| 1 |
// Copyright 2019 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 |
//! `chardetng` is a character encoding detector for legacy Web content. |
| 11 |
//! |
| 12 |
//! It is optimized for binary size in applications that already depend |
| 13 |
//! on `encoding_rs` for other reasons. |
| 14 |
|
| 15 |
use encoding_rs::Decoder; |
| 16 |
use encoding_rs::DecoderResult; |
| 17 |
use encoding_rs::Encoding; |
| 18 |
use encoding_rs::BIG5; |
| 19 |
use encoding_rs::EUC_JP; |
| 20 |
use encoding_rs::EUC_KR; |
| 21 |
use encoding_rs::GBK; |
| 22 |
use encoding_rs::ISO_2022_JP; |
| 23 |
use encoding_rs::ISO_8859_8; |
| 24 |
use encoding_rs::SHIFT_JIS; |
| 25 |
use encoding_rs::UTF_8; |
| 26 |
use encoding_rs::WINDOWS_1255; |
| 27 |
|
| 28 |
mod data; |
| 29 |
mod tld; |
| 30 |
use data::*; |
| 31 |
use tld::classify_tld; |
| 32 |
use tld::Tld; |
| 33 |
|
| 34 |
const LATIN_ADJACENCY_PENALTY: i64 = -50; |
| 35 |
|
| 36 |
const IMPLAUSIBILITY_PENALTY: i64 = -220; |
| 37 |
|
| 38 |
const ORDINAL_BONUS: i64 = 300; |
| 39 |
|
| 40 |
/// Must match the ISO-8859-2 score for " Š ". Note: There |
| 41 |
/// are four Slovenian Wikipedia list page titles where the |
| 42 |
/// list is split by letter so that Š stands alone for the |
| 43 |
/// list part for Š. Let's assume that's a special case not |
| 44 |
/// worth detecting even though the copyright sign detection |
| 45 |
/// makes Slovenian title detection round to one percentage |
| 46 |
/// point worse. |
| 47 |
const COPYRIGHT_BONUS: i64 = 222; |
| 48 |
|
| 49 |
const IMPLAUSIBLE_LATIN_CASE_TRANSITION_PENALTY: i64 = -180; |
| 50 |
|
| 51 |
const NON_LATIN_CAPITALIZATION_BONUS: i64 = 40; |
| 52 |
|
| 53 |
const NON_LATIN_ALL_CAPS_PENALTY: i64 = -40; |
| 54 |
|
| 55 |
const NON_LATIN_MIXED_CASE_PENALTY: i64 = -20; |
| 56 |
|
| 57 |
// Manually calibrated relative to windows-1256 Arabic |
| 58 |
const CJK_BASE_SCORE: i64 = 41; |
| 59 |
|
| 60 |
const CJK_SECONDARY_BASE_SCORE: i64 = 20; // Was 20 |
| 61 |
|
| 62 |
const SHIFT_JIS_SCORE_PER_KANA: i64 = 20; |
| 63 |
|
| 64 |
const SHIFT_JIS_SCORE_PER_LEVEL_1_KANJI: i64 = CJK_BASE_SCORE; |
| 65 |
|
| 66 |
const SHIFT_JIS_SCORE_PER_LEVEL_2_KANJI: i64 = CJK_SECONDARY_BASE_SCORE; |
| 67 |
|
| 68 |
// Manually calibrated relative to windows-1256 Persian and Urdu |
| 69 |
const SHIFT_JIS_INITIAL_HALF_WIDTH_KATAKANA_PENALTY: i64 = -75; |
| 70 |
|
| 71 |
const HALF_WIDTH_KATAKANA_SCORE: i64 = 1; |
| 72 |
|
| 73 |
// Unclear if this is a good idea; seems not harmful, but can't be sure. |
| 74 |
const HALF_WIDTH_KATAKANA_VOICING_SCORE: i64 = 10; |
| 75 |
|
| 76 |
const SHIFT_JIS_PUA_PENALTY: i64 = -(CJK_BASE_SCORE * 10); // Should this be larger? |
| 77 |
|
| 78 |
const SHIFT_JIS_EXTENSION_PENALTY: i64 = SHIFT_JIS_PUA_PENALTY * 2; |
| 79 |
|
| 80 |
const SHIFT_JIS_SINGLE_BYTE_EXTENSION_PENALTY: i64 = SHIFT_JIS_EXTENSION_PENALTY; |
| 81 |
|
| 82 |
const EUC_JP_SCORE_PER_KANA: i64 = CJK_BASE_SCORE + (CJK_BASE_SCORE / 3); // Relative to Big5 |
| 83 |
|
| 84 |
const EUC_JP_SCORE_PER_NEAR_OBSOLETE_KANA: i64 = CJK_BASE_SCORE - 1; |
| 85 |
|
| 86 |
const EUC_JP_SCORE_PER_LEVEL_1_KANJI: i64 = CJK_BASE_SCORE; |
| 87 |
|
| 88 |
const EUC_JP_SCORE_PER_LEVEL_2_KANJI: i64 = CJK_SECONDARY_BASE_SCORE; |
| 89 |
|
| 90 |
const EUC_JP_SCORE_PER_OTHER_KANJI: i64 = CJK_SECONDARY_BASE_SCORE / 4; |
| 91 |
|
| 92 |
const EUC_JP_INITIAL_KANA_PENALTY: i64 = -((CJK_BASE_SCORE / 3) + 1); |
| 93 |
|
| 94 |
const EUC_JP_EXTENSION_PENALTY: i64 = -(CJK_BASE_SCORE * 50); // Needs to be more severe than for Shift_JIS to avoid misdetecting EUC-KR! |
| 95 |
|
| 96 |
const BIG5_SCORE_PER_LEVEL_1_HANZI: i64 = CJK_BASE_SCORE; |
| 97 |
|
| 98 |
const BIG5_SCORE_PER_OTHER_HANZI: i64 = CJK_SECONDARY_BASE_SCORE; |
| 99 |
|
| 100 |
const BIG5_PUA_PENALTY: i64 = -(CJK_BASE_SCORE * 30); // More severe than other PUA penalties to avoid misdetecting EUC-KR! (25 as the multiplier is too little) |
| 101 |
|
| 102 |
const BIG5_SINGLE_BYTE_EXTENSION_PENALTY: i64 = -(CJK_BASE_SCORE * 40); |
| 103 |
|
| 104 |
const EUC_KR_SCORE_PER_EUC_HANGUL: i64 = CJK_BASE_SCORE + 1; |
| 105 |
|
| 106 |
const EUC_KR_SCORE_PER_NON_EUC_HANGUL: i64 = CJK_SECONDARY_BASE_SCORE / 5; |
| 107 |
|
| 108 |
const EUC_KR_SCORE_PER_HANJA: i64 = CJK_SECONDARY_BASE_SCORE / 2; |
| 109 |
|
| 110 |
const EUC_KR_HANJA_AFTER_HANGUL_PENALTY: i64 = -(CJK_BASE_SCORE * 10); |
| 111 |
|
| 112 |
const EUC_KR_LONG_WORD_PENALTY: i64 = -6; |
| 113 |
|
| 114 |
const EUC_KR_PUA_PENALTY: i64 = GBK_PUA_PENALTY - 1; // Break tie in favor of GBK |
| 115 |
|
| 116 |
const EUC_KR_MAC_KOREAN_PENALTY: i64 = EUC_KR_PUA_PENALTY * 2; |
| 117 |
|
| 118 |
const EUC_KR_SINGLE_BYTE_EXTENSION_PENALTY: i64 = EUC_KR_MAC_KOREAN_PENALTY; |
| 119 |
|
| 120 |
const GBK_SCORE_PER_LEVEL_1: i64 = CJK_BASE_SCORE; |
| 121 |
|
| 122 |
const GBK_SCORE_PER_LEVEL_2: i64 = CJK_SECONDARY_BASE_SCORE; |
| 123 |
|
| 124 |
const GBK_SCORE_PER_NON_EUC: i64 = CJK_SECONDARY_BASE_SCORE / 4; |
| 125 |
|
| 126 |
const GBK_PUA_PENALTY: i64 = -(CJK_BASE_SCORE * 10); // Factor should be at least 2, but should it be larger? |
| 127 |
|
| 128 |
const GBK_SINGLE_BYTE_EXTENSION_PENALTY: i64 = GBK_PUA_PENALTY * 4; |
| 129 |
|
| 130 |
const CJK_LATIN_ADJACENCY_PENALTY: i64 = -CJK_BASE_SCORE; // smaller penalty than LATIN_ADJACENCY_PENALTY |
| 131 |
|
| 132 |
const CJ_PUNCTUATION: i64 = CJK_BASE_SCORE / 2; |
| 133 |
|
| 134 |
const CJK_OTHER: i64 = CJK_SECONDARY_BASE_SCORE / 4; |
| 135 |
|
| 136 |
/// Latin letter caseless class |
| 137 |
const LATIN_LETTER: u8 = 1; |
| 138 |
|
| 139 |
fn contains_upper_case_period_or_non_ascii(label: &[u8]) -> bool { |
| 140 |
for &b in label.into_iter() { |
| 141 |
if b >= 0x80 { |
| 142 |
return true; |
| 143 |
} |
| 144 |
if b == b'.' { |
| 145 |
return true; |
| 146 |
} |
| 147 |
if b >= b'A' && b <= b'Z' { |
| 148 |
return true; |
| 149 |
} |
| 150 |
} |
| 151 |
false |
| 152 |
} |
| 153 |
|
| 154 |
// For Latin, we only penalize pairwise bad transitions |
| 155 |
// if one participant is non-ASCII. This avoids violating |
| 156 |
// the principle that ASCII pairs never contribute to the |
| 157 |
// score. (Maybe that's a bad principle, though!) |
| 158 |
#[derive(PartialEq)] |
| 159 |
enum LatinCaseState { |
| 160 |
Space, |
| 161 |
Upper, |
| 162 |
Lower, |
| 163 |
AllCaps, |
| 164 |
} |
| 165 |
|
| 166 |
// Fon non-Latin, we calculate case-related penalty |
| 167 |
// or bonus on a per-non-Latin-word basis. |
| 168 |
#[derive(PartialEq)] |
| 169 |
enum NonLatinCaseState { |
| 170 |
Space, |
| 171 |
Upper, |
| 172 |
Lower, |
| 173 |
UpperLower, |
| 174 |
AllCaps, |
| 175 |
Mix, |
| 176 |
} |
| 177 |
|
| 178 |
struct NonLatinCasedCandidate { |
| 179 |
data: &'static SingleByteData, |
| 180 |
prev: u8, |
| 181 |
case_state: NonLatinCaseState, |
| 182 |
prev_ascii: bool, |
| 183 |
current_word_len: u64, |
| 184 |
longest_word: u64, |
| 185 |
ibm866: bool, |
| 186 |
prev_was_a0: bool, // Only used with IBM866 |
| 187 |
} |
| 188 |
|
| 189 |
impl NonLatinCasedCandidate { |
| 190 |
fn new(data: &'static SingleByteData) -> Self { |
| 191 |
NonLatinCasedCandidate { |
| 192 |
data: data, |
| 193 |
prev: 0, |
| 194 |
case_state: NonLatinCaseState::Space, |
| 195 |
prev_ascii: true, |
| 196 |
current_word_len: 0, |
| 197 |
longest_word: 0, |
| 198 |
ibm866: data == &SINGLE_BYTE_DATA[IBM866_INDEX], |
| 199 |
prev_was_a0: false, |
| 200 |
} |
| 201 |
} |
| 202 |
|
| 203 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 204 |
let mut score = 0i64; |
| 205 |
for &b in buffer { |
| 206 |
let class = self.data.classify(b); |
| 207 |
if class == 255 { |
| 208 |
return None; |
| 209 |
} |
| 210 |
let caseless_class = class & 0x7F; |
| 211 |
|
| 212 |
let ascii = b < 0x80; |
| 213 |
let ascii_pair = self.prev_ascii && ascii; |
| 214 |
|
| 215 |
let non_ascii_alphabetic = self.data.is_non_latin_alphabetic(caseless_class, false); |
| 216 |
|
| 217 |
// The purpose of this state machine is to avoid misdetecting Greek as |
| 218 |
// Cyrillic by: |
| 219 |
// |
| 220 |
// * Giving a small bonus to words that start with an upper-case letter |
| 221 |
// and are lower-case for the rest. |
| 222 |
// * Giving a large penalty to start with one lower-case letter followed |
| 223 |
// by all upper-case (obviously upper and lower case inverted, which |
| 224 |
// unfortunately is possible due to KOI8-U). |
| 225 |
// * Giving a small per-word penalty to all-uppercase KOI8-U (to favor |
| 226 |
// all-lowercase Greek over all-caps KOI8-U). |
| 227 |
// * Giving large penalties for mixed-case other than initial upper-case. |
| 228 |
// This also helps relative to non-cased encodings. |
| 229 |
|
| 230 |
// ASCII doesn't participate in non-Latin casing. |
| 231 |
if caseless_class == LATIN_LETTER { |
| 232 |
// Latin |
| 233 |
// Mark this word as a mess. If there end up being non-Latin |
| 234 |
// letters in this word, the ASCII-adjacency penalty gets |
| 235 |
// applied to Latin/non-Latin pairs and the mix penalty |
| 236 |
// to non-Latin/non-Latin pairs. |
| 237 |
// XXX Apply penalty here |
| 238 |
self.case_state = NonLatinCaseState::Mix; |
| 239 |
} else if !non_ascii_alphabetic { |
| 240 |
// Space |
| 241 |
match self.case_state { |
| 242 |
NonLatinCaseState::Space |
| 243 |
| NonLatinCaseState::Upper |
| 244 |
| NonLatinCaseState::Lower => {} |
| 245 |
NonLatinCaseState::UpperLower => { |
| 246 |
// Intentionally applied only once per word. |
| 247 |
score += NON_LATIN_CAPITALIZATION_BONUS; |
| 248 |
} |
| 249 |
NonLatinCaseState::AllCaps => { |
| 250 |
// Intentionally applied only once per word. |
| 251 |
if self.data == &SINGLE_BYTE_DATA[KOI8_U_INDEX] { |
| 252 |
// Apply only to KOI8-U. |
| 253 |
score += NON_LATIN_ALL_CAPS_PENALTY; |
| 254 |
} |
| 255 |
} |
| 256 |
NonLatinCaseState::Mix => { |
| 257 |
// Per letter |
| 258 |
score += NON_LATIN_MIXED_CASE_PENALTY * (self.current_word_len as i64); |
| 259 |
} |
| 260 |
} |
| 261 |
self.case_state = NonLatinCaseState::Space; |
| 262 |
} else if (class >> 7) == 0 { |
| 263 |
// Lower case |
| 264 |
match self.case_state { |
| 265 |
NonLatinCaseState::Space => { |
| 266 |
self.case_state = NonLatinCaseState::Lower; |
| 267 |
} |
| 268 |
NonLatinCaseState::Upper => { |
| 269 |
self.case_state = NonLatinCaseState::UpperLower; |
| 270 |
} |
| 271 |
NonLatinCaseState::Lower |
| 272 |
| NonLatinCaseState::UpperLower |
| 273 |
| NonLatinCaseState::Mix => {} |
| 274 |
NonLatinCaseState::AllCaps => { |
| 275 |
self.case_state = NonLatinCaseState::Mix; |
| 276 |
} |
| 277 |
} |
| 278 |
} else { |
| 279 |
// Upper case |
| 280 |
match self.case_state { |
| 281 |
NonLatinCaseState::Space => { |
| 282 |
self.case_state = NonLatinCaseState::Upper; |
| 283 |
} |
| 284 |
NonLatinCaseState::Upper => { |
| 285 |
self.case_state = NonLatinCaseState::AllCaps; |
| 286 |
} |
| 287 |
NonLatinCaseState::Lower | NonLatinCaseState::UpperLower => { |
| 288 |
self.case_state = NonLatinCaseState::Mix; |
| 289 |
} |
| 290 |
NonLatinCaseState::AllCaps | NonLatinCaseState::Mix => {} |
| 291 |
} |
| 292 |
} |
| 293 |
|
| 294 |
// XXX Apply penalty if > 16 |
| 295 |
if non_ascii_alphabetic { |
| 296 |
self.current_word_len += 1; |
| 297 |
} else { |
| 298 |
if self.current_word_len > self.longest_word { |
| 299 |
self.longest_word = self.current_word_len; |
| 300 |
} |
| 301 |
self.current_word_len = 0; |
| 302 |
} |
| 303 |
|
| 304 |
let is_a0 = b == 0xA0; |
| 305 |
if !ascii_pair { |
| 306 |
// 0xA0 is no-break space in many other encodings, so avoid |
| 307 |
// assigning score to IBM866 when 0xA0 occurs next to itself |
| 308 |
// or a space-like byte. |
| 309 |
if !(self.ibm866 |
| 310 |
&& ((is_a0 && (self.prev_was_a0 || self.prev == 0)) |
| 311 |
|| caseless_class == 0 && self.prev_was_a0)) |
| 312 |
{ |
| 313 |
score += self.data.score(caseless_class, self.prev, false); |
| 314 |
} |
| 315 |
|
| 316 |
if self.prev == LATIN_LETTER && non_ascii_alphabetic { |
| 317 |
score += LATIN_ADJACENCY_PENALTY; |
| 318 |
} else if caseless_class == LATIN_LETTER |
| 319 |
&& self.data.is_non_latin_alphabetic(self.prev, false) |
| 320 |
{ |
| 321 |
score += LATIN_ADJACENCY_PENALTY; |
| 322 |
} |
| 323 |
} |
| 324 |
|
| 325 |
self.prev_ascii = ascii; |
| 326 |
self.prev = caseless_class; |
| 327 |
self.prev_was_a0 = is_a0; |
| 328 |
} |
| 329 |
Some(score) |
| 330 |
} |
| 331 |
} |
| 332 |
|
| 333 |
enum OrdinalState { |
| 334 |
Other, |
| 335 |
Space, |
| 336 |
PeriodAfterN, |
| 337 |
OrdinalExpectingSpace, |
| 338 |
OrdinalExpectingSpaceUndoImplausibility, |
| 339 |
OrdinalExpectingSpaceOrDigit, |
| 340 |
OrdinalExpectingSpaceOrDigitUndoImplausibily, |
| 341 |
UpperN, |
| 342 |
LowerN, |
| 343 |
FeminineAbbreviationStartLetter, |
| 344 |
Digit, |
| 345 |
Roman, |
| 346 |
Copyright, |
| 347 |
} |
| 348 |
|
| 349 |
struct LatinCandidate { |
| 350 |
data: &'static SingleByteData, |
| 351 |
prev: u8, |
| 352 |
case_state: LatinCaseState, |
| 353 |
prev_non_ascii: u32, |
| 354 |
ordinal_state: OrdinalState, // Used only when `windows1252 == true` |
| 355 |
windows1252: bool, |
| 356 |
} |
| 357 |
|
| 358 |
impl LatinCandidate { |
| 359 |
fn new(data: &'static SingleByteData) -> Self { |
| 360 |
LatinCandidate { |
| 361 |
data: data, |
| 362 |
prev: 0, |
| 363 |
case_state: LatinCaseState::Space, |
| 364 |
prev_non_ascii: 0, |
| 365 |
ordinal_state: OrdinalState::Space, |
| 366 |
windows1252: data == &SINGLE_BYTE_DATA[WINDOWS_1252_INDEX], |
| 367 |
} |
| 368 |
} |
| 369 |
|
| 370 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 371 |
let mut score = 0i64; |
| 372 |
for &b in buffer { |
| 373 |
let class = self.data.classify(b); |
| 374 |
if class == 255 { |
| 375 |
return None; |
| 376 |
} |
| 377 |
let caseless_class = class & 0x7F; |
| 378 |
|
| 379 |
let ascii = b < 0x80; |
| 380 |
let ascii_pair = self.prev_non_ascii == 0 && ascii; |
| 381 |
|
| 382 |
let non_ascii_penalty = match self.prev_non_ascii { |
| 383 |
0 | 1 | 2 => 0, |
| 384 |
3 => -5, |
| 385 |
4 => -20, |
| 386 |
_ => -200, |
| 387 |
}; |
| 388 |
score += non_ascii_penalty; |
| 389 |
// XXX if has Vietnamese-only characters and word length > 7, |
| 390 |
// apply penalty |
| 391 |
|
| 392 |
if !self.data.is_latin_alphabetic(caseless_class) { |
| 393 |
self.case_state = LatinCaseState::Space; |
| 394 |
} else if (class >> 7) == 0 { |
| 395 |
// Penalizing lower case after two upper case |
| 396 |
// is important for avoiding misdetecting |
| 397 |
// windows-1250 as windows-1252 (byte 0x9F). |
| 398 |
if self.case_state == LatinCaseState::AllCaps && !ascii_pair { |
| 399 |
score += IMPLAUSIBLE_LATIN_CASE_TRANSITION_PENALTY; |
| 400 |
} |
| 401 |
self.case_state = LatinCaseState::Lower; |
| 402 |
} else { |
| 403 |
match self.case_state { |
| 404 |
LatinCaseState::Space => { |
| 405 |
self.case_state = LatinCaseState::Upper; |
| 406 |
} |
| 407 |
LatinCaseState::Upper | LatinCaseState::AllCaps => { |
| 408 |
self.case_state = LatinCaseState::AllCaps; |
| 409 |
} |
| 410 |
LatinCaseState::Lower => { |
| 411 |
if !ascii_pair { |
| 412 |
// XXX How bad is this for Irish Gaelic? |
| 413 |
score += IMPLAUSIBLE_LATIN_CASE_TRANSITION_PENALTY; |
| 414 |
} |
| 415 |
self.case_state = LatinCaseState::Upper; |
| 416 |
} |
| 417 |
} |
| 418 |
} |
| 419 |
|
| 420 |
// Treat pairing space-like, which can be non-ASCII, with ASCII as |
| 421 |
// ASCIIish enough not to get a score in order to avoid giving |
| 422 |
// ASCII i and I in windows-1254 next to windows-125x apostrophe/quote |
| 423 |
// a score. This avoids detecting English I’ as Turkish. |
| 424 |
let ascii_ish_pair = ascii_pair |
| 425 |
|| (ascii && self.prev == 0) |
| 426 |
|| (caseless_class == 0 && self.prev_non_ascii == 0); |
| 427 |
|
| 428 |
if !ascii_ish_pair { |
| 429 |
score += self.data.score(caseless_class, self.prev, false); |
| 430 |
} |
| 431 |
|
| 432 |
if self.windows1252 { |
| 433 |
// This state machine assigns score to the sequences |
| 434 |
// * " º " (Spanish) |
| 435 |
// * " ª " (Spanish) |
| 436 |
// * ".ª " (Spanish) |
| 437 |
// * ".º " (Spanish) |
| 438 |
// * "n.º1" (Spanish) |
| 439 |
// * " Mª " (Spanish) |
| 440 |
// * " Dª " (Spanish) |
| 441 |
// * " Nª " (Spanish) |
| 442 |
// * " Sª " (Spanish) |
| 443 |
// * " 3º " (Italian, where 3 is an ASCII digit) |
| 444 |
// * " 3ª " (Italian, where 3 is an ASCII digit) |
| 445 |
// * " Xº " (Italian, where X is a small Roman numeral) |
| 446 |
// * " Xª " (Italian, where X is a small Roman numeral) |
| 447 |
// * " Nº1" (Italian, where 1 is an ASCII digit) |
| 448 |
// * " Nº " (Italian) |
| 449 |
// * " © " (otherwise ASCII-only) |
