041); |
| 228 |
if bmp_minus_hiragana < 0x53 { |
| 229 |
handle.write_two(0x82, 0x9F + bmp_minus_hiragana as u8) |
| 230 |
} else if in_inclusive_range16(bmp, 0x4E00, 0x9FA0) { |
| 231 |
if let Some((lead, trail)) = encode_kanji(bmp) { |
| 232 |
handle.write_two(lead, trail) |
| 233 |
} else { |
| 234 |
return ( |
| 235 |
EncoderResult::unmappable_from_bmp(bmp), |
| 236 |
source.consumed(), |
| 237 |
handle.written(), |
| 238 |
); |
| 239 |
} |
| 240 |
} else { |
| 241 |
let bmp_minus_katakana = bmp.wrapping_sub(0x30A1); |
| 242 |
if bmp_minus_katakana < 0x56 { |
| 243 |
let trail_offset = if bmp_minus_katakana < 0x3F { |
| 244 |
0x40 |
| 245 |
} else { |
| 246 |
0x41 |
| 247 |
}; |
| 248 |
handle.write_two(0x83, (trail_offset + bmp_minus_katakana) as u8) |
| 249 |
} else { |
| 250 |
let bmp_minus_space = bmp.wrapping_sub(0x3000); |
| 251 |
if bmp_minus_space < 3 { |
| 252 |
// fast-track common punctuation |
| 253 |
handle.write_two(0x81, 0x40 + bmp_minus_space as u8) |
| 254 |
} else if bmp == 0xA5 { |
| 255 |
handle.write_one(0x5Cu8) |
| 256 |
} else if bmp == 0x80 { |
| 257 |
handle.write_one(0x80u8) |
| 258 |
} else if bmp == 0x203E { |
| 259 |
handle.write_one(0x7Eu8) |
| 260 |
} else if in_inclusive_range16(bmp, 0xFF61, 0xFF9F) { |
| 261 |
handle.write_one((bmp - (0xFF61 - 0xA1)) as u8) |
| 262 |
} else if bmp == 0x2212 { |
| 263 |
handle.write_two(0x81u8, 0x7Cu8) |
| 264 |
} else { |
| 265 |
let bmp_minus_roman = bmp.wrapping_sub(0x2170); |
| 266 |
let pointer = if bmp_minus_roman <= (0x2179 - 0x2170) { |
| 267 |
10716 + bmp_minus_roman as usize |
| 268 |
} else if let Some(pointer) = jis0208_range_encode(bmp) { |
| 269 |
pointer |
| 270 |
} else if in_inclusive_range16(bmp, 0xFA0E, 0xFA2D) |
| 271 |
|| bmp == 0xF929 |
| 272 |
|| bmp == 0xF9DC |
| 273 |
{ |
| 274 |
// Guaranteed to be found in IBM_KANJI |
| 275 |
let pos = position(&IBM_KANJI[..], bmp).unwrap(); |
| 276 |
10744 + pos |
| 277 |
} else if let Some(pointer) = jis0208_symbol_encode(bmp) { |
| 278 |
pointer |
| 279 |
} else { |
| 280 |
return ( |
| 281 |
EncoderResult::unmappable_from_bmp(bmp), |
| 282 |
source.consumed(), |
| 283 |
handle.written(), |
| 284 |
); |
| 285 |
}; |
| 286 |
let lead = pointer / 188; |
| 287 |
let lead_offset = if lead < 0x1F { 0x81usize } else { 0xC1usize }; |
| 288 |
let trail = pointer % 188; |
| 289 |
let trail_offset = if trail < 0x3F { 0x40usize } else { 0x41usize }; |
| 290 |
handle.write_two((lead + lead_offset) as u8, (trail + trail_offset) as u8) |
| 291 |
} |
| 292 |
} |
| 293 |
} |
| 294 |
}, |
| 295 |
bmp, |
| 296 |
self, |
| 297 |
source, |
| 298 |
handle, |
| 299 |
copy_ascii_to_check_space_two, |
| 300 |
check_space_two, |
| 301 |
false |
| 302 |
); |
| 303 |
} |
| 304 |
|
| 305 |
// Any copyright to the test code below this comment is dedicated to the |
| 306 |
// Public Domain. http://creativecommons.org/publicdomain/zero/1.0/ |
| 307 |
|
| 308 |
#[cfg(all(test, feature = "alloc"))] |
| 309 |
mod tests { |
| 310 |
use super::super::testing::*; |
| 311 |
use super::super::*; |
| 312 |
|
| 313 |
fn decode_shift_jis(bytes: &[u8], expect: &str) { |
| 314 |
decode(SHIFT_JIS, bytes, expect); |
| 315 |
} |
| 316 |
|
| 317 |
fn encode_shift_jis(string: &str, expect: &[u8]) { |
| 318 |
encode(SHIFT_JIS, string, expect); |
| 319 |
} |
| 320 |
|
| 321 |
#[test] |
| 322 |
fn test_shift_jis_decode() { |
| 323 |
// Empty |
| 324 |
decode_shift_jis(b"", &""); |
| 325 |
|
| 326 |
// ASCII |
| 327 |
decode_shift_jis(b"\x61\x62", "\u{0061}\u{0062}"); |
| 328 |
|
| 329 |
// Half-width |
| 330 |
decode_shift_jis(b"\xA1", "\u{FF61}"); |
| 331 |
decode_shift_jis(b"\xDF", "\u{FF9F}"); |