| 450 |
// which are problematic to deal with by pairwise scoring |
| 451 |
// without messing up Romanian detection. |
| 452 |
// Initial sc |
| 453 |
match self.ordinal_state { |
| 454 |
OrdinalState::Other => { |
| 455 |
if caseless_class == 0 { |
| 456 |
self.ordinal_state = OrdinalState::Space; |
| 457 |
} |
| 458 |
} |
| 459 |
OrdinalState::Space => { |
| 460 |
if caseless_class == 0 { |
| 461 |
// pass |
| 462 |
} else if b == 0xAA || b == 0xBA { |
| 463 |
self.ordinal_state = OrdinalState::OrdinalExpectingSpace; |
| 464 |
} else if b == b'M' || b == b'D' || b == b'S' { |
| 465 |
self.ordinal_state = OrdinalState::FeminineAbbreviationStartLetter; |
| 466 |
} else if b == b'N' { |
| 467 |
// numero or Nuestra |
| 468 |
self.ordinal_state = OrdinalState::UpperN; |
| 469 |
} else if b == b'n' { |
| 470 |
// numero |
| 471 |
self.ordinal_state = OrdinalState::LowerN; |
| 472 |
} else if caseless_class == (ASCII_DIGIT as u8) { |
| 473 |
self.ordinal_state = OrdinalState::Digit; |
| 474 |
} else if caseless_class == 9 /* I */ || caseless_class == 22 /* V */ || caseless_class == 24 |
| 475 |
/* X */ |
| 476 |
{ |
| 477 |
self.ordinal_state = OrdinalState::Roman; |
| 478 |
} else if b == 0xA9 { |
| 479 |
self.ordinal_state = OrdinalState::Copyright; |
| 480 |
} else { |
| 481 |
self.ordinal_state = OrdinalState::Other; |
| 482 |
} |
| 483 |
} |
| 484 |
OrdinalState::OrdinalExpectingSpace => { |
| 485 |
if caseless_class == 0 { |
| 486 |
score += ORDINAL_BONUS; |
| 487 |
self.ordinal_state = OrdinalState::Space; |
| 488 |
} else { |
| 489 |
self.ordinal_state = OrdinalState::Other; |
| 490 |
} |
| 491 |
} |
| 492 |
OrdinalState::OrdinalExpectingSpaceUndoImplausibility => { |
| 493 |
if caseless_class == 0 { |
| 494 |
score += ORDINAL_BONUS - IMPLAUSIBILITY_PENALTY; |
| 495 |
self.ordinal_state = OrdinalState::Space; |
| 496 |
} else { |
| 497 |
self.ordinal_state = OrdinalState::Other; |
| 498 |
} |
| 499 |
} |
| 500 |
OrdinalState::OrdinalExpectingSpaceOrDigit => { |
| 501 |
if caseless_class == 0 { |
| 502 |
score += ORDINAL_BONUS; |
| 503 |
self.ordinal_state = OrdinalState::Space; |
| 504 |
} else if caseless_class == (ASCII_DIGIT as u8) { |
| 505 |
score += ORDINAL_BONUS; |
| 506 |
// Deliberately set to `Other` |
| 507 |
self.ordinal_state = OrdinalState::Other; |
| 508 |
} else { |
| 509 |
self.ordinal_state = OrdinalState::Other; |
| 510 |
} |
| 511 |
} |
| 512 |
OrdinalState::OrdinalExpectingSpaceOrDigitUndoImplausibily => { |
| 513 |
if caseless_class == 0 { |
| 514 |
score += ORDINAL_BONUS - IMPLAUSIBILITY_PENALTY; |
| 515 |
self.ordinal_state = OrdinalState::Space; |
| 516 |
} else if caseless_class == (ASCII_DIGIT as u8) { |
| 517 |
score += ORDINAL_BONUS - IMPLAUSIBILITY_PENALTY; |
| 518 |
// Deliberately set to `Other` |
| 519 |
self.ordinal_state = OrdinalState::Other; |
| 520 |
} else { |
| 521 |
self.ordinal_state = OrdinalState::Other; |
| 522 |
} |
| 523 |
} |
| 524 |
OrdinalState::UpperN => { |
| 525 |
if b == 0xAA { |
| 526 |
self.ordinal_state = |
| 527 |
OrdinalState::OrdinalExpectingSpaceUndoImplausibility; |
| 528 |
} else if b == 0xBA { |
| 529 |
self.ordinal_state = |
| 530 |
OrdinalState::OrdinalExpectingSpaceOrDigitUndoImplausibily; |
| 531 |
} else if b == b'.' { |
| 532 |
self.ordinal_state = OrdinalState::PeriodAfterN; |
| 533 |
} else if caseless_class == 0 { |
| 534 |
self.ordinal_state = OrdinalState::Space; |
| 535 |
} else { |
| 536 |
self.ordinal_state = OrdinalState::Other; |
| 537 |
} |
| 538 |
} |
| 539 |
OrdinalState::LowerN => { |
| 540 |
if b == 0xBA { |
| 541 |
self.ordinal_state = |
| 542 |
OrdinalState::OrdinalExpectingSpaceOrDigitUndoImplausibily; |
| 543 |
} else if b == b'.' { |
| 544 |
self.ordinal_state = OrdinalState::PeriodAfterN; |
| 545 |
} else if caseless_class == 0 { |
| 546 |
self.ordinal_state = OrdinalState::Space; |
| 547 |
} else { |
| 548 |
self.ordinal_state = OrdinalState::Other; |
| 549 |
} |
| 550 |
} |
| 551 |
OrdinalState::FeminineAbbreviationStartLetter => { |
| 552 |
if b == 0xAA { |
| 553 |
self.ordinal_state = |
| 554 |
OrdinalState::OrdinalExpectingSpaceUndoImplausibility; |
| 555 |
} else if caseless_class == 0 { |
| 556 |
self.ordinal_state = OrdinalState::Space; |
| 557 |
} else { |
| 558 |
self.ordinal_state = OrdinalState::Other; |
| 559 |
} |
| 560 |
} |
| 561 |
OrdinalState::Digit => { |
| 562 |
if b == 0xAA || b == 0xBA { |
| 563 |
self.ordinal_state = OrdinalState::OrdinalExpectingSpace; |
| 564 |
} else if caseless_class == 0 { |
| 565 |
self.ordinal_state = OrdinalState::Space; |
| 566 |
} else if caseless_class == (ASCII_DIGIT as u8) { |
| 567 |
// pass |
| 568 |
} else { |
| 569 |
self.ordinal_state = OrdinalState::Other; |
| 570 |
} |
| 571 |
} |
| 572 |
OrdinalState::Roman => { |
| 573 |
if b == 0xAA || b == 0xBA { |
| 574 |
self.ordinal_state = |
| 575 |
OrdinalState::OrdinalExpectingSpaceUndoImplausibility; |
| 576 |
} else if caseless_class == 0 { |
| 577 |
self.ordinal_state = OrdinalState::Space; |
| 578 |
} else if caseless_class == 9 /* I */ || caseless_class == 22 /* V */ || caseless_class == 24 |
| 579 |
/* X */ |
| 580 |
{ |
| 581 |
// pass |
| 582 |
} else { |
| 583 |
self.ordinal_state = OrdinalState::Other; |
| 584 |
} |
| 585 |
} |
| 586 |
OrdinalState::PeriodAfterN => { |
| 587 |
if b == 0xBA { |
| 588 |
self.ordinal_state = OrdinalState::OrdinalExpectingSpaceOrDigit; |
| 589 |
} else if caseless_class == 0 { |
| 590 |
self.ordinal_state = OrdinalState::Space; |
| 591 |
} else { |
| 592 |
self.ordinal_state = OrdinalState::Other; |
| 593 |
} |
| 594 |
} |
| 595 |
OrdinalState::Copyright => { |
| 596 |
if caseless_class == 0 { |
| 597 |
score += COPYRIGHT_BONUS; |
| 598 |
self.ordinal_state = OrdinalState::Space; |
| 599 |
} else { |
| 600 |
self.ordinal_state = OrdinalState::Other; |
| 601 |
} |
| 602 |
} |
| 603 |
} |
| 604 |
} |
| 605 |
|
| 606 |
if ascii { |
| 607 |
self.prev_non_ascii = 0; |
| 608 |
} else { |
| 609 |
self.prev_non_ascii += 1; |
| 610 |
} |
| 611 |
self.prev = caseless_class; |
| 612 |
} |
| 613 |
Some(score) |
| 614 |
} |
| 615 |
} |
| 616 |
|
| 617 |
struct ArabicFrenchCandidate { |
| 618 |
data: &'static SingleByteData, |
| 619 |
prev: u8, |
| 620 |
case_state: LatinCaseState, |
| 621 |
prev_ascii: bool, |
| 622 |
current_word_len: u64, |
| 623 |
longest_word: u64, |
| 624 |
} |
| 625 |
|
| 626 |
impl ArabicFrenchCandidate { |
| 627 |
fn new(data: &'static SingleByteData) -> Self { |
| 628 |
ArabicFrenchCandidate { |
| 629 |
data: data, |
| 630 |
prev: 0, |
| 631 |
case_state: LatinCaseState::Space, |
| 632 |
prev_ascii: true, |
| 633 |
current_word_len: 0, |
| 634 |
longest_word: 0, |
| 635 |
} |
| 636 |
} |
| 637 |
|
| 638 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 639 |
let mut score = 0i64; |
| 640 |
for &b in buffer { |
| 641 |
let class = self.data.classify(b); |
| 642 |
if class == 255 { |
| 643 |
return None; |
| 644 |
} |
| 645 |
let caseless_class = class & 0x7F; |
| 646 |
|
| 647 |
let ascii = b < 0x80; |
| 648 |
let ascii_pair = self.prev_ascii && ascii; |
| 649 |
|
| 650 |
if caseless_class != LATIN_LETTER { |
| 651 |
// We compute case penalties for French only |
| 652 |
self.case_state = LatinCaseState::Space; |
| 653 |
} else if (class >> 7) == 0 { |
| 654 |
if self.case_state == LatinCaseState::AllCaps && !ascii_pair { |
| 655 |
score += IMPLAUSIBLE_LATIN_CASE_TRANSITION_PENALTY; |
| 656 |
} |
| 657 |
self.case_state = LatinCaseState::Lower; |
| 658 |
} else { |
| 659 |
match self.case_state { |
| 660 |
LatinCaseState::Space => { |
| 661 |
self.case_state = LatinCaseState::Upper; |
| 662 |
} |
| 663 |
LatinCaseState::Upper | LatinCaseState::AllCaps => { |
| 664 |
self.case_state = LatinCaseState::AllCaps; |
| 665 |
} |
| 666 |
LatinCaseState::Lower => { |
| 667 |
if !ascii_pair { |
| 668 |
score += IMPLAUSIBLE_LATIN_CASE_TRANSITION_PENALTY; |
| 669 |
} |
| 670 |
self.case_state = LatinCaseState::Upper; |
| 671 |
} |
| 672 |
} |
| 673 |
} |
| 674 |
|
| 675 |
// Count only Arabic word length and ignore French |
| 676 |
let non_ascii_alphabetic = self.data.is_non_latin_alphabetic(caseless_class, true); |
| 677 |
// XXX apply penalty if > 23 |
| 678 |
if non_ascii_alphabetic { |
| 679 |
self.current_word_len += 1; |
| 680 |
} else { |
| 681 |
if self.current_word_len > self.longest_word { |
| 682 |
self.longest_word = self.current_word_len; |
| 683 |
} |
| 684 |
self.current_word_len = 0; |
| 685 |
} |
| 686 |
|
| 687 |
if !ascii_pair { |
| 688 |
score += self.data.score(caseless_class, self.prev, true); |
| 689 |
|
| 690 |
if self.prev == LATIN_LETTER && non_ascii_alphabetic { |
| 691 |
score += LATIN_ADJACENCY_PENALTY; |
| 692 |
} else if caseless_class == LATIN_LETTER |
| 693 |
&& self.data.is_non_latin_alphabetic(self.prev, true) |
| 694 |
{ |
| 695 |
score += LATIN_ADJACENCY_PENALTY; |
| 696 |
} |
| 697 |
} |
| 698 |
|
| 699 |
self.prev_ascii = ascii; |
| 700 |
self.prev = caseless_class; |
| 701 |
} |
| 702 |
Some(score) |
| 703 |
} |
| 704 |
} |
| 705 |
|
| 706 |
struct CaselessCandidate { |
| 707 |
data: &'static SingleByteData, |
| 708 |
prev: u8, |
| 709 |
prev_ascii: bool, |
| 710 |
current_word_len: u64, |
| 711 |
longest_word: u64, |
| 712 |
} |
| 713 |
|
| 714 |
impl CaselessCandidate { |
| 715 |
fn new(data: &'static SingleByteData) -> Self { |
| 716 |
CaselessCandidate { |
| 717 |
data: data, |
| 718 |
prev: 0, |
| 719 |
prev_ascii: true, |
| 720 |
current_word_len: 0, |
| 721 |
longest_word: 0, |
| 722 |
} |
| 723 |
} |
| 724 |
|
| 725 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 726 |
let mut score = 0i64; |
| 727 |
for &b in buffer { |
| 728 |
let class = self.data.classify(b); |
| 729 |
if class == 255 { |
| 730 |
return None; |
| 731 |
} |
| 732 |
let caseless_class = class & 0x7F; |
| 733 |
|
| 734 |
let ascii = b < 0x80; |
| 735 |
let ascii_pair = self.prev_ascii && ascii; |
| 736 |
|
| 737 |
let non_ascii_alphabetic = self.data.is_non_latin_alphabetic(caseless_class, false); |
| 738 |
// Apply penalty if > 23 and not Thai |
| 739 |
if non_ascii_alphabetic { |
| 740 |
self.current_word_len += 1; |
| 741 |
} else { |
| 742 |
if self.current_word_len > self.longest_word { |
| 743 |
self.longest_word = self.current_word_len; |
| 744 |
} |
| 745 |
self.current_word_len = 0; |
| 746 |
} |
| 747 |
|
| 748 |
if !ascii_pair { |
| 749 |
score += self.data.score(caseless_class, self.prev, false); |
| 750 |
|
| 751 |
if self.prev == LATIN_LETTER && non_ascii_alphabetic { |
| 752 |
score += LATIN_ADJACENCY_PENALTY; |
| 753 |
} else if caseless_class == LATIN_LETTER |
| 754 |
&& self.data.is_non_latin_alphabetic(self.prev, false) |
| 755 |
{ |
| 756 |
score += LATIN_ADJACENCY_PENALTY; |
| 757 |
} |
| 758 |
} |
| 759 |
|
| 760 |
self.prev_ascii = ascii; |
| 761 |
self.prev = caseless_class; |
| 762 |
} |
| 763 |
Some(score) |
| 764 |
} |
| 765 |
} |
| 766 |
|
| 767 |
fn is_ascii_punctuation(byte: u8) -> bool { |
| 768 |
match byte { |
| 769 |
b'.' | b',' | b':' | b';' | b'?' | b'!' => true, |
| 770 |
_ => false, |
| 771 |
} |
| 772 |
} |
| 773 |
|
| 774 |
struct LogicalCandidate { |
| 775 |
data: &'static SingleByteData, |
| 776 |
prev: u8, |
| 777 |
prev_ascii: bool, |
| 778 |
plausible_punctuation: u64, |
| 779 |
current_word_len: u64, |
| 780 |
longest_word: u64, |
| 781 |
} |
| 782 |
|
| 783 |
impl LogicalCandidate { |
| 784 |
fn new(data: &'static SingleByteData) -> Self { |
| 785 |
LogicalCandidate { |
| 786 |
data: data, |
| 787 |
prev: 0, |
| 788 |
prev_ascii: true, |
| 789 |
plausible_punctuation: 0, |
| 790 |
current_word_len: 0, |
| 791 |
longest_word: 0, |
| 792 |
} |
| 793 |
} |
| 794 |
|
| 795 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 796 |
let mut score = 0i64; |
| 797 |
for &b in buffer { |
| 798 |
let class = self.data.classify(b); |
| 799 |
if class == 255 { |
| 800 |
return None; |
| 801 |
} |
| 802 |
let caseless_class = class & 0x7F; |
| 803 |
|
| 804 |
let ascii = b < 0x80; |
| 805 |
let ascii_pair = self.prev_ascii && ascii; |
| 806 |
|
| 807 |
let non_ascii_alphabetic = self.data.is_non_latin_alphabetic(caseless_class, false); |
| 808 |
// XXX apply penalty if > 22 |
| 809 |
if non_ascii_alphabetic { |
| 810 |
self.current_word_len += 1; |
| 811 |
} else { |
| 812 |
if self.current_word_len > self.longest_word { |
| 813 |
self.longest_word = self.current_word_len; |
| 814 |
} |
| 815 |
self.current_word_len = 0; |
| 816 |
} |
| 817 |
|
| 818 |
if !ascii_pair { |
| 819 |
score += self.data.score(caseless_class, self.prev, false); |
| 820 |
|
| 821 |
let prev_non_ascii_alphabetic = self.data.is_non_latin_alphabetic(self.prev, false); |
| 822 |
if caseless_class == 0 && prev_non_ascii_alphabetic && is_ascii_punctuation(b) { |
| 823 |
self.plausible_punctuation += 1; |
| 824 |
} |
| 825 |
|
| 826 |
if self.prev == LATIN_LETTER && non_ascii_alphabetic { |
| 827 |
score += LATIN_ADJACENCY_PENALTY; |
| 828 |
} else if caseless_class == LATIN_LETTER && prev_non_ascii_alphabetic { |
| 829 |
score += LATIN_ADJACENCY_PENALTY; |
| 830 |
} |
| 831 |
} |
| 832 |
|
| 833 |
self.prev_ascii = ascii; |
| 834 |
self.prev = caseless_class; |
| 835 |
} |
| 836 |
Some(score) |
| 837 |
} |
| 838 |
} |
| 839 |
|
| 840 |
struct VisualCandidate { |
| 841 |
data: &'static SingleByteData, |
| 842 |
prev: u8, |
| 843 |
prev_ascii: bool, |
| 844 |
prev_punctuation: bool, |
| 845 |
plausible_punctuation: u64, |
| 846 |
current_word_len: u64, |
| 847 |
longest_word: u64, |
| 848 |
} |
| 849 |
|
| 850 |
impl VisualCandidate { |
| 851 |
fn new(data: &'static SingleByteData) -> Self { |
| 852 |
VisualCandidate { |
| 853 |
data: data, |
| 854 |
prev: 0, |
| 855 |
prev_ascii: true, |
| 856 |
prev_punctuation: false, |
| 857 |
plausible_punctuation: 0, |
| 858 |
current_word_len: 0, |
| 859 |
longest_word: 0, |
| 860 |
} |
| 861 |
} |
| 862 |
|
| 863 |
fn feed(&mut self, buffer: &[u8]) -> Option<i64> { |
| 864 |
let mut score = 0i64; |
| 865 |
for &b in buffer { |
| 866 |
let class = self.data.classify(b); |
| 867 |
if class == 255 { |
| 868 |
return None; |
| 869 |
} |
| 870 |
let caseless_class = class & 0x7F; |
| 871 |
|
| 872 |
let ascii = b < 0x80; |
| 873 |
let ascii_pair = self.prev_ascii && ascii; |
| 874 |
|
| 875 |
let non_ascii_alphabetic = self.data.is_non_latin_alphabetic(caseless_class, false); |
| 876 |
// XXX apply penalty if > 22 |
| 877 |
if non_ascii_alphabetic { |
| 878 |
self.current_word_len += 1; |
| 879 |
} else { |
| 880 |
if self.current_word_len > self.longest_word { |
| 881 |
self.longest_word = self.current_word_len; |
| 882 |
} |
| 883 |
self.current_word_len = 0; |
| 884 |
} |
| 885 |
|
| 886 |
if !ascii_pair { |
| 887 |
score += self.data.score(caseless_class, self.prev, false); |
| 888 |
|
| 889 |
if non_ascii_alphabetic && self.prev_punctuation { |
| 890 |
self.plausible_punctuation += 1; |
| 891 |
} |
| 892 |
|
| 893 |
if self.prev == LATIN_LETTER && non_ascii_alphabetic { |
| 894 |
score += LATIN_ADJACENCY_PENALTY; |
| 895 |
} else if caseless_class == LATIN_LETTER |
| 896 |
&& self.data.is_non_latin_alphabetic(self.prev, false) |
| 897 |
{ |
| 898 |
score += LATIN_ADJACENCY_PENALTY; |
| 899 |
} |
| 900 |
} |
| 901 |
|
| 902 |
self.prev_ascii = ascii; |
| 903 |
self.prev = caseless_class; |
| 904 |
self.prev_punctuation = caseless_class == 0 && is_ascii_punctuation(b); |
| 905 |
} |
| 906 |
Some(score) |
| 907 |
} |
| 908 |
} |
| 909 |
|
| 910 |
struct Utf8Candidate { |
| 911 |
decoder: Decoder, |
| 912 |
} |
| 913 |
|
| 914 |
impl Utf8Candidate { |
| 915 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 916 |
let mut dst = [0u8; 1024]; |
| 917 |
let mut total_read = 0; |
| 918 |
loop { |
| 919 |
let (result, read, _) = self.decoder.decode_to_utf8_without_replacement( |
| 920 |
&buffer[total_read..], |
| 921 |
&mut dst, |
| 922 |
last, |
| 923 |
); |
| 924 |
total_read += read; |
| 925 |
match result { |
| 926 |
DecoderResult::InputEmpty => { |
| 927 |
return Some(0); |
| 928 |
} |
| 929 |
DecoderResult::Malformed(_, _) => { |
| 930 |
return None; |
| 931 |
} |
| 932 |
DecoderResult::OutputFull => { |
| 933 |
continue; |
| 934 |
} |
| 935 |
} |
| 936 |
} |
| 937 |
} |
| 938 |
} |
| 939 |
|
| 940 |
struct Iso2022Candidate { |
| 941 |
decoder: Decoder, |
| 942 |
} |
| 943 |
|
| 944 |
impl Iso2022Candidate { |
| 945 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 946 |
let mut dst = [0u16; 1024]; |
| 947 |
let mut total_read = 0; |
| 948 |
loop { |
| 949 |
let (result, read, _) = self.decoder.decode_to_utf16_without_replacement( |
| 950 |
&buffer[total_read..], |
| 951 |
&mut dst, |
| 952 |
last, |
| 953 |
); |
| 954 |
total_read += read; |
| 955 |
match result { |
| 956 |
DecoderResult::InputEmpty => { |
| 957 |
return Some(0); |
| 958 |
} |
| 959 |
DecoderResult::Malformed(_, _) => { |
| 960 |
return None; |
| 961 |
} |
| 962 |
DecoderResult::OutputFull => { |
| 963 |
continue; |
| 964 |
} |
| 965 |
} |
| 966 |
} |
| 967 |
} |
| 968 |
} |
| 969 |
|
| 970 |
#[derive(PartialEq)] |
| 971 |
enum LatinCj { |
| 972 |
AsciiLetter, |
| 973 |
Cj, |
| 974 |
Other, |
| 975 |
} |
| 976 |
|
| 977 |
#[derive(PartialEq, Copy, Clone)] |
| 978 |
enum HalfWidthKatakana { |
| 979 |
DakutenForbidden, |
| 980 |
DakutenAllowed, |
| 981 |
DakutenOrHandakutenAllowed, |
| 982 |
} |
| 983 |
|
| 984 |
#[derive(PartialEq)] |
| 985 |
enum LatinKorean { |
| 986 |
AsciiLetter, |
| 987 |
Hangul, |
| 988 |
Hanja, |
| 989 |
Other, |
| 990 |
} |
| 991 |
|
| 992 |
fn cjk_extra_score(u: u16, table: &'static [u16; 128]) -> i64 { |
| 993 |
if let Some(pos) = table.iter().position(|&x| x == u) { |
| 994 |
((128 - pos) / 16) as i64 |
| 995 |
} else { |
| 996 |
0 |
| 997 |
} |
| 998 |
} |
| 999 |
|
| 1000 |
struct GbkCandidate { |
| 1001 |
decoder: Decoder, |
| 1002 |
prev_byte: u8, |
| 1003 |
prev: LatinCj, |
| 1004 |
pending_score: Option<i64>, |
| 1005 |
} |
| 1006 |
|
| 1007 |
impl GbkCandidate { |
| 1008 |
fn maybe_set_as_pending(&mut self, s: i64) -> i64 { |
| 1009 |
assert!(self.pending_score.is_none()); |
| 1010 |
if self.prev == LatinCj::Cj || !more_problematic_lead(self.prev_byte) { |
| 1011 |
s |
| 1012 |
} else { |
| 1013 |
self.pending_score = Some(s); |
| 1014 |
0 |
| 1015 |
} |
| 1016 |
} |
| 1017 |
|
| 1018 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1019 |
let mut score = 0i64; |
| 1020 |
let mut src = [0u8]; |
| 1021 |
let mut dst = [0u16; 2]; |
| 1022 |
for &b in buffer { |
| 1023 |
src[0] = b; |
| 1024 |
let (result, read, written) = self |
| 1025 |
.decoder |
| 1026 |
.decode_to_utf16_without_replacement(&src, &mut dst, false); |
| 1027 |
if written == 1 { |
| 1028 |
let u = dst[0]; |
| 1029 |
if (u >= u16::from(b'a') && u <= u16::from(b'z')) |
| 1030 |
|| (u >= u16::from(b'A') && u <= u16::from(b'Z')) |
| 1031 |
{ |
| 1032 |
self.pending_score = None; // Discard pending score |
| 1033 |
if self.prev == LatinCj::Cj { |
| 1034 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1035 |
} |
| 1036 |
self.prev = LatinCj::AsciiLetter; |
| 1037 |
} else if u == 0x20AC { |
| 1038 |
// euro sign |
| 1039 |
self.pending_score = None; // Discard pending score |
| 1040 |
// Should there even be a penalty? |
| 1041 |
self.prev = LatinCj::Other; |
| 1042 |
} else if u >= 0x4E00 && u <= 0x9FA5 { |
| 1043 |
if let Some(pending) = self.pending_score { |
| 1044 |
score += pending; |
| 1045 |
self.pending_score = None; |
| 1046 |
} |
| 1047 |
if b >= 0xA1 && b <= 0xFE { |
| 1048 |
match self.prev_byte { |
| 1049 |
0xA1..=0xD7 => { |
| 1050 |
score += GBK_SCORE_PER_LEVEL_1; |
| 1051 |
score += |
| 1052 |
cjk_extra_score(u, &data::DETECTOR_DATA.frequent_simplified); |
| 1053 |
} |
| 1054 |
0xD8..=0xFE => score += GBK_SCORE_PER_LEVEL_2, |
| 1055 |
_ => { |
| 1056 |
score += GBK_SCORE_PER_NON_EUC; |
| 1057 |
} |
| 1058 |
} |
| 1059 |
} else { |
| 1060 |
score += self.maybe_set_as_pending(GBK_SCORE_PER_NON_EUC); |
| 1061 |
} |
| 1062 |
if self.prev == LatinCj::AsciiLetter { |
| 1063 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1064 |
} |
| 1065 |
self.prev = LatinCj::Cj; |
| 1066 |
} else if (u >= 0x3400 && u < 0xA000) || (u >= 0xF900 && u < 0xFB00) { |
| 1067 |
if let Some(pending) = self.pending_score { |
| 1068 |
score += pending; |
| 1069 |
self.pending_score = None; |
| 1070 |
} |
| 1071 |
// XXX score? |
| 1072 |
if self.prev == LatinCj::AsciiLetter { |
| 1073 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1074 |
} |
| 1075 |
self.prev = LatinCj::Cj; |
| 1076 |
} else if u >= 0xE000 && u < 0xF900 { |
| 1077 |
if let Some(pending) = self.pending_score { |
| 1078 |
score += pending; |
| 1079 |
self.pending_score = None; |
| 1080 |
} |
| 1081 |
// Treat the GB18030-required PUA mappings as non-EUC ideographs. |
| 1082 |
match u { |
| 1083 |
0xE78D..=0xE796 |
| 1084 |
| 0xE816..=0xE818 |
| 1085 |
| 0xE81E |
| 1086 |
| 0xE826 |
| 1087 |
| 0xE82B |
| 1088 |
| 0xE82C |
| 1089 |
| 0xE831 |
| 1090 |
| 0xE832 |
| 1091 |
| 0xE83B |
| 1092 |
| 0xE843 |
| 1093 |
| 0xE854 |
| 1094 |
| 0xE855 |
| 1095 |
| 0xE864 => { |
| 1096 |
score += GBK_SCORE_PER_NON_EUC; |
| 1097 |
if self.prev == LatinCj::AsciiLetter { |
| 1098 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1099 |
} |
| 1100 |
self.prev = LatinCj::Cj; |
| 1101 |
} |
| 1102 |
_ => { |
| 1103 |
score += GBK_PUA_PENALTY; |
| 1104 |
self.prev = LatinCj::Other; |
| 1105 |
} |
| 1106 |
} |
| 1107 |
} else { |
| 1108 |
match u { |
| 1109 |
0x3000 // Distinct from Korean, space |
| 1110 |
| 0x3001 // Distinct from Korean, enumeration comma |
| 1111 |
| 0x3002 // Distinct from Korean, full stop |
| 1112 |
| 0xFF08 // Distinct from Korean, parenthesis |
| 1113 |
| 0xFF09 // Distinct from Korean, parenthesis |
| 1114 |
| 0xFF01 // Distinct from Japanese, exclamation |
| 1115 |
| 0xFF0C // Distinct from Japanese, comma |
| 1116 |
| 0xFF1B // Distinct from Japanese, semicolon |
| 1117 |
| 0xFF1F // Distinct from Japanese, question |
| 1118 |
=> { |
| 1119 |
if let Some(pending) = self.pending_score { |
| 1120 |
score += pending; |
| 1121 |
self.pending_score = None; |
| 1122 |
} |
| 1123 |
score += CJ_PUNCTUATION; |
| 1124 |
} |
| 1125 |
0..=0x7F => { |
| 1126 |
self.pending_score = None; // Discard pending score |
| 1127 |
} |
| 1128 |
_ => { |
| 1129 |
if let Some(pending) = self.pending_score { |
| 1130 |
score += pending; |
| 1131 |
self.pending_score = None; |
| 1132 |
} |
| 1133 |
score += CJK_OTHER; |
| 1134 |
} |
| 1135 |
} |
| 1136 |
self.prev = LatinCj::Other; |
| 1137 |
} |
| 1138 |
} else if written == 2 { |
| 1139 |
if let Some(pending) = self.pending_score { |
| 1140 |
score += pending; |
| 1141 |
self.pending_score = None; |
| 1142 |
} |
| 1143 |
let u = dst[0]; |
| 1144 |
if u >= 0xDB80 && u <= 0xDBFF { |
| 1145 |
score += GBK_PUA_PENALTY; |
| 1146 |
self.prev = LatinCj::Other; |
| 1147 |
} else if u >= 0xD480 && u < 0xD880 { |
| 1148 |
score += GBK_SCORE_PER_NON_EUC; |
| 1149 |
if self.prev == LatinCj::AsciiLetter { |
| 1150 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1151 |
} |
| 1152 |
self.prev = LatinCj::Cj; |
| 1153 |
} else { |
| 1154 |
score += CJK_OTHER; |
| 1155 |
self.prev = LatinCj::Other; |
| 1156 |
} |
| 1157 |
} |
| 1158 |
match result { |
| 1159 |
DecoderResult::InputEmpty => { |
| 1160 |
assert_eq!(read, 1); |
| 1161 |
} |
| 1162 |
DecoderResult::Malformed(malformed_len, _) => { |
| 1163 |
if (self.prev_byte == 0xA0 || self.prev_byte == 0xFE || self.prev_byte == 0xFD) |
| 1164 |
&& (b < 0x80 || b == 0xFF) |
| 1165 |
{ |
| 1166 |
// Mac OS Chinese Simplified single-byte that conflicts with code page GBK lead byte |
| 1167 |
// followed by ASCII or a non-conflicting single-byte extension. |
| 1168 |
self.pending_score = None; // Just in case |
| 1169 |
score += GBK_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1170 |
if (b >= b'a' && b <= b'z') || (b >= b'A' && b <= b'Z') { |
| 1171 |
self.prev = LatinCj::AsciiLetter; |
| 1172 |
} else if b == 0xFF { |
| 1173 |
score += GBK_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1174 |
self.prev = LatinCj::Other; |
| 1175 |
} else { |
| 1176 |
self.prev = LatinCj::Other; |
| 1177 |
} |
| 1178 |
// The GBK decoder has the pending ASCII concept, which is |
| 1179 |
// a problem with this trickery, so let's reset the state. |
| 1180 |
self.decoder = GBK.new_decoder_without_bom_handling(); |
| 1181 |
} else if malformed_len == 1 && b == 0xFF { |
| 1182 |
// Mac OS Chinese Simplified single-byte extension that doesn't conflict with lead bytes |
| 1183 |
self.pending_score = None; // Just in case |
| 1184 |
score += GBK_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1185 |
self.prev = LatinCj::Other; |
| 1186 |
// The GBK decoder has the pending ASCII concept, which is |
| 1187 |
// a problem with this trickery, so let's reset the state. |
| 1188 |
self.decoder = GBK.new_decoder_without_bom_handling(); |
| 1189 |
} else { |
| 1190 |
return None; |
| 1191 |
} |
| 1192 |
} |
| 1193 |
DecoderResult::OutputFull => { |
| 1194 |
unreachable!(); |
| 1195 |
} |
| 1196 |
} |
| 1197 |
self.prev_byte = b; |
| 1198 |
} |
| 1199 |
if last { |
| 1200 |
let (result, _, _) = self |
| 1201 |
.decoder |
| 1202 |
.decode_to_utf16_without_replacement(b"", &mut dst, true); |
| 1203 |
match result { |
| 1204 |
DecoderResult::InputEmpty => {} |
| 1205 |
DecoderResult::Malformed(_, _) => { |
| 1206 |
return None; |
| 1207 |
} |
| 1208 |
DecoderResult::OutputFull => { |
| 1209 |
unreachable!(); |
| 1210 |
} |
| 1211 |
} |
| 1212 |
} |
| 1213 |
Some(score) |
| 1214 |
} |
| 1215 |
} |
| 1216 |
|
| 1217 |
// Shift_JIS and Big5 |
| 1218 |
fn problematic_lead(b: u8) -> bool { |
| 1219 |
match b { |
| 1220 |
0x91..=0x97 | 0x9A | 0x8A | 0x9B | 0x8B | 0x9E | 0x8E | 0xB0 => true, |
| 1221 |
_ => false, |
| 1222 |
} |
| 1223 |
} |
| 1224 |
|
| 1225 |
// GBK and EUC-KR |
| 1226 |
fn more_problematic_lead(b: u8) -> bool { |
| 1227 |
problematic_lead(b) || b == 0x82 || b == 0x84 || b == 0x85 || b == 0xA0 |
| 1228 |
} |
| 1229 |
|
| 1230 |
struct ShiftJisCandidate { |
| 1231 |
decoder: Decoder, |
| 1232 |
half_width_katakana_seen: bool, |
| 1233 |
half_width_katakana_state: HalfWidthKatakana, |
| 1234 |
prev: LatinCj, |
| 1235 |
prev_byte: u8, |
| 1236 |
pending_score: Option<i64>, |
| 1237 |
} |
| 1238 |
|
| 1239 |
impl ShiftJisCandidate { |
| 1240 |
fn maybe_set_as_pending(&mut self, s: i64) -> i64 { |
| 1241 |
assert!(self.pending_score.is_none()); |
| 1242 |
if self.prev == LatinCj::Cj || !problematic_lead(self.prev_byte) { |
| 1243 |
s |
| 1244 |
} else { |
| 1245 |
self.pending_score = Some(s); |
| 1246 |
0 |
| 1247 |
} |
| 1248 |
} |
| 1249 |
|
| 1250 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1251 |
let mut score = 0i64; |
| 1252 |
let mut src = [0u8]; |
| 1253 |
let mut dst = [0u16; 2]; |
| 1254 |
for &b in buffer { |
| 1255 |
src[0] = b; |
| 1256 |
let (result, read, written) = self |
| 1257 |
.decoder |
| 1258 |
.decode_to_utf16_without_replacement(&src, &mut dst, false); |
| 1259 |
if written > 0 { |
| 1260 |
let half_width_katakana_state = self.half_width_katakana_state; |
| 1261 |
self.half_width_katakana_state = HalfWidthKatakana::DakutenForbidden; |
| 1262 |
let u = dst[0]; |
| 1263 |
if (u >= u16::from(b'a') && u <= u16::from(b'z')) |
| 1264 |
|| (u >= u16::from(b'A') && u <= u16::from(b'Z')) |
| 1265 |
{ |
| 1266 |
self.pending_score = None; // Discard pending score |
| 1267 |
if self.prev == LatinCj::Cj { |
| 1268 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1269 |
} |
| 1270 |
self.prev = LatinCj::AsciiLetter; |
| 1271 |
} else if u >= 0xFF61 && u <= 0xFF9F { |
| 1272 |
if !self.half_width_katakana_seen { |
| 1273 |
self.half_width_katakana_seen = true; |
| 1274 |
// To avoid misdetecting title-length inputs |
| 1275 |
score += SHIFT_JIS_INITIAL_HALF_WIDTH_KATAKANA_PENALTY; |
| 1276 |
} |
| 1277 |
self.pending_score = None; // Discard pending score |
| 1278 |
score += HALF_WIDTH_KATAKANA_SCORE; |
| 1279 |
|
| 1280 |
if (u >= 0xFF76 && u <= 0xFF84) || u == 0xFF73 { |
| 1281 |
self.half_width_katakana_state = HalfWidthKatakana::DakutenAllowed; |
| 1282 |
} else if u >= 0xFF8A && u <= 0xFF8E { |
| 1283 |
self.half_width_katakana_state = |
| 1284 |
HalfWidthKatakana::DakutenOrHandakutenAllowed; |
| 1285 |
} else if u == 0xFF9E { |
| 1286 |
if half_width_katakana_state == HalfWidthKatakana::DakutenForbidden { |
| 1287 |
score += IMPLAUSIBILITY_PENALTY; |
| 1288 |
} else { |
| 1289 |
score += HALF_WIDTH_KATAKANA_VOICING_SCORE; |
| 1290 |
} |
| 1291 |
} else if u == 0xFF9F { |
| 1292 |
if half_width_katakana_state |
| 1293 |
!= HalfWidthKatakana::DakutenOrHandakutenAllowed |
| 1294 |
{ |
| 1295 |
score += IMPLAUSIBILITY_PENALTY; |
| 1296 |
} else { |
| 1297 |
score += HALF_WIDTH_KATAKANA_VOICING_SCORE; |
| 1298 |
} |
| 1299 |
} |
| 1300 |
|
| 1301 |
if self.prev == LatinCj::AsciiLetter { |
| 1302 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1303 |
} |
| 1304 |
self.prev = LatinCj::Cj; |
| 1305 |
} else if u >= 0x3040 && u < 0x3100 { |
| 1306 |
if let Some(pending) = self.pending_score { |
| 1307 |
score += pending; |
| 1308 |
self.pending_score = None; |
| 1309 |
} |
| 1310 |
score += SHIFT_JIS_SCORE_PER_KANA; |
| 1311 |
if self.prev == LatinCj::AsciiLetter { |
| 1312 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1313 |
} |
| 1314 |
self.prev = LatinCj::Cj; |
| 1315 |
} else if (u >= 0x3400 && u < 0xA000) || (u >= 0xF900 && u < 0xFB00) { |
| 1316 |
if let Some(pending) = self.pending_score { |
| 1317 |
score += pending; |
| 1318 |
self.pending_score = None; |
| 1319 |
} |
| 1320 |
if self.prev_byte < 0x98 || (self.prev_byte == 0x98 && b < 0x73) { |
| 1321 |
score += self.maybe_set_as_pending( |
| 1322 |
SHIFT_JIS_SCORE_PER_LEVEL_1_KANJI |
| 1323 |
+ cjk_extra_score(u, &data::DETECTOR_DATA.frequent_kanji), |
| 1324 |
); |
| 1325 |
} else { |
| 1326 |
score += self.maybe_set_as_pending(SHIFT_JIS_SCORE_PER_LEVEL_2_KANJI); |
| 1327 |
} |
| 1328 |
if self.prev == LatinCj::AsciiLetter { |
| 1329 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1330 |
} |
| 1331 |
self.prev = LatinCj::Cj; |
| 1332 |
} else if u >= 0xE000 && u < 0xF900 { |
| 1333 |
if let Some(pending) = self.pending_score { |
| 1334 |
score += pending; |
| 1335 |
self.pending_score = None; |
| 1336 |
} |
| 1337 |
score += SHIFT_JIS_PUA_PENALTY; |
| 1338 |
self.prev = LatinCj::Other; |
| 1339 |
} else { |
| 1340 |
match u { |
| 1341 |
0x3000 // Distinct from Korean, space |
| 1342 |
| 0x3001 // Distinct from Korean, enumeration comma |
| 1343 |
| 0x3002 // Distinct from Korean, full stop |
| 1344 |
| 0xFF08 // Distinct from Korean, parenthesis |
| 1345 |
| 0xFF09 // Distinct from Korean, parenthesis |
| 1346 |
=> { |
| 1347 |
if let Some(pending) = self.pending_score { |
| 1348 |
score += pending; |
| 1349 |
self.pending_score = None; |
| 1350 |
} |
| 1351 |
// Not really needed for CJK distinction |
| 1352 |
// but let's give non-zero score for these |
| 1353 |
// common byte pairs anyway. |
| 1354 |
score += CJ_PUNCTUATION; |
| 1355 |
} |
| 1356 |
0..=0x7F => { |
| 1357 |
self.pending_score = None; // Discard pending score |
| 1358 |
} |
| 1359 |
0x80 => { |
| 1360 |
// This is a control character that overlaps euro |
| 1361 |
// in windows-1252 and happens to be a non-error |
| 1362 |
// is Shift_JIS. |
| 1363 |
self.pending_score = None; // Discard pending score |
| 1364 |
score += IMPLAUSIBILITY_PENALTY; |
| 1365 |
} |
| 1366 |
_ => { |
| 1367 |
if let Some(pending) = self.pending_score { |
| 1368 |
score += pending; |
| 1369 |
self.pending_score = None; |
| 1370 |
} |
| 1371 |
score += CJK_OTHER; |
| 1372 |
} |
| 1373 |
} |
| 1374 |
self.prev = LatinCj::Other; |
| 1375 |
} |
| 1376 |
} |
| 1377 |
match result { |
| 1378 |
DecoderResult::InputEmpty => { |
| 1379 |
assert_eq!(read, 1); |
| 1380 |
} |
| 1381 |
DecoderResult::Malformed(malformed_len, _) => { |
| 1382 |
if (((self.prev_byte >= 0x81 && self.prev_byte <= 0x9F) |
| 1383 |
|| (self.prev_byte >= 0xE0 && self.prev_byte <= 0xFC)) |
| 1384 |
&& ((b >= 0x40 && b <= 0x7E) || (b >= 0x80 && b <= 0xFC))) |
| 1385 |
&& !((self.prev_byte == 0x82 && b >= 0xFA) |
| 1386 |
|| (self.prev_byte == 0x84 && ((b >= 0xDD && b <= 0xE4) || b >= 0xFB)) |
| 1387 |
|| (self.prev_byte == 0x86 && b >= 0xF2 && b <= 0xFA) |
| 1388 |
|| (self.prev_byte == 0x87 && b >= 0x77 && b <= 0x7D) |
| 1389 |
|| (self.prev_byte == 0xFC && b >= 0xF5)) |
| 1390 |
{ |
| 1391 |
// Shift_JIS2004 or MacJapanese |
| 1392 |
if let Some(pending) = self.pending_score { |
| 1393 |
score += pending; |
| 1394 |
self.pending_score = None; |
| 1395 |
} |
| 1396 |
score += SHIFT_JIS_EXTENSION_PENALTY; |
| 1397 |
// Approximate boundary |
| 1398 |
if self.prev_byte < 0x87 { |
| 1399 |
self.prev = LatinCj::Other; |
| 1400 |
} else { |
| 1401 |
if self.prev == LatinCj::AsciiLetter { |
| 1402 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1403 |
} |
| 1404 |
self.prev = LatinCj::Cj; |
| 1405 |
} |
| 1406 |
} else if malformed_len == 1 && (b == 0xA0 || b >= 0xFD) { |
| 1407 |
self.pending_score = None; // Just in case |
| 1408 |
score += SHIFT_JIS_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1409 |
self.prev = LatinCj::Other; |
| 1410 |
} else { |
| 1411 |
return None; |