| 332 |
decode_shift_jis(b"\xA0", "\u{FFFD}"); |
| 333 |
decode_shift_jis(b"\xE0", "\u{FFFD}"); |
| 334 |
decode_shift_jis(b"\xA0+", "\u{FFFD}+"); |
| 335 |
decode_shift_jis(b"\xE0+", "\u{FFFD}+"); |
| 336 |
|
| 337 |
// EUDC |
| 338 |
decode_shift_jis(b"\xF0\x40", "\u{E000}"); |
| 339 |
decode_shift_jis(b"\xF9\xFC", "\u{E757}"); |
| 340 |
decode_shift_jis(b"\xEF\xFC", "\u{FFFD}"); |
| 341 |
decode_shift_jis(b"\xFA\x40", "\u{2170}"); |
| 342 |
|
| 343 |
// JIS 0208 |
| 344 |
decode_shift_jis(b"\x81\x40", "\u{3000}"); |
| 345 |
decode_shift_jis(b"\x81\x3F", "\u{FFFD}?"); |
| 346 |
decode_shift_jis(b"\xEE\xFC", "\u{FF02}"); |
| 347 |
decode_shift_jis(b"\xEE\xFD", "\u{FFFD}"); |
| 348 |
decode_shift_jis(b"\xFA\x40", "\u{2170}"); |
| 349 |
decode_shift_jis(b"\xFA\x3F", "\u{FFFD}?"); |
| 350 |
decode_shift_jis(b"\xFC\x4B", "\u{9ED1}"); |
| 351 |
decode_shift_jis(b"\xFC\x4C", "\u{FFFD}L"); |
| 352 |
// |
| 353 |
} |
| 354 |
|
| 355 |
#[test] |
| 356 |
fn test_shift_jis_encode() { |
| 357 |
// Empty |
| 358 |
encode_shift_jis("", b""); |
| 359 |
|
| 360 |
// ASCII |
| 361 |
encode_shift_jis("\u{0061}\u{0062}", b"\x61\x62"); |
| 362 |
|
| 363 |
// Exceptional code points |
| 364 |
encode_shift_jis("\u{0080}", b"\x80"); |
| 365 |
encode_shift_jis("\u{00A5}", b"\x5C"); |
| 366 |
encode_shift_jis("\u{203E}", b"\x7E"); |
| 367 |
encode_shift_jis("\u{2212}", b"\x81\x7C"); |
| 368 |
|
| 369 |
// Half-width |
| 370 |
encode_shift_jis("\u{FF61}", b"\xA1"); |
| 371 |
encode_shift_jis("\u{FF9F}", b"\xDF"); |
| 372 |
|
| 373 |
// EUDC |
| 374 |
encode_shift_jis("\u{E000}", b""); |
| 375 |
encode_shift_jis("\u{E757}", b""); |
| 376 |
|
| 377 |
// JIS 0212 |
| 378 |
encode_shift_jis("\u{02D8}", b"˘"); |
| 379 |
|
| 380 |
// JIS 0208 |
| 381 |
encode_shift_jis("\u{3000}", b"\x81\x40"); |
| 382 |
encode_shift_jis("\u{FF02}", b"\xFA\x57"); |
| 383 |
encode_shift_jis("\u{2170}", b"\xFA\x40"); |
| 384 |
encode_shift_jis("\u{9ED1}", b"\xFC\x4B"); |
| 385 |
} |
| 386 |
|
| 387 |
#[test] |
| 388 |
#[cfg_attr(miri, ignore)] // Miri is too slow |
| 389 |
fn test_shift_jis_decode_all() { |
| 390 |
let input = include_bytes!("test_data/shift_jis_in.txt"); |
| 391 |
let expectation = include_str!("test_data/shift_jis_in_ref.txt"); |
| 392 |
let (cow, had_errors) = SHIFT_JIS.decode_without_bom_handling(input); |
| 393 |
assert!(had_errors, "Should have had errors."); |
| 394 |
assert_eq!(&cow[..], expectation); |
| 395 |
} |
| 396 |
|
| 397 |
#[test] |
| 398 |
#[cfg_attr(miri, ignore)] // Miri is too slow |
| 399 |
fn test_shift_jis_encode_all() { |
| 400 |
let input = include_str!("test_data/shift_jis_out.txt"); |
| 401 |
let expectation = include_bytes!("test_data/shift_jis_out_ref.txt"); |
| 402 |
let (cow, encoding, had_errors) = SHIFT_JIS.encode(input); |
| 403 |
assert!(!had_errors, "Should not have had errors."); |
| 404 |
assert_eq!(encoding, SHIFT_JIS); |
| 405 |
assert_eq!(&cow[..], &expectation[..]); |
| 406 |
} |
| 407 |
|
| 408 |
#[test] |
| 409 |
fn test_shift_jis_half_width_katakana_length() { |
| 410 |
let mut output = [0u8; 20]; |
| 411 |
let mut decoder = SHIFT_JIS.new_decoder(); |
| 412 |
{ |
| 413 |
let needed = decoder |
| 414 |
.max_utf8_buffer_length_without_replacement(1) |
| 415 |
.unwrap(); |
| 416 |
let (result, read, written) = |
| 417 |
decoder.decode_to_utf8_without_replacement(b"\xA1", &mut output[..needed], true); |
| 418 |
assert_eq!(result, DecoderResult::InputEmpty); |
| 419 |
assert_eq!(read, 1); |
| 420 |
assert_eq!(written, 3); |
| 421 |
assert_eq!(output[0], 0xEF); |
| 422 |
assert_eq!(output[1], 0xBD); |
| 423 |
assert_eq!(output[2], 0xA1); |
| 424 |
} |
| 425 |
} |
| 426 |
} |