| 1412 |
} |
| 1413 |
} |
| 1414 |
DecoderResult::OutputFull => { |
| 1415 |
unreachable!(); |
| 1416 |
} |
| 1417 |
} |
| 1418 |
self.prev_byte = b; |
| 1419 |
} |
| 1420 |
if last { |
| 1421 |
let (result, _, _) = self |
| 1422 |
.decoder |
| 1423 |
.decode_to_utf16_without_replacement(b"", &mut dst, true); |
| 1424 |
match result { |
| 1425 |
DecoderResult::InputEmpty => {} |
| 1426 |
DecoderResult::Malformed(_, _) => { |
| 1427 |
return None; |
| 1428 |
} |
| 1429 |
DecoderResult::OutputFull => { |
| 1430 |
unreachable!(); |
| 1431 |
} |
| 1432 |
} |
| 1433 |
} |
| 1434 |
Some(score) |
| 1435 |
} |
| 1436 |
} |
| 1437 |
|
| 1438 |
struct EucJpCandidate { |
| 1439 |
decoder: Decoder, |
| 1440 |
non_ascii_seen: bool, |
| 1441 |
half_width_katakana_state: HalfWidthKatakana, |
| 1442 |
prev: LatinCj, |
| 1443 |
prev_byte: u8, |
| 1444 |
prev_prev_byte: u8, |
| 1445 |
} |
| 1446 |
|
| 1447 |
impl EucJpCandidate { |
| 1448 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1449 |
let mut score = 0i64; |
| 1450 |
let mut src = [0u8]; |
| 1451 |
let mut dst = [0u16; 2]; |
| 1452 |
for &b in buffer { |
| 1453 |
src[0] = b; |
| 1454 |
let (result, read, written) = self |
| 1455 |
.decoder |
| 1456 |
.decode_to_utf16_without_replacement(&src, &mut dst, false); |
| 1457 |
if written > 0 { |
| 1458 |
let half_width_katakana_state = self.half_width_katakana_state; |
| 1459 |
self.half_width_katakana_state = HalfWidthKatakana::DakutenForbidden; |
| 1460 |
let u = dst[0]; |
| 1461 |
if !self.non_ascii_seen && u >= 0x80 { |
| 1462 |
self.non_ascii_seen = true; |
| 1463 |
if u >= 0x3040 && u < 0x3100 { |
| 1464 |
// Remove the kana advantage over initial Big5 |
| 1465 |
// hanzi. |
| 1466 |
score += EUC_JP_INITIAL_KANA_PENALTY; |
| 1467 |
} |
| 1468 |
} |
| 1469 |
if (u >= u16::from(b'a') && u <= u16::from(b'z')) |
| 1470 |
|| (u >= u16::from(b'A') && u <= u16::from(b'Z')) |
| 1471 |
{ |
| 1472 |
if self.prev == LatinCj::Cj { |
| 1473 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1474 |
} |
| 1475 |
self.prev = LatinCj::AsciiLetter; |
| 1476 |
} else if u >= 0xFF61 && u <= 0xFF9F { |
| 1477 |
score += HALF_WIDTH_KATAKANA_SCORE; |
| 1478 |
|
| 1479 |
if (u >= 0xFF76 && u <= 0xFF84) || u == 0xFF73 { |
| 1480 |
self.half_width_katakana_state = HalfWidthKatakana::DakutenAllowed; |
| 1481 |
} else if u >= 0xFF8A && u <= 0xFF8E { |
| 1482 |
self.half_width_katakana_state = |
| 1483 |
HalfWidthKatakana::DakutenOrHandakutenAllowed; |
| 1484 |
} else if u == 0xFF9E { |
| 1485 |
if half_width_katakana_state == HalfWidthKatakana::DakutenForbidden { |
| 1486 |
score += IMPLAUSIBILITY_PENALTY; |
| 1487 |
} else { |
| 1488 |
score += HALF_WIDTH_KATAKANA_VOICING_SCORE; |
| 1489 |
} |
| 1490 |
} else if u == 0xFF9F { |
| 1491 |
if half_width_katakana_state |
| 1492 |
!= HalfWidthKatakana::DakutenOrHandakutenAllowed |
| 1493 |
{ |
| 1494 |
score += IMPLAUSIBILITY_PENALTY; |
| 1495 |
} else { |
| 1496 |
score += HALF_WIDTH_KATAKANA_VOICING_SCORE; |
| 1497 |
} |
| 1498 |
} |
| 1499 |
|
| 1500 |
if self.prev == LatinCj::AsciiLetter { |
| 1501 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1502 |
} |
| 1503 |
self.prev = LatinCj::Other; |
| 1504 |
} else if (u >= 0x3041 && u <= 0x3093) || (u >= 0x30A1 && u <= 0x30F6) { |
| 1505 |
match u { |
| 1506 |
0x3090 // hiragana wi |
| 1507 |
| 0x3091 // hiragana we |
| 1508 |
| 0x30F0 // katakana wi |
| 1509 |
| 0x30F1 // katakana we |
| 1510 |
=> { |
| 1511 |
// Remove advantage over Big5 Hanzi |
| 1512 |
score += EUC_JP_SCORE_PER_NEAR_OBSOLETE_KANA; |
| 1513 |
} |
| 1514 |
_ => { |
| 1515 |
score += EUC_JP_SCORE_PER_KANA; |
| 1516 |
} |
| 1517 |
} |
| 1518 |
if self.prev == LatinCj::AsciiLetter { |
| 1519 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1520 |
} |
| 1521 |
self.prev = LatinCj::Cj; |
| 1522 |
} else if (u >= 0x3400 && u < 0xA000) || (u >= 0xF900 && u < 0xFB00) { |
| 1523 |
if self.prev_prev_byte == 0x8F { |
| 1524 |
score += EUC_JP_SCORE_PER_OTHER_KANJI; |
| 1525 |
} else if self.prev_byte < 0xD0 { |
| 1526 |
score += EUC_JP_SCORE_PER_LEVEL_1_KANJI; |
| 1527 |
score += cjk_extra_score(u, &data::DETECTOR_DATA.frequent_kanji); |
| 1528 |
} else { |
| 1529 |
score += EUC_JP_SCORE_PER_LEVEL_2_KANJI; |
| 1530 |
} |
| 1531 |
if self.prev == LatinCj::AsciiLetter { |
| 1532 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1533 |
} |
| 1534 |
self.prev = LatinCj::Cj; |
| 1535 |
} else { |
| 1536 |
match u { |
| 1537 |
0x3000 // Distinct from Korean, space |
| 1538 |
| 0x3001 // Distinct from Korean, enumeration comma |
| 1539 |
| 0x3002 // Distinct from Korean, full stop |
| 1540 |
| 0xFF08 // Distinct from Korean, parenthesis |
| 1541 |
| 0xFF09 // Distinct from Korean, parenthesis |
| 1542 |
=> { |
| 1543 |
score += CJ_PUNCTUATION; |
| 1544 |
} |
| 1545 |
0..=0x7F => {} |
| 1546 |
_ => { |
| 1547 |
score += CJK_OTHER; |
| 1548 |
} |
| 1549 |
} |
| 1550 |
self.prev = LatinCj::Other; |
| 1551 |
} |
| 1552 |
} |
| 1553 |
match result { |
| 1554 |
DecoderResult::InputEmpty => { |
| 1555 |
assert_eq!(read, 1); |
| 1556 |
} |
| 1557 |
DecoderResult::Malformed(_, _) => { |
| 1558 |
if b >= 0xA1 |
| 1559 |
&& b <= 0xFE |
| 1560 |
&& self.prev_byte >= 0xA1 |
| 1561 |
&& self.prev_byte <= 0xFE |
| 1562 |
&& ((self.prev_prev_byte != 0x8F |
| 1563 |
&& !(self.prev_byte == 0xA8 && b >= 0xDF && b <= 0xE6) |
| 1564 |
&& !(self.prev_byte == 0xAC && b >= 0xF4 && b <= 0xFC) |
| 1565 |
&& !(self.prev_byte == 0xAD && b >= 0xD8 && b <= 0xDE)) |
| 1566 |
|| (self.prev_prev_byte == 0x8F |
| 1567 |
&& self.prev_byte != 0xA2 |
| 1568 |
&& self.prev_byte != 0xA6 |
| 1569 |
&& self.prev_byte != 0xA7 |
| 1570 |
&& self.prev_byte != 0xA9 |
| 1571 |
&& self.prev_byte != 0xAA |
| 1572 |
&& self.prev_byte != 0xAB |
| 1573 |
&& self.prev_byte != 0xED |
| 1574 |
&& !(self.prev_byte == 0xFE && b >= 0xF7))) |
| 1575 |
{ |
| 1576 |
score += EUC_JP_EXTENSION_PENALTY; |
| 1577 |
if self.prev == LatinCj::AsciiLetter { |
| 1578 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1579 |
} |
| 1580 |
self.prev = LatinCj::Cj; |
| 1581 |
} else { |
| 1582 |
return None; |
| 1583 |
} |
| 1584 |
} |
| 1585 |
DecoderResult::OutputFull => { |
| 1586 |
unreachable!(); |
| 1587 |
} |
| 1588 |
} |
| 1589 |
self.prev_prev_byte = self.prev_byte; |
| 1590 |
self.prev_byte = b; |
| 1591 |
} |
| 1592 |
if last { |
| 1593 |
let (result, _, _) = self |
| 1594 |
.decoder |
| 1595 |
.decode_to_utf16_without_replacement(b"", &mut dst, true); |
| 1596 |
match result { |
| 1597 |
DecoderResult::InputEmpty => {} |
| 1598 |
DecoderResult::Malformed(_, _) => { |
| 1599 |
return None; |
| 1600 |
} |
| 1601 |
DecoderResult::OutputFull => { |
| 1602 |
unreachable!(); |
| 1603 |
} |
| 1604 |
} |
| 1605 |
} |
| 1606 |
Some(score) |
| 1607 |
} |
| 1608 |
} |
| 1609 |
|
| 1610 |
struct Big5Candidate { |
| 1611 |
decoder: Decoder, |
| 1612 |
prev: LatinCj, |
| 1613 |
prev_byte: u8, |
| 1614 |
pending_score: Option<i64>, |
| 1615 |
} |
| 1616 |
|
| 1617 |
impl Big5Candidate { |
| 1618 |
fn maybe_set_as_pending(&mut self, s: i64) -> i64 { |
| 1619 |
assert!(self.pending_score.is_none()); |
| 1620 |
if self.prev == LatinCj::Cj || !problematic_lead(self.prev_byte) { |
| 1621 |
s |
| 1622 |
} else { |
| 1623 |
self.pending_score = Some(s); |
| 1624 |
0 |
| 1625 |
} |
| 1626 |
} |
| 1627 |
|
| 1628 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1629 |
let mut score = 0i64; |
| 1630 |
let mut src = [0u8]; |
| 1631 |
let mut dst = [0u16; 2]; |
| 1632 |
for &b in buffer { |
| 1633 |
src[0] = b; |
| 1634 |
let (result, read, written) = self |
| 1635 |
.decoder |
| 1636 |
.decode_to_utf16_without_replacement(&src, &mut dst, false); |
| 1637 |
if written == 1 { |
| 1638 |
let u = dst[0]; |
| 1639 |
if (u >= u16::from(b'a') && u <= u16::from(b'z')) |
| 1640 |
|| (u >= u16::from(b'A') && u <= u16::from(b'Z')) |
| 1641 |
{ |
| 1642 |
self.pending_score = None; // Discard pending score |
| 1643 |
if self.prev == LatinCj::Cj { |
| 1644 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1645 |
} |
| 1646 |
self.prev = LatinCj::AsciiLetter; |
| 1647 |
} else if (u >= 0x3400 && u < 0xA000) || (u >= 0xF900 && u < 0xFB00) { |
| 1648 |
if let Some(pending) = self.pending_score { |
| 1649 |
score += pending; |
| 1650 |
self.pending_score = None; |
| 1651 |
} |
| 1652 |
match self.prev_byte { |
| 1653 |
0xA4..=0xC6 => { |
| 1654 |
score += self.maybe_set_as_pending(BIG5_SCORE_PER_LEVEL_1_HANZI); |
| 1655 |
// score += cjk_extra_score(u, &data::DETECTOR_DATA.frequent_traditional); |
| 1656 |
} |
| 1657 |
_ => { |
| 1658 |
score += self.maybe_set_as_pending(BIG5_SCORE_PER_OTHER_HANZI); |
| 1659 |
} |
| 1660 |
} |
| 1661 |
if self.prev == LatinCj::AsciiLetter { |
| 1662 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1663 |
} |
| 1664 |
self.prev = LatinCj::Cj; |
| 1665 |
} else { |
| 1666 |
match u { |
| 1667 |
0x3000 // Distinct from Korean, space |
| 1668 |
| 0x3001 // Distinct from Korean, enumeration comma |
| 1669 |
| 0x3002 // Distinct from Korean, full stop |
| 1670 |
| 0xFF08 // Distinct from Korean, parenthesis |
| 1671 |
| 0xFF09 // Distinct from Korean, parenthesis |
| 1672 |
| 0xFF01 // Distinct from Japanese, exclamation |
| 1673 |
| 0xFF0C // Distinct from Japanese, comma |
| 1674 |
| 0xFF1B // Distinct from Japanese, semicolon |
| 1675 |
| 0xFF1F // Distinct from Japanese, question |
| 1676 |
=> { |
| 1677 |
if let Some(pending) = self.pending_score { |
| 1678 |
score += pending; |
| 1679 |
self.pending_score = None; |
| 1680 |
} |
| 1681 |
// Not really needed for CJK distinction |
| 1682 |
// but let's give non-zero score for these |
| 1683 |
// common byte pairs anyway. |
| 1684 |
score += CJ_PUNCTUATION; |
| 1685 |
} |
| 1686 |
0..=0x7F => { |
| 1687 |
self.pending_score = None; // Discard pending score |
| 1688 |
} |
| 1689 |
_ => { |
| 1690 |
if let Some(pending) = self.pending_score { |
| 1691 |
score += pending; |
| 1692 |
self.pending_score = None; |
| 1693 |
} |
| 1694 |
score += CJK_OTHER; |
| 1695 |
} |
| 1696 |
} |
| 1697 |
self.prev = LatinCj::Other; |
| 1698 |
} |
| 1699 |
} else if written == 2 { |
| 1700 |
if let Some(pending) = self.pending_score { |
| 1701 |
score += pending; |
| 1702 |
self.pending_score = None; |
| 1703 |
} |
| 1704 |
if dst[0] == 0xCA || dst[0] == 0xEA { |
| 1705 |
score += CJK_OTHER; |
| 1706 |
self.prev = LatinCj::Other; |
| 1707 |
} else { |
| 1708 |
debug_assert!(dst[0] >= 0xD480 && dst[0] < 0xD880); |
| 1709 |
score += self.maybe_set_as_pending(BIG5_SCORE_PER_OTHER_HANZI); |
| 1710 |
if self.prev == LatinCj::AsciiLetter { |
| 1711 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1712 |
} |
| 1713 |
self.prev = LatinCj::Cj; |
| 1714 |
} |
| 1715 |
} |
| 1716 |
match result { |
| 1717 |
DecoderResult::InputEmpty => { |
| 1718 |
assert_eq!(read, 1); |
| 1719 |
} |
| 1720 |
DecoderResult::Malformed(malformed_len, _) => { |
| 1721 |
if self.prev_byte >= 0x81 |
| 1722 |
&& self.prev_byte <= 0xFE |
| 1723 |
&& ((b >= 0x40 && b <= 0x7E) || (b >= 0xA1 && b <= 0xFE)) |
| 1724 |
{ |
| 1725 |
// The byte pair is in the Big5 range but unmapped. |
| 1726 |
// Treat as PUA to avoid rejecting Big5-UAO, etc. |
| 1727 |
// We don't reprocess `b` even if ASCII, since it's |
| 1728 |
// logically part of the pair. |
| 1729 |
if let Some(pending) = self.pending_score { |
| 1730 |
score += pending; |
| 1731 |
self.pending_score = None; |
| 1732 |
} |
| 1733 |
score += BIG5_PUA_PENALTY; |
| 1734 |
// Assume Hanzi semantics |
| 1735 |
if self.prev == LatinCj::AsciiLetter { |
| 1736 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1737 |
} |
| 1738 |
self.prev = LatinCj::Cj; |
| 1739 |
} else if (self.prev_byte == 0xA0 |
| 1740 |
|| self.prev_byte == 0xFD |
| 1741 |
|| self.prev_byte == 0xFE) |
| 1742 |
&& (b < 0x80 || b == 0xFF) |
| 1743 |
{ |
| 1744 |
// Mac OS Chinese Traditional single-byte that conflicts with code page Big5 lead byte |
| 1745 |
// followed by ASCII or a non-conflicting single-byte extension. |
| 1746 |
self.pending_score = None; // Just in case |
| 1747 |
score += BIG5_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1748 |
if (b >= b'a' && b <= b'z') || (b >= b'A' && b <= b'Z') { |
| 1749 |
self.prev = LatinCj::AsciiLetter; |
| 1750 |
} else if b == 0xFF { |
| 1751 |
score += BIG5_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1752 |
self.prev = LatinCj::Other; |
| 1753 |
} else { |
| 1754 |
self.prev = LatinCj::Other; |
| 1755 |
} |
| 1756 |
} else if malformed_len == 1 && b == 0xFF { |
| 1757 |
// Mac OS Chinese Traditional single-byte extension that doesn't conflict with lead bytes |
| 1758 |
self.pending_score = None; // Just in case |
| 1759 |
score += BIG5_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1760 |
self.prev = LatinCj::Other; |
| 1761 |
} else { |
| 1762 |
return None; |
| 1763 |
} |
| 1764 |
} |
| 1765 |
DecoderResult::OutputFull => { |
| 1766 |
unreachable!(); |
| 1767 |
} |
| 1768 |
} |
| 1769 |
self.prev_byte = b; |
| 1770 |
} |
| 1771 |
if last { |
| 1772 |
let (result, _, _) = self |
| 1773 |
.decoder |
| 1774 |
.decode_to_utf16_without_replacement(b"", &mut dst, true); |
| 1775 |
match result { |
| 1776 |
DecoderResult::InputEmpty => {} |
| 1777 |
DecoderResult::Malformed(_, _) => { |
| 1778 |
return None; |
| 1779 |
} |
| 1780 |
DecoderResult::OutputFull => { |
| 1781 |
unreachable!(); |
| 1782 |
} |
| 1783 |
} |
| 1784 |
} |
| 1785 |
Some(score) |
| 1786 |
} |
| 1787 |
} |
| 1788 |
|
| 1789 |
struct EucKrCandidate { |
| 1790 |
decoder: Decoder, |
| 1791 |
prev_byte: u8, |
| 1792 |
prev_was_euc_range: bool, |
| 1793 |
prev: LatinKorean, |
| 1794 |
current_word_len: u64, |
| 1795 |
pending_score: Option<i64>, |
| 1796 |
} |
| 1797 |
|
| 1798 |
impl EucKrCandidate { |
| 1799 |
fn maybe_set_as_pending(&mut self, s: i64) -> i64 { |
| 1800 |
assert!(self.pending_score.is_none()); |
| 1801 |
if self.prev == LatinKorean::Hangul || !more_problematic_lead(self.prev_byte) { |
| 1802 |
s |
| 1803 |
} else { |
| 1804 |
self.pending_score = Some(s); |
| 1805 |
0 |
| 1806 |
} |
| 1807 |
} |
| 1808 |
|
| 1809 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1810 |
let mut score = 0i64; |
| 1811 |
let mut src = [0u8]; |
| 1812 |
let mut dst = [0u16; 2]; |
| 1813 |
for &b in buffer { |
| 1814 |
let in_euc_range = b >= 0xA1 && b <= 0xFE; |
| 1815 |
src[0] = b; |
| 1816 |
let (result, read, written) = self |
| 1817 |
.decoder |
| 1818 |
.decode_to_utf16_without_replacement(&src, &mut dst, false); |
| 1819 |
if written > 0 { |
| 1820 |
let u = dst[0]; |
| 1821 |
if (u >= u16::from(b'a') && u <= u16::from(b'z')) |
| 1822 |
|| (u >= u16::from(b'A') && u <= u16::from(b'Z')) |
| 1823 |
{ |
| 1824 |
self.pending_score = None; // Discard pending score |
| 1825 |
match self.prev { |
| 1826 |
LatinKorean::Hangul | LatinKorean::Hanja => { |
| 1827 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1828 |
} |
| 1829 |
_ => {} |
| 1830 |
} |
| 1831 |
self.prev = LatinKorean::AsciiLetter; |
| 1832 |
self.current_word_len = 0; |
| 1833 |
} else if u >= 0xAC00 && u <= 0xD7A3 { |
| 1834 |
if let Some(pending) = self.pending_score { |
| 1835 |
score += pending; |
| 1836 |
self.pending_score = None; |
| 1837 |
} |
| 1838 |
if self.prev_was_euc_range && in_euc_range { |
| 1839 |
score += EUC_KR_SCORE_PER_EUC_HANGUL; |
| 1840 |
score += cjk_extra_score(u, &data::DETECTOR_DATA.frequent_hangul); |
| 1841 |
} else { |
| 1842 |
score += self.maybe_set_as_pending(EUC_KR_SCORE_PER_NON_EUC_HANGUL); |
| 1843 |
} |
| 1844 |
if self.prev == LatinKorean::AsciiLetter { |
| 1845 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1846 |
} |
| 1847 |
self.prev = LatinKorean::Hangul; |
| 1848 |
self.current_word_len += 1; |
| 1849 |
if self.current_word_len > 5 { |
| 1850 |
score += EUC_KR_LONG_WORD_PENALTY; |
| 1851 |
} |
| 1852 |
} else if (u >= 0x4E00 && u < 0xAC00) || (u >= 0xF900 && u <= 0xFA0B) { |
| 1853 |
if let Some(pending) = self.pending_score { |
| 1854 |
score += pending; |
| 1855 |
self.pending_score = None; |
| 1856 |
} |
| 1857 |
score += EUC_KR_SCORE_PER_HANJA; |
| 1858 |
match self.prev { |
| 1859 |
LatinKorean::AsciiLetter => { |
| 1860 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1861 |
} |
| 1862 |
LatinKorean::Hangul => { |
| 1863 |
score += EUC_KR_HANJA_AFTER_HANGUL_PENALTY; |
| 1864 |
} |
| 1865 |
_ => {} |
| 1866 |
} |
| 1867 |
self.prev = LatinKorean::Hanja; |
| 1868 |
self.current_word_len += 1; |
| 1869 |
if self.current_word_len > 5 { |
| 1870 |
score += EUC_KR_LONG_WORD_PENALTY; |
| 1871 |
} |
| 1872 |
} else { |
| 1873 |
if u >= 0x80 { |
| 1874 |
if let Some(pending) = self.pending_score { |
| 1875 |
score += pending; |
| 1876 |
self.pending_score = None; |
| 1877 |
} |
| 1878 |
score += CJK_OTHER; |
| 1879 |
} else { |
| 1880 |
self.pending_score = None; // Discard pending score |
| 1881 |
} |
| 1882 |
self.prev = LatinKorean::Other; |
| 1883 |
self.current_word_len = 0; |
| 1884 |
} |
| 1885 |
} |
| 1886 |
match result { |
| 1887 |
DecoderResult::InputEmpty => { |
| 1888 |
assert_eq!(read, 1); |
| 1889 |
} |
| 1890 |
DecoderResult::Malformed(malformed_len, _) => { |
| 1891 |
if (self.prev_byte == 0xC9 || self.prev_byte == 0xFE) && b >= 0xA1 && b <= 0xFE |
| 1892 |
{ |
| 1893 |
if let Some(pending) = self.pending_score { |
| 1894 |
score += pending; |
| 1895 |
self.pending_score = None; |
| 1896 |
} |
| 1897 |
// The byte pair is in code page 949 EUDC range |
| 1898 |
score += EUC_KR_PUA_PENALTY; |
| 1899 |
// Assume Hanja semantics |
| 1900 |
match self.prev { |
| 1901 |
LatinKorean::AsciiLetter => { |
| 1902 |
score += CJK_LATIN_ADJACENCY_PENALTY; |
| 1903 |
} |
| 1904 |
LatinKorean::Hangul => { |
| 1905 |
score += EUC_KR_HANJA_AFTER_HANGUL_PENALTY; |
| 1906 |
} |
| 1907 |
_ => {} |
| 1908 |
} |
| 1909 |
self.prev = LatinKorean::Hanja; |
| 1910 |
self.current_word_len += 1; |
| 1911 |
if self.current_word_len > 5 { |
| 1912 |
score += EUC_KR_LONG_WORD_PENALTY; |
| 1913 |
} |
| 1914 |
} else if (self.prev_byte == 0xA1 |
| 1915 |
|| (self.prev_byte >= 0xA3 && self.prev_byte <= 0xA8) |
| 1916 |
|| (self.prev_byte >= 0xAA && self.prev_byte <= 0xAD)) |
| 1917 |
&& (b >= 0x7B && b <= 0x7D) |
| 1918 |
{ |
| 1919 |
if let Some(pending) = self.pending_score { |
| 1920 |
score += pending; |
| 1921 |
self.pending_score = None; |
| 1922 |
} |
| 1923 |
// MacKorean symbols in range not part of code page 949 |
| 1924 |
score += EUC_KR_MAC_KOREAN_PENALTY; |
| 1925 |
self.prev = LatinKorean::Other; |
| 1926 |
self.current_word_len = 0; |
| 1927 |
} else if (self.prev_byte >= 0x81 && self.prev_byte <= 0x84) |
| 1928 |
&& (b <= 0x80 || b == 0xFF) |
| 1929 |
{ |
| 1930 |
// MacKorean single-byte that conflicts with code page 949 lead byte |
| 1931 |
// followed by ASCII or a non-conflicting single-byte extension. |
| 1932 |
self.pending_score = None; // Just in case |
| 1933 |
score += EUC_KR_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1934 |
if (b >= b'a' && b <= b'z') || (b >= b'A' && b <= b'Z') { |
| 1935 |
self.prev = LatinKorean::AsciiLetter; |
| 1936 |
} else if b == 0x80 || b == 0xFF { |
| 1937 |
score += EUC_KR_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1938 |
self.prev = LatinKorean::Other; |
| 1939 |
} else { |
| 1940 |
self.prev = LatinKorean::Other; |
| 1941 |
} |
| 1942 |
self.current_word_len = 0; |
| 1943 |
} else if malformed_len == 1 && (b == 0x80 || b == 0xFF) { |
| 1944 |
// MacKorean single-byte extensions that don't conflict with lead bytes |
| 1945 |
self.pending_score = None; // Just in case |
| 1946 |
score += EUC_KR_SINGLE_BYTE_EXTENSION_PENALTY; |
| 1947 |
self.prev = LatinKorean::Other; |
| 1948 |
self.current_word_len = 0; |
| 1949 |
} else { |
| 1950 |
return None; |
| 1951 |
} |
| 1952 |
} |
| 1953 |
DecoderResult::OutputFull => { |
| 1954 |
unreachable!(); |
| 1955 |
} |
| 1956 |
} |
| 1957 |
self.prev_was_euc_range = in_euc_range; |
| 1958 |
self.prev_byte = b; |
| 1959 |
} |
| 1960 |
if last { |
| 1961 |
let (result, _, _) = self |
| 1962 |
.decoder |
| 1963 |
.decode_to_utf16_without_replacement(b"", &mut dst, true); |
| 1964 |
match result { |
| 1965 |
DecoderResult::InputEmpty => {} |
| 1966 |
DecoderResult::Malformed(_, _) => { |
| 1967 |
return None; |
| 1968 |
} |
| 1969 |
DecoderResult::OutputFull => { |
| 1970 |
unreachable!(); |
| 1971 |
} |
| 1972 |
} |
| 1973 |
} |
| 1974 |
Some(score) |
| 1975 |
} |
| 1976 |
} |
| 1977 |
|
| 1978 |
enum InnerCandidate { |
| 1979 |
Latin(LatinCandidate), |
| 1980 |
NonLatinCased(NonLatinCasedCandidate), |
| 1981 |
Caseless(CaselessCandidate), |
| 1982 |
ArabicFrench(ArabicFrenchCandidate), |
| 1983 |
Logical(LogicalCandidate), |
| 1984 |
Visual(VisualCandidate), |
| 1985 |
Utf8(Utf8Candidate), |
| 1986 |
Iso2022(Iso2022Candidate), |
| 1987 |
Shift(ShiftJisCandidate), |
| 1988 |
EucJp(EucJpCandidate), |
| 1989 |
EucKr(EucKrCandidate), |
| 1990 |
Big5(Big5Candidate), |
| 1991 |
Gbk(GbkCandidate), |
| 1992 |
} |
| 1993 |
|
| 1994 |
impl InnerCandidate { |
| 1995 |
fn feed(&mut self, buffer: &[u8], last: bool) -> Option<i64> { |
| 1996 |
match self { |
| 1997 |
InnerCandidate::Latin(c) => { |
| 1998 |
if let Some(new_score) = c.feed(buffer) { |
| 1999 |
if last { |
| 2000 |
// Treat EOF as space-like |
| 2001 |
if let Some(additional_score) = c.feed(b" ") { |
| 2002 |
Some(new_score + additional_score) |
| 2003 |
} else { |
| 2004 |
None |
| 2005 |
} |
| 2006 |
} else { |
| 2007 |
Some(new_score) |
| 2008 |
} |
| 2009 |
} else { |
| 2010 |
None |
| 2011 |
} |
| 2012 |
} |
| 2013 |
InnerCandidate::NonLatinCased(c) => { |
| 2014 |
if let Some(new_score) = c.feed(buffer) { |
| 2015 |
if last { |
| 2016 |
// Treat EOF as space-like |
| 2017 |
if let Some(additional_score) = c.feed(b" ") { |
| 2018 |
Some(new_score + additional_score) |
| 2019 |
} else { |
| 2020 |
None |
| 2021 |
} |
| 2022 |
} else { |
| 2023 |
Some(new_score) |
| 2024 |
} |
| 2025 |
} else { |
| 2026 |
None |
| 2027 |
} |
| 2028 |
} |
| 2029 |
InnerCandidate::Caseless(c) => { |
| 2030 |
if let Some(new_score) = c.feed(buffer) { |
| 2031 |
if last { |
| 2032 |
// Treat EOF as space-like |
| 2033 |
if let Some(additional_score) = c.feed(b" ") { |
| 2034 |
Some(new_score + additional_score) |
| 2035 |
} else { |
| 2036 |
None |
| 2037 |
} |
| 2038 |
} else { |
| 2039 |
Some(new_score) |
| 2040 |
} |
| 2041 |
} else { |
| 2042 |
None |
| 2043 |
} |
| 2044 |
} |
| 2045 |
InnerCandidate::ArabicFrench(c) => { |
| 2046 |
if let Some(new_score) = c.feed(buffer) { |
| 2047 |
if last { |
| 2048 |
// Treat EOF as space-like |
| 2049 |
if let Some(additional_score) = c.feed(b" ") { |
| 2050 |
Some(new_score + additional_score) |
| 2051 |
} else { |
| 2052 |
None |
| 2053 |
} |
| 2054 |
} else { |
| 2055 |
Some(new_score) |
| 2056 |
} |
| 2057 |
} else { |
| 2058 |
None |
| 2059 |
} |
| 2060 |
} |
| 2061 |
InnerCandidate::Logical(c) => { |
| 2062 |
if let Some(new_score) = c.feed(buffer) { |
| 2063 |
if last { |
| 2064 |
// Treat EOF as space-like |
| 2065 |
if let Some(additional_score) = c.feed(b" ") { |
| 2066 |
Some(new_score + additional_score) |
| 2067 |
} else { |
| 2068 |
None |
| 2069 |
} |
| 2070 |
} else { |
| 2071 |
Some(new_score) |
| 2072 |
} |
| 2073 |
} else { |
| 2074 |
None |
| 2075 |
} |
| 2076 |
} |
| 2077 |
InnerCandidate::Visual(c) => { |
| 2078 |
if let Some(new_score) = c.feed(buffer) { |
| 2079 |
if last { |
| 2080 |
// Treat EOF as space-like |
| 2081 |
if let Some(additional_score) = c.feed(b" ") { |
| 2082 |
Some(new_score + additional_score) |
| 2083 |
} else { |
| 2084 |
None |
| 2085 |
} |
| 2086 |
} else { |
| 2087 |
Some(new_score) |
| 2088 |
} |
| 2089 |
} else { |
| 2090 |
None |
| 2091 |
} |
| 2092 |
} |
| 2093 |
InnerCandidate::Utf8(c) => c.feed(buffer, last), |
| 2094 |
InnerCandidate::Iso2022(c) => c.feed(buffer, last), |
| 2095 |
InnerCandidate::Shift(c) => c.feed(buffer, last), |
| 2096 |
InnerCandidate::EucJp(c) => c.feed(buffer, last), |
| 2097 |
InnerCandidate::EucKr(c) => c.feed(buffer, last), |
| 2098 |
InnerCandidate::Big5(c) => c.feed(buffer, last), |
| 2099 |
InnerCandidate::Gbk(c) => c.feed(buffer, last), |
| 2100 |
} |
| 2101 |
} |
| 2102 |
} |
| 2103 |
|
| 2104 |
fn encoding_for_tld(tld: Tld) -> usize { |
| 2105 |
match tld { |
| 2106 |
Tld::CentralWindows | Tld::CentralCyrillic => EncodingDetector::CENTRAL_WINDOWS_INDEX, |
| 2107 |
Tld::Cyrillic => EncodingDetector::CYRILLIC_WINDOWS_INDEX, |
| 2108 |
Tld::Generic | Tld::Western | Tld::WesternCyrillic | Tld::WesternArabic | Tld::Eu => { |
| 2109 |
EncodingDetector::WESTERN_INDEX |
| 2110 |
} |
| 2111 |
Tld::IcelandicFaroese => EncodingDetector::ICELANDIC_INDEX, |
| 2112 |
Tld::Greek => EncodingDetector::GREEK_ISO_INDEX, |
| 2113 |
Tld::TurkishAzeri => EncodingDetector::TURKISH_INDEX, |
| 2114 |
Tld::Hebrew => EncodingDetector::LOGICAL_INDEX, |
| 2115 |
Tld::Arabic => EncodingDetector::ARABIC_WINDOWS_INDEX, |
| 2116 |
Tld::Baltic => EncodingDetector::BALTIC_WINDOWS_INDEX, |
| 2117 |
Tld::Vietnamese => EncodingDetector::VIETNAMESE_INDEX, |
| 2118 |
Tld::Thai => EncodingDetector::THAI_INDEX, |
| 2119 |
Tld::Simplified | Tld::SimplifiedTraditional => EncodingDetector::GBK_INDEX, |
| 2120 |
Tld::Traditional | Tld::TraditionalSimplified => EncodingDetector::BIG5_INDEX, |
| 2121 |
Tld::Japanese => EncodingDetector::SHIFT_JIS_INDEX, |
| 2122 |
Tld::Korean => EncodingDetector::EUC_KR_INDEX, |
| 2123 |
Tld::CentralIso => EncodingDetector::CENTRAL_ISO_INDEX, |
| 2124 |
} |
| 2125 |
} |
| 2126 |
|
| 2127 |
fn encoding_is_native_to_tld(tld: Tld, encoding: usize) -> bool { |
| 2128 |
match tld { |
| 2129 |
Tld::CentralWindows => encoding == EncodingDetector::CENTRAL_WINDOWS_INDEX, |
| 2130 |
Tld::Cyrillic => { |
| 2131 |
encoding == EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2132 |
|| encoding == EncodingDetector::CYRILLIC_KOI_INDEX |
| 2133 |
|| encoding == EncodingDetector::CYRILLIC_IBM_INDEX |
| 2134 |
|| encoding == EncodingDetector::CYRILLIC_ISO_INDEX |
| 2135 |
} |
| 2136 |
Tld::Western => encoding == EncodingDetector::WESTERN_INDEX, |
| 2137 |
Tld::Greek => { |
| 2138 |
encoding == EncodingDetector::GREEK_WINDOWS_INDEX |
| 2139 |
|| encoding == EncodingDetector::GREEK_ISO_INDEX |
| 2140 |
} |
| 2141 |
Tld::TurkishAzeri => encoding == EncodingDetector::TURKISH_INDEX, |
| 2142 |
Tld::Hebrew => encoding == EncodingDetector::LOGICAL_INDEX, |
| 2143 |
Tld::Arabic => { |
| 2144 |
encoding == EncodingDetector::ARABIC_WINDOWS_INDEX |
| 2145 |
|| encoding == EncodingDetector::ARABIC_ISO_INDEX |
| 2146 |
} |
| 2147 |
Tld::Baltic => { |
| 2148 |
encoding == EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2149 |
|| encoding == EncodingDetector::BALTIC_ISO13_INDEX |
| 2150 |
|| encoding == EncodingDetector::BALTIC_ISO4_INDEX |
| 2151 |
} |
| 2152 |
Tld::Vietnamese => encoding == EncodingDetector::VIETNAMESE_INDEX, |
| 2153 |
Tld::Thai => encoding == EncodingDetector::THAI_INDEX, |
| 2154 |
Tld::Simplified => encoding == EncodingDetector::GBK_INDEX, |
| 2155 |
Tld::Traditional => encoding == EncodingDetector::BIG5_INDEX, |
| 2156 |
Tld::Japanese => { |
| 2157 |
encoding == EncodingDetector::SHIFT_JIS_INDEX |
| 2158 |
|| encoding == EncodingDetector::EUC_JP_INDEX |
| 2159 |
} |
| 2160 |
Tld::Korean => encoding == EncodingDetector::EUC_KR_INDEX, |
| 2161 |
Tld::SimplifiedTraditional | Tld::TraditionalSimplified => { |
| 2162 |
encoding == EncodingDetector::GBK_INDEX || encoding == EncodingDetector::BIG5_INDEX |
| 2163 |
} |
| 2164 |
Tld::CentralIso => encoding == EncodingDetector::CENTRAL_ISO_INDEX, |
| 2165 |
Tld::IcelandicFaroese => encoding == EncodingDetector::ICELANDIC_INDEX, |
| 2166 |
Tld::WesternCyrillic => { |
| 2167 |
encoding == EncodingDetector::WESTERN_INDEX |
| 2168 |
|| encoding == EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2169 |
|| encoding == EncodingDetector::CYRILLIC_KOI_INDEX |
| 2170 |
|| encoding == EncodingDetector::CYRILLIC_IBM_INDEX |
| 2171 |
|| encoding == EncodingDetector::CYRILLIC_ISO_INDEX |
| 2172 |
} |
| 2173 |
Tld::CentralCyrillic => { |
| 2174 |
encoding == EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2175 |
|| encoding == EncodingDetector::CENTRAL_ISO_INDEX |
| 2176 |
|| encoding == EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2177 |
|| encoding == EncodingDetector::CYRILLIC_KOI_INDEX |
| 2178 |
|| encoding == EncodingDetector::CYRILLIC_IBM_INDEX |
| 2179 |
|| encoding == EncodingDetector::CYRILLIC_ISO_INDEX |
| 2180 |
} |
| 2181 |
Tld::WesternArabic => { |
| 2182 |
encoding == EncodingDetector::WESTERN_INDEX |
| 2183 |
|| encoding == EncodingDetector::ARABIC_WINDOWS_INDEX |
| 2184 |
|| encoding == EncodingDetector::ARABIC_ISO_INDEX |
| 2185 |
} |
| 2186 |
Tld::Eu => { |
| 2187 |
encoding == EncodingDetector::WESTERN_INDEX |
| 2188 |
|| encoding == EncodingDetector::ICELANDIC_INDEX |
| 2189 |
|| encoding == EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2190 |
|| encoding == EncodingDetector::CENTRAL_ISO_INDEX |
| 2191 |
|| encoding == EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2192 |
|| encoding == EncodingDetector::CYRILLIC_KOI_INDEX |
| 2193 |
|| encoding == EncodingDetector::CYRILLIC_IBM_INDEX |
| 2194 |
|| encoding == EncodingDetector::CYRILLIC_ISO_INDEX |
| 2195 |
|| encoding == EncodingDetector::GREEK_WINDOWS_INDEX |
| 2196 |
|| encoding == EncodingDetector::GREEK_ISO_INDEX |
| 2197 |
|| encoding == EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2198 |
|| encoding == EncodingDetector::BALTIC_ISO13_INDEX |
| 2199 |
|| encoding == EncodingDetector::BALTIC_ISO4_INDEX |
| 2200 |
} |
| 2201 |
Tld::Generic => false, |
| 2202 |
} |
| 2203 |
} |
| 2204 |
|
| 2205 |
fn score_adjustment(score: i64, encoding: usize, tld: Tld) -> i64 { |
| 2206 |
if score < 1 { |
| 2207 |
return 0; |
| 2208 |
} |
| 2209 |
// This is the most ad hoc part of this library. |
| 2210 |
let (divisor, constant) = match tld { |
| 2211 |
Tld::Generic => { |
| 2212 |
unreachable!(); |
| 2213 |
} |
| 2214 |
Tld::CentralWindows | Tld::CentralIso => { |
| 2215 |
match encoding { |
| 2216 |
EncodingDetector::WESTERN_INDEX |
| 2217 |
| EncodingDetector::ICELANDIC_INDEX |
| 2218 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2219 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2220 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2221 |
| EncodingDetector::VIETNAMESE_INDEX |
| 2222 |
| EncodingDetector::TURKISH_INDEX => { |
| 2223 |
// XXX Tune this better instead of this kind of absolute. |
| 2224 |
return score; |
| 2225 |
} |
| 2226 |
_ => (50, 60), |
| 2227 |
} |
| 2228 |
} |
| 2229 |
Tld::Cyrillic => { |
| 2230 |
match encoding { |
| 2231 |
EncodingDetector::BIG5_INDEX |
| 2232 |
| EncodingDetector::GBK_INDEX |
| 2233 |
| EncodingDetector::EUC_JP_INDEX |
| 2234 |
| EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2235 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2236 |
| EncodingDetector::GREEK_WINDOWS_INDEX |
| 2237 |
| EncodingDetector::GREEK_ISO_INDEX |
| 2238 |
| EncodingDetector::VISUAL_INDEX |
| 2239 |
| EncodingDetector::LOGICAL_INDEX |
| 2240 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2241 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2242 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2243 |
| EncodingDetector::TURKISH_INDEX => { |
| 2244 |
// XXX Tune this better instead of this kind of absolute. |
| 2245 |
return score; |
| 2246 |
} |
| 2247 |
_ => (50, 60), |
| 2248 |
} |
| 2249 |
} |
| 2250 |
Tld::Western | Tld::WesternCyrillic | Tld::WesternArabic => { |
| 2251 |
match encoding { |
| 2252 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2253 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2254 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2255 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2256 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2257 |
| EncodingDetector::TURKISH_INDEX |
| 2258 |
| EncodingDetector::VIETNAMESE_INDEX => { |
| 2259 |
// XXX Tune this better instead of this kind of absolute. |
| 2260 |
return score; |
| 2261 |
} |
| 2262 |
_ => (50, 60), |
| 2263 |
} |
| 2264 |
} |
| 2265 |
Tld::Greek => { |
| 2266 |
match encoding { |
| 2267 |
EncodingDetector::BIG5_INDEX |
| 2268 |
| EncodingDetector::GBK_INDEX |
| 2269 |
| EncodingDetector::EUC_JP_INDEX |
| 2270 |
| EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2271 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2272 |
| EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2273 |
| EncodingDetector::CYRILLIC_ISO_INDEX |
| 2274 |
| EncodingDetector::CYRILLIC_KOI_INDEX |
| 2275 |
| EncodingDetector::CYRILLIC_IBM_INDEX |
| 2276 |
| EncodingDetector::VISUAL_INDEX |
| 2277 |
| EncodingDetector::LOGICAL_INDEX |
| 2278 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2279 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2280 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2281 |
| EncodingDetector::TURKISH_INDEX => { |
| 2282 |
// XXX Tune this better instead of this kind of absolute. |
| 2283 |
return score; |
| 2284 |
} |
| 2285 |
_ => (50, 60), |
| 2286 |
} |
| 2287 |
} |
| 2288 |
Tld::TurkishAzeri => { |
| 2289 |
match encoding { |
| 2290 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2291 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2292 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2293 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2294 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2295 |
| EncodingDetector::VIETNAMESE_INDEX |
| 2296 |
| EncodingDetector::ICELANDIC_INDEX => { |
| 2297 |
// XXX Tune this better instead of this kind of absolute. |
| 2298 |
return score; |
| 2299 |
} |
| 2300 |
_ => (50, 60), |
| 2301 |
} |
| 2302 |
} |
| 2303 |
Tld::Hebrew => { |
| 2304 |
match encoding { |
| 2305 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2306 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2307 |
| EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2308 |
| EncodingDetector::CYRILLIC_ISO_INDEX |
| 2309 |
| EncodingDetector::CYRILLIC_KOI_INDEX |
| 2310 |
| EncodingDetector::CYRILLIC_IBM_INDEX |
| 2311 |
| EncodingDetector::GREEK_WINDOWS_INDEX |
| 2312 |
| EncodingDetector::GREEK_ISO_INDEX |
| 2313 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2314 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2315 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2316 |
| EncodingDetector::VIETNAMESE_INDEX |
| 2317 |
| EncodingDetector::TURKISH_INDEX => { |
| 2318 |
// XXX Tune this better instead of this kind of absolute. |
| 2319 |
return score; |
| 2320 |
} |
| 2321 |
_ => (50, 60), |
| 2322 |
} |
| 2323 |
} |
| 2324 |
Tld::Arabic => { |
| 2325 |
match encoding { |
| 2326 |
EncodingDetector::BIG5_INDEX |
| 2327 |
| EncodingDetector::GBK_INDEX |
| 2328 |
| EncodingDetector::EUC_JP_INDEX |
| 2329 |
| EncodingDetector::EUC_KR_INDEX |
| 2330 |
| EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2331 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2332 |
| EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2333 |
| EncodingDetector::CYRILLIC_ISO_INDEX |
| 2334 |
| EncodingDetector::CYRILLIC_KOI_INDEX |
| 2335 |
| EncodingDetector::CYRILLIC_IBM_INDEX |
| 2336 |
| EncodingDetector::GREEK_WINDOWS_INDEX |
| 2337 |
| EncodingDetector::GREEK_ISO_INDEX |
| 2338 |
| EncodingDetector::VISUAL_INDEX |
| 2339 |
| EncodingDetector::LOGICAL_INDEX |
| 2340 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2341 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2342 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2343 |
| EncodingDetector::VIETNAMESE_INDEX |
| 2344 |
| EncodingDetector::TURKISH_INDEX => { |
| 2345 |
// XXX Tune this better instead of this kind of absolute. |
| 2346 |
return score; |
| 2347 |
} |
| 2348 |
_ => (50, 60), |
| 2349 |
} |
| 2350 |
} |
| 2351 |
Tld::Baltic => { |
| 2352 |
match encoding { |
| 2353 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2354 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2355 |
| EncodingDetector::ICELANDIC_INDEX |
| 2356 |
| EncodingDetector::TURKISH_INDEX |
| 2357 |
| EncodingDetector::VIETNAMESE_INDEX => { |
| 2358 |
// XXX Tune this better instead of this kind of absolute. |
| 2359 |
return score; |
| 2360 |
} |
| 2361 |
_ => (50, 60), |
| 2362 |
} |
| 2363 |
} |
| 2364 |
Tld::Vietnamese => { |
| 2365 |
match encoding { |
| 2366 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2367 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2368 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2369 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2370 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2371 |
| EncodingDetector::TURKISH_INDEX |
| 2372 |
| EncodingDetector::ICELANDIC_INDEX => { |
| 2373 |
// XXX Tune this better instead of this kind of absolute. |
| 2374 |
return score; |
| 2375 |
} |
| 2376 |
_ => (50, 60), |
| 2377 |
} |
| 2378 |
} |
| 2379 |
Tld::Thai => { |
| 2380 |
match encoding { |
| 2381 |
EncodingDetector::BIG5_INDEX |
| 2382 |
| EncodingDetector::GBK_INDEX |
| 2383 |
| EncodingDetector::EUC_JP_INDEX |
| 2384 |
| EncodingDetector::EUC_KR_INDEX |
| 2385 |
| EncodingDetector::SHIFT_JIS_INDEX |
| 2386 |
| EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2387 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2388 |
| EncodingDetector::CYRILLIC_WINDOWS_INDEX |
| 2389 |
| EncodingDetector::CYRILLIC_ISO_INDEX |
| 2390 |
| EncodingDetector::CYRILLIC_KOI_INDEX |
| 2391 |
| EncodingDetector::CYRILLIC_IBM_INDEX |
| 2392 |
| EncodingDetector::GREEK_WINDOWS_INDEX |
| 2393 |
| EncodingDetector::GREEK_ISO_INDEX |
| 2394 |
| EncodingDetector::ARABIC_WINDOWS_INDEX |
| 2395 |
| EncodingDetector::ARABIC_ISO_INDEX |
| 2396 |
| EncodingDetector::VISUAL_INDEX |
| 2397 |
| EncodingDetector::LOGICAL_INDEX |
| 2398 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2399 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2400 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2401 |
| EncodingDetector::TURKISH_INDEX => { |
| 2402 |
// XXX Tune this better instead of this kind of absolute. |
| 2403 |
return score; |
| 2404 |
} |
| 2405 |
_ => (50, 60), |
| 2406 |
} |
| 2407 |
} |
| 2408 |
Tld::Simplified |
| 2409 |
| Tld::Traditional |
| 2410 |
| Tld::TraditionalSimplified |
| 2411 |
| Tld::SimplifiedTraditional |
| 2412 |
| Tld::Japanese |
| 2413 |
| Tld::Korean => { |
| 2414 |
// If TLD default is valid, everything else scores zero |
| 2415 |
return score; |
| 2416 |
} |
| 2417 |
Tld::IcelandicFaroese => { |
| 2418 |
match encoding { |
| 2419 |
EncodingDetector::CENTRAL_WINDOWS_INDEX |
| 2420 |
| EncodingDetector::CENTRAL_ISO_INDEX |
| 2421 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2422 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2423 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2424 |
| EncodingDetector::TURKISH_INDEX |
| 2425 |
| EncodingDetector::VIETNAMESE_INDEX => { |
| 2426 |
// XXX Tune this better instead of this kind of absolute. |
| 2427 |
return score; |
| 2428 |
} |
| 2429 |
_ => (50, 60), |
| 2430 |
} |
| 2431 |
} |
| 2432 |
Tld::CentralCyrillic => { |
| 2433 |
match encoding { |
| 2434 |
EncodingDetector::BIG5_INDEX |
| 2435 |
| EncodingDetector::GBK_INDEX |
| 2436 |
| EncodingDetector::EUC_JP_INDEX |
| 2437 |
| EncodingDetector::GREEK_WINDOWS_INDEX |
| 2438 |
| EncodingDetector::GREEK_ISO_INDEX |
| 2439 |
| EncodingDetector::VISUAL_INDEX |
| 2440 |
| EncodingDetector::LOGICAL_INDEX |
| 2441 |
| EncodingDetector::BALTIC_WINDOWS_INDEX |
| 2442 |
| EncodingDetector::BALTIC_ISO4_INDEX |
| 2443 |
| EncodingDetector::BALTIC_ISO13_INDEX |
| 2444 |
| EncodingDetector::TURKISH_INDEX => { |
| 2445 |
// XXX Tune this better instead of this kind of absolute. |
| 2446 |
return score; |
| 2447 |
} |
| 2448 |
_ => (50, 60), |
| 2449 |
} |
| 2450 |
} |
| 2451 |
Tld::Eu => { |
| 2452 |
match encoding { |
| 2453 |
EncodingDetector::BIG5_INDEX |
| 2454 |
| EncodingDetector::GBK_INDEX |
| 2455 |
| EncodingDetector::EUC_JP_INDEX |
| 2456 |
| EncodingDetector::TURKISH_INDEX |
| 2457 |
| EncodingDetector::VIETNAMESE_INDEX => { |
| 2458 |
// XXX Tune this better instead of this kind of absolute. |
| 2459 |
return score; |
| 2460 |
} |
| 2461 |
_ => (50, 60), |
| 2462 |
} |
| 2463 |
} |
| 2464 |
}; |
| 2465 |
(score / divisor) + constant |
| 2466 |
} |
| 2467 |
|
| 2468 |
struct Candidate { |
| 2469 |
inner: InnerCandidate, |
| 2470 |
score: Option<i64>, |
| 2471 |
} |
| 2472 |
|
| 2473 |
impl Candidate { |
| 2474 |
fn feed(&mut self, buffer: &[u8], last: bool) { |
| 2475 |
if let Some(old_score) = self.score { |
| 2476 |
if let Some(new_score) = self.inner.feed(buffer, last) { |
| 2477 |
self.score = Some(old_score + new_score); |
| 2478 |
} else { |
| 2479 |
self.score = None; |
| 2480 |
} |
| 2481 |
} |
| 2482 |
} |
| 2483 |
|
| 2484 |
fn new_latin(data: &'static SingleByteData) -> Self { |
| 2485 |
Candidate { |
| 2486 |
inner: InnerCandidate::Latin(LatinCandidate::new(data)), |
| 2487 |
score: Some(0), |
| 2488 |
} |
| 2489 |
} |
| 2490 |
|
| 2491 |
fn new_non_latin_cased(data: &'static SingleByteData) -> Self { |
| 2492 |
Candidate { |
| 2493 |
inner: InnerCandidate::NonLatinCased(NonLatinCasedCandidate::new(data)), |
| 2494 |
score: Some(0), |
| 2495 |
} |
| 2496 |
} |
| 2497 |
|
| 2498 |
fn new_caseless(data: &'static SingleByteData) -> Self { |
| 2499 |
Candidate { |
| 2500 |
inner: InnerCandidate::Caseless(CaselessCandidate::new(data)), |
| 2501 |
score: Some(0), |
| 2502 |
} |
| 2503 |
} |
| 2504 |
|
| 2505 |
fn new_arabic_french(data: &'static SingleByteData) -> Self { |
| 2506 |
Candidate { |
| 2507 |
inner: InnerCandidate::ArabicFrench(ArabicFrenchCandidate::new(data)), |
| 2508 |
score: Some(0), |
| 2509 |
} |
| 2510 |
} |
| 2511 |
|
| 2512 |
fn new_logical(data: &'static SingleByteData) -> Self { |
| 2513 |
Candidate { |
| 2514 |
inner: InnerCandidate::Logical(LogicalCandidate::new(data)), |
| 2515 |
score: Some(0), |
| 2516 |
} |
| 2517 |
} |
| 2518 |
|
| 2519 |
fn new_visual(data: &'static SingleByteData) -> Self { |
| 2520 |
Candidate { |
| 2521 |
inner: InnerCandidate::Visual(VisualCandidate::new(data)), |
| 2522 |
score: Some(0), |
| 2523 |
} |
| 2524 |
} |
| 2525 |
|
| 2526 |
fn new_utf_8() -> Self { |
| 2527 |
Candidate { |
| 2528 |
inner: InnerCandidate::Utf8(Utf8Candidate { |
| 2529 |
decoder: UTF_8.new_decoder_without_bom_handling(), |
| 2530 |
}), |
| 2531 |
score: Some(0), |
| 2532 |
} |
| 2533 |
} |
| 2534 |
|
| 2535 |
fn new_iso_2022_jp() -> Self { |
| 2536 |
Candidate { |
| 2537 |
inner: InnerCandidate::Iso2022(Iso2022Candidate { |
| 2538 |
decoder: ISO_2022_JP.new_decoder_without_bom_handling(), |
| 2539 |
}), |
| 2540 |
score: Some(0), |
| 2541 |
} |
| 2542 |
} |
| 2543 |
|
| 2544 |
fn new_shift_jis() -> Self { |
| 2545 |
Candidate { |
| 2546 |
inner: InnerCandidate::Shift(ShiftJisCandidate { |
| 2547 |
decoder: SHIFT_JIS.new_decoder_without_bom_handling(), |
| 2548 |
half_width_katakana_seen: false, |
| 2549 |
half_width_katakana_state: HalfWidthKatakana::DakutenForbidden, |
| 2550 |
prev: LatinCj::Other, |
| 2551 |
prev_byte: 0, |
| 2552 |
pending_score: None, |
| 2553 |
}), |
| 2554 |
score: Some(0), |
| 2555 |
} |
| 2556 |
} |
| 2557 |
|
| 2558 |
fn new_euc_jp() -> Self { |
| 2559 |
Candidate { |
| 2560 |
inner: InnerCandidate::EucJp(EucJpCandidate { |
| 2561 |
decoder: EUC_JP.new_decoder_without_bom_handling(), |
| 2562 |
non_ascii_seen: false, |
| 2563 |
half_width_katakana_state: HalfWidthKatakana::DakutenForbidden, |
| 2564 |
prev: LatinCj::Other, |
| 2565 |
prev_byte: 0, |
| 2566 |
prev_prev_byte: 0, |
| 2567 |
}), |
| 2568 |
score: Some(0), |
| 2569 |
} |
| 2570 |
} |
| 2571 |
|
| 2572 |
fn new_euc_kr() -> Self { |
| 2573 |
Candidate { |
| 2574 |
inner: InnerCandidate::EucKr(EucKrCandidate { |
| 2575 |
decoder: EUC_KR.new_decoder_without_bom_handling(), |
| 2576 |
prev_byte: 0, |
| 2577 |
prev_was_euc_range: false, |
| 2578 |
prev: LatinKorean::Other, |
| 2579 |
current_word_len: 0, |
| 2580 |
pending_score: None, |
| 2581 |
}), |
| 2582 |
score: Some(0), |
| 2583 |
} |
| 2584 |
} |
| 2585 |
|
| 2586 |
fn new_big5() -> Self { |
| 2587 |
Candidate { |
| 2588 |
inner: InnerCandidate::Big5(Big5Candidate { |
| 2589 |
decoder: BIG5.new_decoder_without_bom_handling(), |
| 2590 |
prev: LatinCj::Other, |
| 2591 |
prev_byte: 0, |
| 2592 |
pending_score: None, |
| 2593 |
}), |
| 2594 |
score: Some(0), |
| 2595 |
} |
| 2596 |
} |
| 2597 |
|
| 2598 |
fn new_gbk() -> Self { |
| 2599 |
Candidate { |
| 2600 |
inner: InnerCandidate::Gbk(GbkCandidate { |
| 2601 |
decoder: GBK.new_decoder_without_bom_handling(), |
| 2602 |
prev: LatinCj::Other, |
| 2603 |
prev_byte: 0, |
| 2604 |
pending_score: None, |
| 2605 |
}), |
| 2606 |
score: Some(0), |
| 2607 |
} |
| 2608 |
} |
| 2609 |
|
| 2610 |
fn score(&self, encoding: usize, tld: Tld, expectation_is_valid: bool) -> Option<i64> { |
| 2611 |
match &self.inner { |
| 2612 |
InnerCandidate::NonLatinCased(c) => { |
| 2613 |
if c.longest_word < 2 { |
| 2614 |
return None; |
| 2615 |
} |
| 2616 |
} |
| 2617 |
InnerCandidate::Caseless(c) => { |
| 2618 |
if c.longest_word < 2 && !encoding_is_native_to_tld(tld, encoding) { |
| 2619 |
return None; |
| 2620 |
} |
| 2621 |
} |
| 2622 |
InnerCandidate::ArabicFrench(c) => { |
| 2623 |
if c.longest_word < 2 && !encoding_is_native_to_tld(tld, encoding) { |
| 2624 |
return None; |
| 2625 |
} |
| 2626 |
} |
| 2627 |
InnerCandidate::Logical(c) => { |
| 2628 |
if c.longest_word < 2 && !encoding_is_native_to_tld(tld, encoding) { |
| 2629 |
return None; |
| 2630 |
} |
| 2631 |
} |
| 2632 |
InnerCandidate::Visual(c) => { |
| 2633 |
if c.longest_word < 2 && !encoding_is_native_to_tld(tld, encoding) { |
| 2634 |
return None; |
| 2635 |
} |
| 2636 |
} |
| 2637 |
_ => {} |
| 2638 |
} |
| 2639 |
if tld == Tld::Generic { |
| 2640 |
return self.score; |
| 2641 |
} |
| 2642 |
if let Some(score) = self.score { |
| 2643 |
if encoding == encoding_for_tld(tld) { |
| 2644 |
return Some(score + 1); |
| 2645 |
} |
| 2646 |
if encoding_is_native_to_tld(tld, encoding) { |
| 2647 |
return Some(score); |
| 2648 |
} |
| 2649 |
if expectation_is_valid { |
| 2650 |
return Some(score - score_adjustment(score, encoding, tld)); |
| 2651 |
} |
| 2652 |
// If expectation is no longer valid, fall back to |
| 2653 |
// generic behavior. |
| 2654 |
// XXX Flipped Chinese and Central |
| 2655 |
return Some(score); |
| 2656 |
} |
| 2657 |
None |
| 2658 |
} |
| 2659 |
|
| 2660 |
fn plausible_punctuation(&self) -> u64 { |
| 2661 |
match &self.inner { |
| 2662 |
InnerCandidate::Logical(c) => { |
| 2663 |
return c.plausible_punctuation; |
| 2664 |
} |
| 2665 |
InnerCandidate::Visual(c) => { |
| 2666 |
return c.plausible_punctuation; |
| 2667 |
} |
| 2668 |
_ => { |
| 2669 |
unreachable!(); |
| 2670 |
} |
| 2671 |
} |
| 2672 |
} |
| 2673 |
|
| 2674 |
fn encoding(&self) -> &'static Encoding { |
| 2675 |
match &self.inner { |
| 2676 |
InnerCandidate::Latin(c) => { |
| 2677 |
return c.data.encoding; |
| 2678 |
} |
| 2679 |
InnerCandidate::NonLatinCased(c) => { |
| 2680 |
return c.data.encoding; |
| 2681 |
} |
| 2682 |
InnerCandidate::Caseless(c) => { |
| 2683 |
return c.data.encoding; |
| 2684 |
} |
| 2685 |
InnerCandidate::ArabicFrench(c) => { |
| 2686 |
return c.data.encoding; |
| 2687 |
} |
| 2688 |
InnerCandidate::Logical(c) => { |
| 2689 |
return c.data.encoding; |
| 2690 |
} |
| 2691 |
InnerCandidate::Visual(c) => { |
| 2692 |
return c.data.encoding; |
| 2693 |
} |
| 2694 |
InnerCandidate::Shift(_) => { |
| 2695 |
return SHIFT_JIS; |
| 2696 |
} |
| 2697 |
InnerCandidate::EucJp(_) => { |
| 2698 |
return EUC_JP; |
| 2699 |
} |
| 2700 |
InnerCandidate::Big5(_) => { |
| 2701 |
return BIG5; |
| 2702 |
} |
| 2703 |
InnerCandidate::EucKr(_) => { |
| 2704 |
return EUC_KR; |
| 2705 |
} |
| 2706 |
InnerCandidate::Gbk(_) => { |
| 2707 |
return GBK; |
| 2708 |
} |
| 2709 |
InnerCandidate::Utf8(_) => { |
| 2710 |
return UTF_8; |
| 2711 |
} |
| 2712 |
InnerCandidate::Iso2022(_) => { |
| 2713 |
return ISO_2022_JP; |
| 2714 |
} |
| 2715 |
} |
| 2716 |
} |
| 2717 |
} |
| 2718 |
|
| 2719 |
fn count_non_ascii(buffer: &[u8]) -> u64 { |
| 2720 |
let mut count = 0; |
| 2721 |
for &b in buffer { |
| 2722 |
if b >= 0x80 { |
| 2723 |
count += 1; |
| 2724 |
} |
| 2725 |
} |
| 2726 |
count |
| 2727 |
} |
| 2728 |
|
| 2729 |
#[derive(Clone, Copy)] |
| 2730 |
enum BeforeNonAscii { |
| 2731 |
None, |
| 2732 |
One([u8; 1]), |
| 2733 |
Two([u8; 2]), |
| 2734 |
} |
| 2735 |
|
| 2736 |
impl BeforeNonAscii { |
| 2737 |
fn as_slice(&self) -> &[u8] { |
| 2738 |
match self { |
| 2739 |
BeforeNonAscii::None => b"", |
| 2740 |
BeforeNonAscii::One(arr) => &arr[..], |
| 2741 |
BeforeNonAscii::Two(arr) => &arr[..], |
| 2742 |
} |
| 2743 |
} |
| 2744 |
|
| 2745 |
fn push(&mut self, buffer: &[u8]) { |
| 2746 |
let len = buffer.len(); |
| 2747 |
if len >= 2 { |
| 2748 |
let arr = [buffer[len - 2], buffer[len - 1]]; |
| 2749 |
*self = BeforeNonAscii::Two(arr); |
| 2750 |
} else if len == 1 { |
| 2751 |
match self { |
| 2752 |
BeforeNonAscii::None => { |
| 2753 |
let arr = [buffer[0]]; |
| 2754 |
*self = BeforeNonAscii::One(arr); |
| 2755 |
} |
| 2756 |
BeforeNonAscii::One(first) => { |
| 2757 |
let arr = [first[0], buffer[0]]; |
| 2758 |
*self = BeforeNonAscii::Two(arr); |
| 2759 |
} |
| 2760 |
BeforeNonAscii::Two(first) => { |
| 2761 |
let arr = [first[1], buffer[0]]; |
| 2762 |
*self = BeforeNonAscii::Two(arr); |
| 2763 |
} |
| 2764 |
} |
| 2765 |
} |
| 2766 |
} |
| 2767 |
} |
| 2768 |
|
| 2769 |
/// A Web browser-oriented detector for guessing what character |
| 2770 |
/// encoding a stream of bytes is encoded in. |
| 2771 |
/// |
| 2772 |
/// The bytes are fed to the detector incrementally using the `feed` |
| 2773 |
/// method. The current guess of the detector can be queried using |
| 2774 |
/// the `guess` method. The guessing parameters are arguments to the |
| 2775 |
/// `guess` method rather than arguments to the constructor in order |
| 2776 |
/// to enable the application to check if the arguments affect the |
| 2777 |
/// guessing outcome. (The specific use case is to disable UI for |
| 2778 |
/// re-running the detector with UTF-8 allowed and the top-level |
| 2779 |
/// domain name ignored if those arguments don't change the guess.) |
| 2780 |
pub struct EncodingDetector { |
| 2781 |
candidates: [Candidate; 27], |
| 2782 |
non_ascii_seen: u64, |
| 2783 |
// We need to feed up to two bytes of context before non-ASCII |
| 2784 |
// thanks to Spanish n.º. |
| 2785 |
last_before_non_ascii: BeforeNonAscii, |
| 2786 |
esc_seen: bool, |
| 2787 |
closed: bool, |
| 2788 |
} |
| 2789 |
|
| 2790 |
impl EncodingDetector { |
| 2791 |
fn feed_impl(&mut self, buffer: &[u8], last: bool) { |
| 2792 |
for candidate in self.candidates.iter_mut() { |
| 2793 |
candidate.feed(buffer, last); |
| 2794 |
} |
| 2795 |
self.non_ascii_seen += count_non_ascii(buffer); |
| 2796 |
} |
| 2797 |
|
| 2798 |
/// Inform the detector of a chunk of input. |
| 2799 |
/// |
| 2800 |
/// The byte stream is represented as a sequence of calls to this |
| 2801 |
/// method such that the concatenation of the arguments to this |
| 2802 |
/// method form the byte stream. It does not matter how the application |
| 2803 |
/// chooses to chunk the stream. It is OK to call this method with |
| 2804 |
/// a zero-length byte slice. |
| 2805 |
/// |
| 2806 |
/// The end of the stream is indicated by calling this method with |
| 2807 |
/// `last` set to `true`. In that case, the end of the stream is |
| 2808 |
/// considered to occur after the last byte of the `buffer` (which |
| 2809 |
/// may be zero-length) passed in the same call. Once this method |
| 2810 |
/// has been called with `last` set to `true` this method must not |
| 2811 |
/// be called again. |
| 2812 |
/// |
| 2813 |
/// If you want to perform detection on just the prefix of a longer |
| 2814 |
/// stream, do not pass `last=true` after the prefix if the stream |
| 2815 |
/// actually still continues. |
| 2816 |
/// |
| 2817 |
/// Returns `true` if after processing `buffer` the stream has |
| 2818 |
/// contained at least one non-ASCII byte and `false` if only |
| 2819 |
/// ASCII has been seen so far. |
| 2820 |
/// |
| 2821 |
/// # Panics |
| 2822 |
/// |
| 2823 |
/// If this method has previously been called with `last` set to `true`. |
| 2824 |
pub fn feed(&mut self, buffer: &[u8], last: bool) -> bool { |
| 2825 |
assert!( |
| 2826 |
!self.closed, |
| 2827 |
"Must not feed again after feeding with last equaling true." |
| 2828 |
); |
| 2829 |
if last { |
| 2830 |
self.closed = true; |
| 2831 |
} |
| 2832 |
let start = if self.non_ascii_seen == 0 && !self.esc_seen { |
| 2833 |
let up_to = Encoding::ascii_valid_up_to(buffer); |
| 2834 |
let start = if let Some(escape) = memchr::memchr(0x1B, &buffer[..up_to]) { |
| 2835 |
self.esc_seen = true; |
| 2836 |
escape |
| 2837 |
} else { |
| 2838 |
up_to |
| 2839 |
}; |
| 2840 |
if start == buffer.len() { |
| 2841 |
self.last_before_non_ascii.push(buffer); |
| 2842 |
return self.non_ascii_seen != 0; |
| 2843 |
} |
| 2844 |
if start == 0 || start == 1 { |
| 2845 |
let last_before = self.last_before_non_ascii; |
| 2846 |
self.last_before_non_ascii = BeforeNonAscii::None; |
| 2847 |
self.feed_impl(last_before.as_slice(), false); |
| 2848 |
0 |
| 2849 |
} else { |
| 2850 |
start - 2 |
| 2851 |
} |
| 2852 |
} else { |
| 2853 |
0 |
| 2854 |
}; |
| 2855 |
self.feed_impl(&buffer[start..], last); |
| 2856 |
self.non_ascii_seen != 0 |
| 2857 |
} |
| 2858 |
|
| 2859 |
/// Guess the encoding given the bytes pushed to the detector so far |
| 2860 |
/// (via `feed()`), the top-level domain name from which the bytes were |
| 2861 |
/// loaded, and an indication of whether to consider UTF-8 as a permissible |
| 2862 |
/// guess. |
| 2863 |
/// |
| 2864 |
/// The `tld` argument takes the rightmost DNS label of the hostname of the |
| 2865 |
/// host the stream was loaded from in lower-case ASCII form. That is, if |
| 2866 |
/// the label is an internationalized top-level domain name, it must be |
| 2867 |
/// provided in its Punycode form. If the TLD that the stream was loaded |
| 2868 |
/// from is unavalable, `None` may be passed instead, which is equivalent |
| 2869 |
/// to passing `Some(b"com")`. |
| 2870 |
/// |
| 2871 |
/// If the `allow_utf8` argument is set to `false`, the return value of |
| 2872 |
/// this method won't be `encoding_rs::UTF_8`. When performing detection |
| 2873 |
/// on `text/html` on non-`file:` URLs, Web browsers must pass `false`, |
| 2874 |
/// unless the user has taken a specific contextual action to request an |
| 2875 |
/// override. This way, Web developers cannot start depending on UTF-8 |
| 2876 |
/// detection. Such reliance would make the Web Platform more brittle. |
| 2877 |
/// |
| 2878 |
/// Returns the guessed encoding. |
| 2879 |
/// |
| 2880 |
/// # Panics |
| 2881 |
/// |
| 2882 |
/// If `tld` contains non-ASCII, period, or upper-case letters. (The panic |
| 2883 |
/// condition is intentionally limited to signs of failing to extract the |
| 2884 |
/// label correctly, failing to provide it in its Punycode form, and failure |
| 2885 |
/// to lower-case it. Full DNS label validation is intentionally not performed |
| 2886 |
/// to avoid panics when the reality doesn't match the specs.) |
| 2887 |
pub fn guess(&self, tld: Option<&[u8]>, allow_utf8: bool) -> &'static Encoding { |
| 2888 |
let mut tld_type = tld.map_or(Tld::Generic, |tld| { |
| 2889 |
assert!(!contains_upper_case_period_or_non_ascii(tld)); |
| 2890 |
classify_tld(tld) |
| 2891 |
}); |
| 2892 |
|
| 2893 |
if self.non_ascii_seen == 0 |
| 2894 |
&& self.esc_seen |
| 2895 |
&& self.candidates[Self::ISO_2022_JP_INDEX].score.is_some() |
| 2896 |
{ |
| 2897 |
return ISO_2022_JP; |
| 2898 |
} |
| 2899 |
|
| 2900 |
if self.candidates[Self::UTF_8_INDEX].score.is_some() { |
| 2901 |
if allow_utf8 { |
| 2902 |
return UTF_8; |
| 2903 |
} |
| 2904 |
// Various test cases that prohibit UTF-8 detection want to |
| 2905 |
// see windows-1252 specifically. These tests run on generic |
| 2906 |
// domains. However, if we returned windows-1252 on |
| 2907 |
// some non-generic domains, we'd cause reloads. |
| 2908 |
return self.candidates[encoding_for_tld(tld_type)].encoding(); |
| 2909 |
} |
| 2910 |
|
| 2911 |
let mut encoding = self.candidates[encoding_for_tld(tld_type)].encoding(); |
| 2912 |
let mut max = 0i64; |
| 2913 |
let mut expectation_is_valid = false; |
| 2914 |
if tld_type != Tld::Generic { |
| 2915 |
for (i, candidate) in self.candidates.iter().enumerate().skip(Self::FIRST_NORMAL) { |
| 2916 |
if encoding_is_native_to_tld(tld_type, i) && candidate.score.is_some() { |
| 2917 |
expectation_is_valid = true; |
| 2918 |
break; |
| 2919 |
} |
| 2920 |
} |
| 2921 |
} |
| 2922 |
if !expectation_is_valid { |
| 2923 |
// Flip Chinese and Central around |
| 2924 |
match tld_type { |
| 2925 |
Tld::Simplified => { |
| 2926 |
if self.candidates[Self::BIG5_INDEX].score.is_some() { |
| 2927 |
tld_type = Tld::Traditional; |
| 2928 |
expectation_is_valid = true; |
| 2929 |
} |
| 2930 |
} |
| 2931 |
Tld::Traditional => { |
| 2932 |
if self.candidates[Self::GBK_INDEX].score.is_some() { |
| 2933 |
tld_type = Tld::Simplified; |
| 2934 |
expectation_is_valid = true; |
| 2935 |
} |
| 2936 |
} |
| 2937 |
Tld::CentralWindows => { |
| 2938 |
if self.candidates[Self::CENTRAL_ISO_INDEX].score.is_some() { |
| 2939 |
tld_type = Tld::CentralIso; |
| 2940 |
expectation_is_valid = true; |
| 2941 |
} |
| 2942 |
} |
| 2943 |
Tld::CentralIso => { |
| 2944 |
if self.candidates[Self::CENTRAL_WINDOWS_INDEX].score.is_some() { |
| 2945 |
tld_type = Tld::CentralWindows; |
| 2946 |
expectation_is_valid = true; |
| 2947 |
} |
| 2948 |
} |
| 2949 |
_ => {} |
| 2950 |
} |
| 2951 |
} |
| 2952 |
for (i, candidate) in self.candidates.iter().enumerate().skip(Self::FIRST_NORMAL) { |
| 2953 |
if let Some(score) = candidate.score(i, tld_type, expectation_is_valid) { |
| 2954 |
if score > max { |
| 2955 |
max = score; |
| 2956 |
encoding = candidate.encoding(); |
| 2957 |
} |
| 2958 |
} |
| 2959 |
} |
| 2960 |
let visual = &self.candidates[Self::VISUAL_INDEX]; |
| 2961 |
if let Some(visual_score) = visual.score(Self::VISUAL_INDEX, tld_type, expectation_is_valid) |
| 2962 |
{ |
| 2963 |
if (visual_score > max || encoding == WINDOWS_1255) |
| 2964 |
&& visual.plausible_punctuation() |
| 2965 |
> self.candidates[Self::LOGICAL_INDEX].plausible_punctuation() |
| 2966 |
{ |
| 2967 |
// max = visual_score; |
| 2968 |
encoding = ISO_8859_8; |
| 2969 |
} |
| 2970 |
} |
| 2971 |
|
| 2972 |
encoding |
| 2973 |
} |
| 2974 |
|
| 2975 |
// XXX Test-only API |
| 2976 |
#[cfg(feature = "testing-only-no-semver-guarantees-do-not-use")] |
| 2977 |
pub fn find_score(&self, encoding: &'static Encoding) -> Option<i64> { |
| 2978 |
let mut tld_type = Tld::Generic; |
| 2979 |
let mut expectation_is_valid = false; |
| 2980 |
if tld_type != Tld::Generic { |
| 2981 |
for (i, candidate) in self.candidates.iter().enumerate().skip(Self::FIRST_NORMAL) { |
| 2982 |
if encoding_is_native_to_tld(tld_type, i) && candidate.score.is_some() { |
| 2983 |
expectation_is_valid = true; |
| 2984 |
break; |
| 2985 |
} |
| 2986 |
} |
| 2987 |
} |
| 2988 |
if !expectation_is_valid { |
| 2989 |
// Flip Chinese and Central around |
| 2990 |
match tld_type { |
| 2991 |
Tld::Simplified => { |
| 2992 |
if self.candidates[Self::BIG5_INDEX].score.is_some() { |
| 2993 |
tld_type = Tld::Traditional; |
| 2994 |
expectation_is_valid = true; |
| 2995 |
} |
| 2996 |
} |
| 2997 |
Tld::Traditional => { |
| 2998 |
if self.candidates[Self::GBK_INDEX].score.is_some() { |
| 2999 |
tld_type = Tld::Simplified; |
| 3000 |
expectation_is_valid = true; |
| 3001 |
} |
| 3002 |
} |
| 3003 |
Tld::CentralWindows => { |
| 3004 |
if self.candidates[Self::CENTRAL_ISO_INDEX].score.is_some() { |
| 3005 |
tld_type = Tld::CentralIso; |
| 3006 |
expectation_is_valid = true; |
| 3007 |
} |
| 3008 |
} |
| 3009 |
Tld::CentralIso => { |
| 3010 |
if self.candidates[Self::CENTRAL_WINDOWS_INDEX].score.is_some() { |
| 3011 |
tld_type = Tld::CentralWindows; |
| 3012 |
expectation_is_valid = true; |
| 3013 |
} |
| 3014 |
} |
| 3015 |
_ => {} |
| 3016 |
} |
| 3017 |
} |
| 3018 |
for (i, candidate) in self.candidates.iter().enumerate() { |
| 3019 |
if encoding == candidate.encoding() { |
| 3020 |
return candidate.score(i, tld_type, expectation_is_valid); |
| 3021 |
} |
| 3022 |
} |
| 3023 |
Some(0) |
| 3024 |
} |
| 3025 |
|
| 3026 |
const FIRST_NORMAL: usize = 3; |
| 3027 |
|
| 3028 |
const UTF_8_INDEX: usize = 0; |
| 3029 |
|
| 3030 |
const ISO_2022_JP_INDEX: usize = 1; |
| 3031 |
|
| 3032 |
const VISUAL_INDEX: usize = 2; |
| 3033 |
|
| 3034 |
const GBK_INDEX: usize = 3; |
| 3035 |
|
| 3036 |
const EUC_JP_INDEX: usize = 4; |
| 3037 |
|
| 3038 |
const EUC_KR_INDEX: usize = 5; |
| 3039 |
|
| 3040 |
const SHIFT_JIS_INDEX: usize = 6; |
| 3041 |
|
| 3042 |
const BIG5_INDEX: usize = 7; |
| 3043 |
|
| 3044 |
const WESTERN_INDEX: usize = 8; |
| 3045 |
|
| 3046 |
const CYRILLIC_WINDOWS_INDEX: usize = 9; |
| 3047 |
|
| 3048 |
const CENTRAL_WINDOWS_INDEX: usize = 10; |
| 3049 |
|
| 3050 |
const CENTRAL_ISO_INDEX: usize = 11; |
| 3051 |
|
| 3052 |
const ARABIC_WINDOWS_INDEX: usize = 12; |
| 3053 |
|
| 3054 |
const ICELANDIC_INDEX: usize = 13; |
| 3055 |
|
| 3056 |
const TURKISH_INDEX: usize = 14; |
| 3057 |
|
| 3058 |
const THAI_INDEX: usize = 15; |
| 3059 |
|
| 3060 |
const LOGICAL_INDEX: usize = 16; |
| 3061 |
|
| 3062 |
const GREEK_WINDOWS_INDEX: usize = 17; |
| 3063 |
|
| 3064 |
const GREEK_ISO_INDEX: usize = 18; |
| 3065 |
|
| 3066 |
const BALTIC_WINDOWS_INDEX: usize = 19; |
| 3067 |
|
| 3068 |
const BALTIC_ISO13_INDEX: usize = 20; |
| 3069 |
|
| 3070 |
const CYRILLIC_KOI_INDEX: usize = 21; |
| 3071 |
|
| 3072 |
const CYRILLIC_IBM_INDEX: usize = 22; |
| 3073 |
|
| 3074 |
const ARABIC_ISO_INDEX: usize = 23; |
| 3075 |
|
| 3076 |
const VIETNAMESE_INDEX: usize = 24; |
| 3077 |
|
| 3078 |
const BALTIC_ISO4_INDEX: usize = 25; |
| 3079 |
|
| 3080 |
const CYRILLIC_ISO_INDEX: usize = 26; |
| 3081 |
|
| 3082 |
/// Creates a new instance of the detector. |
| 3083 |
pub fn new() -> Self { |
| 3084 |
EncodingDetector { |
| 3085 |
candidates: [ |
| 3086 |
Candidate::new_utf_8(), // 0 |
| 3087 |
Candidate::new_iso_2022_jp(), // 1 |
| 3088 |
Candidate::new_visual(&SINGLE_BYTE_DATA[ISO_8859_8_INDEX]), // 2 |
| 3089 |
Candidate::new_gbk(), // 3 |
| 3090 |
Candidate::new_euc_jp(), // 4 |
| 3091 |
Candidate::new_euc_kr(), // 5 |
| 3092 |
Candidate::new_shift_jis(), // 6 |
| 3093 |
Candidate::new_big5(), // 7 |
| 3094 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1252_INDEX]), // 8 |
| 3095 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[WINDOWS_1251_INDEX]), // 9 |
| 3096 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1250_INDEX]), // 10 |
| 3097 |
Candidate::new_latin(&SINGLE_BYTE_DATA[ISO_8859_2_INDEX]), // 11 |
| 3098 |
Candidate::new_arabic_french(&SINGLE_BYTE_DATA[WINDOWS_1256_INDEX]), // 12 |
| 3099 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1252_ICELANDIC_INDEX]), // 13 |
| 3100 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1254_INDEX]), // 14 |
| 3101 |
Candidate::new_caseless(&SINGLE_BYTE_DATA[WINDOWS_874_INDEX]), // 15 |
| 3102 |
Candidate::new_logical(&SINGLE_BYTE_DATA[WINDOWS_1255_INDEX]), // 16 |
| 3103 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[WINDOWS_1253_INDEX]), // 17 |
| 3104 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[ISO_8859_7_INDEX]), // 18 |
| 3105 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1257_INDEX]), // 19 |
| 3106 |
Candidate::new_latin(&SINGLE_BYTE_DATA[ISO_8859_13_INDEX]), // 20 |
| 3107 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[KOI8_U_INDEX]), // 21 |
| 3108 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[IBM866_INDEX]), // 22 |
| 3109 |
Candidate::new_caseless(&SINGLE_BYTE_DATA[ISO_8859_6_INDEX]), // 23 |
| 3110 |
Candidate::new_latin(&SINGLE_BYTE_DATA[WINDOWS_1258_INDEX]), // 24 |
| 3111 |
Candidate::new_latin(&SINGLE_BYTE_DATA[ISO_8859_4_INDEX]), // 25 |
| 3112 |
Candidate::new_non_latin_cased(&SINGLE_BYTE_DATA[ISO_8859_5_INDEX]), // 26 |
| 3113 |
], |
| 3114 |
non_ascii_seen: 0, |
| 3115 |
last_before_non_ascii: BeforeNonAscii::None, |
| 3116 |
esc_seen: false, |
| 3117 |
closed: false, |
| 3118 |
} |
| 3119 |
} |
| 3120 |
|
| 3121 |
/// Queries whether the TLD is considered non-generic and could affect the guess. |
| 3122 |
pub fn tld_may_affect_guess(tld: Option<&[u8]>) -> bool { |
| 3123 |
if let Some(tld) = tld { |
| 3124 |
classify_tld(tld) != Tld::Generic |
| 3125 |
} else { |
| 3126 |
false |
| 3127 |
} |
| 3128 |
} |
| 3129 |
} |
| 3130 |
|
| 3131 |
#[cfg(test)] |
| 3132 |
mod tests { |
| 3133 |
use super::*; |
| 3134 |
use detone::IterDecomposeVietnamese; |
| 3135 |
use encoding_rs::IBM866; |
| 3136 |
use encoding_rs::ISO_8859_2; |
| 3137 |
use encoding_rs::ISO_8859_4; |
| 3138 |
use encoding_rs::ISO_8859_5; |
| 3139 |
use encoding_rs::ISO_8859_6; |
| 3140 |
use encoding_rs::ISO_8859_7; |
| 3141 |
use encoding_rs::KOI8_U; |
| 3142 |
use encoding_rs::WINDOWS_1250; |
| 3143 |
use encoding_rs::WINDOWS_1251; |
| 3144 |
use encoding_rs::WINDOWS_1252; |
| 3145 |
use encoding_rs::WINDOWS_1253; |
| 3146 |
use encoding_rs::WINDOWS_1254; |
| 3147 |
use encoding_rs::WINDOWS_1256; |
| 3148 |
use encoding_rs::WINDOWS_1257; |
| 3149 |
use encoding_rs::WINDOWS_1258; |
| 3150 |
use encoding_rs::WINDOWS_874; |
| 3151 |
|
| 3152 |
fn check_bytes(bytes: &[u8], encoding: &'static Encoding) { |
| 3153 |
let mut det = EncodingDetector::new(); |
| 3154 |
det.feed(bytes, true); |
| 3155 |
let enc = det.guess(None, false); |
| 3156 |
let (decoded, _) = enc.decode_without_bom_handling(bytes); |
| 3157 |
println!("{:?}", decoded); |
| 3158 |
assert_eq!(enc, encoding); |
| 3159 |
} |
| 3160 |
|
| 3161 |
fn check(input: &str, encoding: &'static Encoding) { |
| 3162 |
let orthographic; |
| 3163 |
let (bytes, _, _) = if encoding == WINDOWS_1258 { |
| 3164 |
orthographic = input |
| 3165 |
.chars() |
| 3166 |
.decompose_vietnamese_tones(true) |
| 3167 |
.collect::<String>(); |
| 3168 |
encoding.encode(&orthographic) |
| 3169 |
} else { |
| 3170 |
encoding.encode(input) |
| 3171 |
}; |
| 3172 |
check_bytes(&bytes, encoding); |
| 3173 |
} |
| 3174 |
|
| 3175 |
#[test] |
| 3176 |
fn test_i_apostrophe() { |
| 3177 |
let mut det = EncodingDetector::new(); |
| 3178 |
det.feed(b"I\x92", true); |
| 3179 |
let enc = det.guess(None, false); |
| 3180 |
assert_eq!(enc, WINDOWS_1252); |
| 3181 |
} |
| 3182 |
|
| 3183 |
#[test] |
| 3184 |
fn test_streaming_numero_one_by_one() { |
| 3185 |
let mut det = EncodingDetector::new(); |
| 3186 |
det.feed(b"n", false); |
| 3187 |
det.feed(b".", false); |
| 3188 |
det.feed(b"\xBA", false); |
| 3189 |
det.feed(b"1", true); |
| 3190 |
let enc = det.guess(None, false); |
| 3191 |
assert_eq!(enc, WINDOWS_1252); |
| 3192 |
} |
| 3193 |
|
| 3194 |
#[test] |
| 3195 |
fn test_streaming_numero_two_together() { |
| 3196 |
let mut det = EncodingDetector::new(); |
| 3197 |
det.feed(b"n.", false); |
| 3198 |
det.feed(b"\xBA", false); |
| 3199 |
det.feed(b"1", true); |
| 3200 |
let enc = det.guess(None, false); |
| 3201 |
assert_eq!(enc, WINDOWS_1252); |
| 3202 |
} |
| 3203 |
|
| 3204 |
#[test] |
| 3205 |
fn test_streaming_numero_one_by_one_extra_before() { |
| 3206 |
let mut det = EncodingDetector::new(); |
| 3207 |
det.feed(b" n", false); |
| 3208 |
det.feed(b".", false); |
| 3209 |
det.feed(b"\xBA", false); |
| 3210 |
det.feed(b"1", true); |
| 3211 |
let enc = det.guess(None, false); |
| 3212 |
assert_eq!(enc, WINDOWS_1252); |
| 3213 |
} |
| 3214 |
|
| 3215 |
#[test] |
| 3216 |
fn test_streaming_numero_one_before() { |
| 3217 |
let mut det = EncodingDetector::new(); |
| 3218 |
det.feed(b"n", false); |
| 3219 |
det.feed(b".\xBA", false); |
| 3220 |
det.feed(b"1", true); |
| 3221 |
let enc = det.guess(None, false); |
| 3222 |
assert_eq!(enc, WINDOWS_1252); |
| 3223 |
} |
| 3224 |
|
| 3225 |
#[test] |
| 3226 |
fn test_streaming_numero_longer_first_buffer() { |
| 3227 |
let mut det = EncodingDetector::new(); |
| 3228 |
det.feed(b"rrn.", false); |
| 3229 |
det.feed(b"\xBA", false); |
| 3230 |
det.feed(b"1", true); |
| 3231 |
let enc = det.guess(None, false); |
| 3232 |
assert_eq!(enc, WINDOWS_1252); |
| 3233 |
} |
| 3234 |
|
| 3235 |
#[test] |
| 3236 |
fn test_empty() { |
| 3237 |
let mut det = EncodingDetector::new(); |
| 3238 |
let seen_non_ascii = det.feed(b"", true); |
| 3239 |
let enc = det.guess(None, false); |
| 3240 |
assert_eq!(enc, WINDOWS_1252); |
| 3241 |
assert!(!seen_non_ascii); |
| 3242 |
} |
| 3243 |
|
| 3244 |
#[test] |
| 3245 |
fn test_fi() { |
| 3246 |
check("Ääni", WINDOWS_1252); |
| 3247 |
} |
| 3248 |
|
| 3249 |
#[test] |
| 3250 |
fn test_fi_bis() { |
| 3251 |
check("Tämä", WINDOWS_1252); |
| 3252 |
} |
| 3253 |
|
| 3254 |
#[test] |
| 3255 |
fn test_pt() { |
| 3256 |
check( |
| 3257 |
"Este é um teste de codificação de caracteres.", |
| 3258 |
WINDOWS_1252, |
| 3259 |
); |
| 3260 |
} |
| 3261 |
|
| 3262 |
#[test] |
| 3263 |
fn test_is() { |
| 3264 |
check("Þetta er kóðunarpróf á staf. Fyrir sum tungumál sem nota latneska stafi þurfum við meira inntak til að taka ákvörðunina.", WINDOWS_1252); |
| 3265 |
} |
| 3266 |
|
| 3267 |
#[test] |
| 3268 |
fn test_ru_short() { |
| 3269 |
check("Русский", WINDOWS_1251); |
| 3270 |
} |
| 3271 |
|
| 3272 |
#[test] |
| 3273 |
fn test_ru() { |
| 3274 |
check("Это тест кодировки символов.", WINDOWS_1251); |
| 3275 |
} |
| 3276 |
|
| 3277 |
#[test] |
| 3278 |
fn test_ru_iso() { |
| 3279 |
check("Это тест кодировки символов.", ISO_8859_5); |
| 3280 |
} |
| 3281 |
|
| 3282 |
#[test] |
| 3283 |
fn test_ru_ibm() { |
| 3284 |
check("Это тест кодировки символов.", IBM866); |
| 3285 |
} |
| 3286 |
|
| 3287 |
#[test] |
| 3288 |
fn test_ru_koi() { |
| 3289 |
check("Это тест кодировки символов.", KOI8_U); |
| 3290 |
} |
| 3291 |
|
| 3292 |
#[test] |
| 3293 |
fn test_uk() { |
| 3294 |
check("Це тест на кодування символів.", WINDOWS_1251); |
| 3295 |
} |
| 3296 |
|
| 3297 |
#[test] |
| 3298 |
fn test_uk_koi() { |
| 3299 |
check("Це тест на кодування символів.", KOI8_U); |
| 3300 |
} |
| 3301 |
|
| 3302 |
#[test] |
| 3303 |
fn test_el_short() { |
| 3304 |
check("Ελληνικά", WINDOWS_1253); |
| 3305 |
} |
| 3306 |
|
| 3307 |
#[test] |
| 3308 |
fn test_el() { |
| 3309 |
check( |
| 3310 |
"Πρόκειται για δοκιμή κωδικοποίησης χαρακτήρων: Άρης", |
| 3311 |
WINDOWS_1253, |
| 3312 |
); |
| 3313 |
} |
| 3314 |
|
| 3315 |
#[test] |
| 3316 |
fn test_el_iso() { |
| 3317 |
check( |
| 3318 |
"Πρόκειται για δοκιμή κωδικοποίησης χαρακτήρων: Άρης", |
| 3319 |
ISO_8859_7, |
| 3320 |
); |
| 3321 |
} |
| 3322 |
|
| 3323 |
#[test] |
| 3324 |
fn test_de() { |
| 3325 |
check("Straße", WINDOWS_1252); |
| 3326 |
} |
| 3327 |
|
| 3328 |
#[test] |
| 3329 |
fn test_he() { |
| 3330 |
check("\u{5E2}\u{5D1}\u{5E8}\u{5D9}\u{5EA}", WINDOWS_1255); |
| 3331 |
} |
| 3332 |
|
| 3333 |
#[test] |
| 3334 |
fn test_2022() { |
| 3335 |
check("日本語", ISO_2022_JP); |
| 3336 |
} |
| 3337 |
|
| 3338 |
#[test] |
| 3339 |
fn test_th() { |
| 3340 |
check("นี่คือการทดสอบการเข้ารหัสอักขระ", WINDOWS_874); |
| 3341 |
} |
| 3342 |
|
| 3343 |
#[test] |
| 3344 |
fn test_vi() { |
| 3345 |
check("Đây là một thử nghiệm mã hóa ký tự.", WINDOWS_1258); |
| 3346 |
} |
| 3347 |
|
| 3348 |
#[test] |
| 3349 |
fn test_tr() { |
| 3350 |
check("Bu bir karakter kodlama testidir. Latince karakterleri kullanan bazı dillerde karar vermek için daha fazla girdiye ihtiyacımız var.", WINDOWS_1254); |
| 3351 |
} |
| 3352 |
|
| 3353 |
#[test] |
| 3354 |
fn test_simplified() { |
| 3355 |
check("这是一个字符编码测试。", GBK); |
| 3356 |
} |
| 3357 |
|
| 3358 |
#[test] |
| 3359 |
fn test_traditional() { |
| 3360 |
check("這是一個字符編碼測試。", BIG5); |
| 3361 |
} |
| 3362 |
|
| 3363 |
#[test] |
| 3364 |
fn test_ko() { |
| 3365 |
check("이것은 문자 인코딩 테스트입니다.", EUC_KR); |
| 3366 |
} |
| 3367 |
|
| 3368 |
#[test] |
| 3369 |
fn test_shift() { |
| 3370 |
check("これは文字実験です。", SHIFT_JIS); |
| 3371 |
} |
| 3372 |
|
| 3373 |
#[test] |
| 3374 |
fn test_euc() { |
| 3375 |
check("これは文字実験です。", EUC_JP); |
| 3376 |
} |
| 3377 |
|
| 3378 |
#[test] |
| 3379 |
fn test_ar() { |
| 3380 |
check("هذا هو اختبار ترميز الأحرف.", WINDOWS_1256); |
| 3381 |
} |
| 3382 |
|
| 3383 |
#[test] |
| 3384 |
fn test_ar_iso() { |
| 3385 |
check("هذا هو اختبار ترميز الأحرف.", ISO_8859_6); |
| 3386 |
} |
| 3387 |
|
| 3388 |
#[test] |
| 3389 |
fn test_fa() { |
| 3390 |
check("این یک تست رمزگذاری کاراکتر است.", WINDOWS_1256); |
| 3391 |
} |
| 3392 |
|
| 3393 |
#[test] |
| 3394 |
fn test_visual() { |
| 3395 |
check(".םיוות דודיק ןחבמ והז", ISO_8859_8); |
| 3396 |
} |
| 3397 |
|
| 3398 |
#[test] |
| 3399 |
fn test_yi() { |
| 3400 |
check("דאָס איז אַ טעסט פֿאַר קאָדירונג פון כאַראַקטער.", WINDOWS_1255); |
| 3401 |
} |
| 3402 |
|
| 3403 |
#[test] |
| 3404 |
fn test_it() { |
| 3405 |
check("è", WINDOWS_1252); |
| 3406 |
} |
| 3407 |
|
| 3408 |
#[test] |
| 3409 |
fn test_en() { |
| 3410 |
check("isn’t", WINDOWS_1252); |
| 3411 |
} |
| 3412 |
|
| 3413 |
#[test] |
| 3414 |
fn test_en_bis() { |
| 3415 |
check("Rock ’n Roll", WINDOWS_1252); |
| 3416 |
} |
| 3417 |
|
| 3418 |
#[test] |
| 3419 |
fn test_ca() { |
| 3420 |
check("Codificació de caràcters", WINDOWS_1252); |
| 3421 |
} |
| 3422 |
|
| 3423 |
#[test] |
| 3424 |
fn test_et() { |
| 3425 |
check("või", WINDOWS_1252); |
| 3426 |
} |
| 3427 |
|
| 3428 |
#[test] |
| 3429 |
fn test_pl_iso() { |
| 3430 |
check("To jest test kodowania znaków. W przypadku niektórych języków, które używają znaków łacińskich, potrzebujemy więcej danych, aby podjąć decyzję.", ISO_8859_2); |
| 3431 |
} |
| 3432 |
|
| 3433 |
#[test] |
| 3434 |
fn test_pl() { |
| 3435 |
check("To jest test kodowania znaków. W przypadku niektórych języków, które używają znaków łacińskich, potrzebujemy więcej danych, aby podjąć decyzję.", WINDOWS_1250); |
| 3436 |
} |
| 3437 |
|
| 3438 |
#[test] |
| 3439 |
fn test_lt() { |
| 3440 |
check("Tai simbolių kodavimo testas. Kai kurioms kalboms, naudojančioms lotyniškus rašmenis, mums reikia daugiau informacijos, kad galėtume priimti sprendimą.", WINDOWS_1257); |
| 3441 |
} |
| 3442 |
|
| 3443 |
// TODO: Detected as ISO-8859-2. |
| 3444 |
// #[test] |
| 3445 |
// fn test_lt_windows_iso_8859_4() { |
| 3446 |
// check("Tai simbolių kodavimo testas. Kai kurioms kalboms, naudojančioms lotyniškus rašmenis, mums reikia daugiau informacijos, kad galėtume priimti sprendimą.", ISO_8859_4); |
| 3447 |
// } |
| 3448 |
|
| 3449 |
#[test] |
| 3450 |
fn test_lv() { |
| 3451 |
check("Šis ir rakstzīmju kodēšanas tests. Dažās valodās, kurās tiek izmantotas latīņu valodas burti, lēmuma pieņemšanai mums ir nepieciešams vairāk ieguldījuma.", WINDOWS_1257); |
| 3452 |
} |
| 3453 |
|
| 3454 |
#[test] |
| 3455 |
fn test_lv_iso_8859_4() { |
| 3456 |
check("Šis ir rakstzīmju kodēšanas tests. Dažās valodās, kurās tiek izmantotas latīņu valodas burti, lēmuma pieņemšanai mums ir nepieciešams vairāk ieguldījuma.", ISO_8859_4); |
| 3457 |
} |
| 3458 |
|
| 3459 |
#[test] |
| 3460 |
fn test_a0() { |
| 3461 |
// Test that this isn't IBM866. TODO: What about GBK with fully paired 0xA0? |
| 3462 |
check("\u{A0}\u{A0} \u{A0}", WINDOWS_1252); |
| 3463 |
} |
| 3464 |
|
| 3465 |
#[test] |
| 3466 |
fn test_a0a0() { |
| 3467 |
// Test that this isn't GBK or EUC-KR. |
| 3468 |
check("\u{A0}\u{A0}", WINDOWS_1252); |
| 3469 |
} |
| 3470 |
|
| 3471 |
#[test] |
| 3472 |
fn test_space_copyright_space() { |
| 3473 |
check(" © ", WINDOWS_1252); |
| 3474 |
} |
| 3475 |
|
| 3476 |
#[test] |
| 3477 |
fn test_space_masculine_space() { |
| 3478 |
check(" º ", WINDOWS_1252); |
| 3479 |
} |
| 3480 |
|
| 3481 |
#[test] |
| 3482 |
fn test_space_feminine_space() { |
| 3483 |
check(" ª ", WINDOWS_1252); |
| 3484 |
} |
| 3485 |
|
| 3486 |
#[test] |
| 3487 |
fn test_period_masculine_space() { |
| 3488 |
check(".º ", WINDOWS_1252); |
| 3489 |
} |
| 3490 |
|
| 3491 |
#[test] |
| 3492 |
fn test_period_feminine_space() { |
| 3493 |
check(".ª ", WINDOWS_1252); |
| 3494 |
} |
| 3495 |
|
| 3496 |
#[test] |
| 3497 |
fn test_maria() { |
| 3498 |
check(" Mª ", WINDOWS_1252); |
| 3499 |
} |
| 3500 |
|
| 3501 |
#[test] |
| 3502 |
fn test_dona() { |
| 3503 |
check(" Dª ", WINDOWS_1252); |
| 3504 |
} |
| 3505 |
|
| 3506 |
#[test] |
| 3507 |
fn test_nuestra() { |
| 3508 |
check(" Nª ", WINDOWS_1252); |
| 3509 |
} |
| 3510 |
|
| 3511 |
#[test] |
| 3512 |
fn test_senora() { |
| 3513 |
check(" Sª ", WINDOWS_1252); |
| 3514 |
} |
| 3515 |
|
| 3516 |
#[test] |
| 3517 |
fn test_digit_feminine() { |
| 3518 |
check(" 42ª ", WINDOWS_1252); |
| 3519 |
} |
| 3520 |
|
| 3521 |
#[test] |
| 3522 |
fn test_digit_masculine() { |
| 3523 |
check(" 42º ", WINDOWS_1252); |
| 3524 |
} |
| 3525 |
|
| 3526 |
#[test] |
| 3527 |
fn test_roman_feminine() { |
| 3528 |
check(" XIVª ", WINDOWS_1252); |
| 3529 |
} |
| 3530 |
|
| 3531 |
#[test] |
| 3532 |
fn test_roman_masculine() { |
| 3533 |
check(" XIVº ", WINDOWS_1252); |
| 3534 |
} |
| 3535 |
|
| 3536 |
#[test] |
| 3537 |
fn test_numero_uno() { |
| 3538 |
check("Nº1", WINDOWS_1252); |
| 3539 |
} |
| 3540 |
|
| 3541 |
#[test] |
| 3542 |
fn test_numero() { |
| 3543 |
check("Nº", WINDOWS_1252); |
| 3544 |
} |
| 3545 |
|
| 3546 |
#[test] |
| 3547 |
fn test_euro() { |
| 3548 |
check(" €9", WINDOWS_1252); |
| 3549 |
} |
| 3550 |
|
| 3551 |
#[test] |
| 3552 |
fn test_shift_jis_half_width_katakana() { |
| 3553 |
check("ハードウェアハードウェアハードウェアハードウェアハードウェア", SHIFT_JIS); |
| 3554 |
} |
| 3555 |
|
| 3556 |
#[test] |
| 3557 |
fn test_big5_pua() { |
| 3558 |
let mut v = Vec::new(); |
| 3559 |
for _ in 0..40 { |
| 3560 |
v.extend_from_slice(b"\xA4\x40"); |
| 3561 |
} |
| 3562 |
v.extend_from_slice(b"\x81\x40\xA4\x40"); |
| 3563 |
check_bytes(&v, BIG5); |
| 3564 |
} |
| 3565 |
|
| 3566 |
#[test] |
| 3567 |
fn test_big5_single_byte_a0() { |
| 3568 |
let mut v = Vec::new(); |
| 3569 |
for _ in 0..80 { |
| 3570 |
v.extend_from_slice(b"\xA4\x40"); |
| 3571 |
} |
| 3572 |
v.extend_from_slice(b"\x81\x40\xA0 "); |
| 3573 |
check_bytes(&v, BIG5); |
| 3574 |
} |
| 3575 |
|
| 3576 |
#[test] |
| 3577 |
fn test_big5_single_byte_ff() { |
| 3578 |
let mut v = Vec::new(); |
| 3579 |
for _ in 0..80 { |
| 3580 |
v.extend_from_slice(b"\xA4\x40"); |
| 3581 |
} |
| 3582 |
v.extend_from_slice(b"\x81\x40\xFF "); |
| 3583 |
check_bytes(&v, BIG5); |
| 3584 |
} |
| 3585 |
|
| 3586 |
#[test] |
| 3587 |
fn test_not_big5() { |
| 3588 |
let mut v = Vec::new(); |
| 3589 |
for _ in 0..40 { |
| 3590 |
v.extend_from_slice(b"\xA4\x40"); |
| 3591 |
} |
| 3592 |
v.extend_from_slice(b"\x81\x40\xA0\xA0"); |
| 3593 |
check_bytes(&v, IBM866); |
| 3594 |
} |
| 3595 |
|
| 3596 |
#[test] |
| 3597 |
fn test_euc_kr_pua() { |
| 3598 |
let mut v = Vec::new(); |
| 3599 |
v.extend_from_slice(b"\xC9\xA1\xB0\xA1 "); |
| 3600 |
for _ in 0..40 { |
| 3601 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3602 |
} |
| 3603 |
check_bytes(&v, EUC_KR); |
| 3604 |
} |
| 3605 |
|
| 3606 |
#[test] |
| 3607 |
fn test_euc_kr_pua_bis() { |
| 3608 |
let mut v = Vec::new(); |
| 3609 |
v.extend_from_slice(b"\xFE\xA1\xB0\xA1 "); |
| 3610 |
for _ in 0..40 { |
| 3611 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3612 |
} |
| 3613 |
check_bytes(&v, EUC_KR); |
| 3614 |
} |
| 3615 |
|
| 3616 |
#[test] |
| 3617 |
fn test_euc_kr_single_byte_ff() { |
| 3618 |
let mut v = Vec::new(); |
| 3619 |
v.extend_from_slice(b"\xFF "); |
| 3620 |
for _ in 0..40 { |
| 3621 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3622 |
} |
| 3623 |
check_bytes(&v, EUC_KR); |
| 3624 |
} |
| 3625 |
|
| 3626 |
#[test] |
| 3627 |
fn test_euc_kr_single_byte_81() { |
| 3628 |
let mut v = Vec::new(); |
| 3629 |
v.extend_from_slice(b"\x81 "); |
| 3630 |
for _ in 0..40 { |
| 3631 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3632 |
} |
| 3633 |
check_bytes(&v, EUC_KR); |
| 3634 |
} |
| 3635 |
|
| 3636 |
#[test] |
| 3637 |
fn test_euc_kr_single_byte_84() { |
| 3638 |
let mut v = Vec::new(); |
| 3639 |
v.extend_from_slice(b"\x84 "); |
| 3640 |
for _ in 0..40 { |
| 3641 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3642 |
} |
| 3643 |
check_bytes(&v, EUC_KR); |
| 3644 |
} |
| 3645 |
|
| 3646 |
#[test] |
| 3647 |
fn test_not_euc_kr() { |
| 3648 |
let mut v = Vec::new(); |
| 3649 |
v.extend_from_slice(b"\xC9\xA0\xB0\xA1 "); |
| 3650 |
for _ in 0..40 { |
| 3651 |
v.extend_from_slice(b"\xC5\xD7\xBD\xBA\xC6\xAE. "); |
| 3652 |
} |
| 3653 |
check_bytes(&v, GBK); |
| 3654 |
} |
| 3655 |
|
| 3656 |
#[test] |
| 3657 |
fn test_shift_jis_x0213() { |
| 3658 |
let mut v = Vec::new(); |
| 3659 |
v.extend_from_slice(b"\x87\xE5"); |
| 3660 |
for _ in 0..40 { |
| 3661 |
v.extend_from_slice(b"\x82\xC9\x82\xD9\x82\xF1\x82\xB2"); |
| 3662 |
} |
| 3663 |
check_bytes(&v, SHIFT_JIS); |
| 3664 |
} |
| 3665 |
|
| 3666 |
#[test] |
| 3667 |
fn test_shift_jis_single_byte_fd() { |
| 3668 |
let mut v = Vec::new(); |
| 3669 |
v.extend_from_slice(b"\xFD"); |
| 3670 |
for _ in 0..40 { |
| 3671 |
v.extend_from_slice(b"\x82\xC9\x82\xD9\x82\xF1\x82\xB2"); |
| 3672 |
} |
| 3673 |
check_bytes(&v, SHIFT_JIS); |
| 3674 |
} |
| 3675 |
|
| 3676 |
#[test] |
| 3677 |
fn test_not_shift_jis() { |
| 3678 |
let mut v = Vec::new(); |
| 3679 |
v.extend_from_slice(b"\x84\xE0"); |
| 3680 |
for _ in 0..40 { |
| 3681 |
v.extend_from_slice(b"\x82\xC9\x82\xD9\x82\xF1\x82\xB2"); |
| 3682 |
} |
| 3683 |
check_bytes(&v, GBK); |
| 3684 |
} |
| 3685 |
|
| 3686 |
#[test] |
| 3687 |
fn test_not_shift_jis_bis() { |
| 3688 |
let mut v = Vec::new(); |
| 3689 |
v.extend_from_slice(b"\x87\x7D"); |
| 3690 |
for _ in 0..40 { |
| 3691 |
v.extend_from_slice(b"\x82\xC9\x82\xD9\x82\xF1\x82\xB2"); |
| 3692 |
} |
| 3693 |
check_bytes(&v, GBK); |
| 3694 |
} |
| 3695 |
|
| 3696 |
#[test] |
| 3697 |
fn test_euc_jp_x0213() { |
| 3698 |
let mut v = Vec::new(); |
| 3699 |
v.extend_from_slice(b"\xAD\xBF"); |
| 3700 |
for _ in 0..80 { |
| 3701 |
v.extend_from_slice(b"\xA4\xCB\xA4\xDB\xA4\xF3\xA4\xB4"); |
| 3702 |
} |
| 3703 |
check_bytes(&v, EUC_JP); |
| 3704 |
} |
| 3705 |
|
| 3706 |
#[test] |
| 3707 |
fn test_euc_jp_x0213_other_plane() { |
| 3708 |
let mut v = Vec::new(); |
| 3709 |
v.extend_from_slice(b"\x8F\xFE\xF6"); |
| 3710 |
for _ in 0..80 { |
| 3711 |
v.extend_from_slice(b"\xA4\xCB\xA4\xDB\xA4\xF3\xA4\xB4"); |
| 3712 |
} |
| 3713 |
check_bytes(&v, EUC_JP); |
| 3714 |
} |
| 3715 |
|
| 3716 |
#[test] |
| 3717 |
fn test_not_euc_jp() { |
| 3718 |
let mut v = Vec::new(); |
| 3719 |
v.extend_from_slice(b"\x8F\xFE\xF7"); |
| 3720 |
for _ in 0..80 { |
| 3721 |
v.extend_from_slice(b"\xA4\xCB\xA4\xDB\xA4\xF3\xA4\xB4"); |
| 3722 |
} |
| 3723 |
check_bytes(&v, WINDOWS_1252); |
| 3724 |
} |
| 3725 |
|
| 3726 |
#[test] |
| 3727 |
fn test_not_euc_jp_bis() { |
| 3728 |
let mut v = Vec::new(); |
| 3729 |
v.extend_from_slice(b"\xA8\xDF"); |
| 3730 |
for _ in 0..80 { |
| 3731 |
v.extend_from_slice(b"\xA4\xCB\xA4\xDB\xA4\xF3\xA4\xB4"); |
| 3732 |
} |
| 3733 |
check_bytes(&v, BIG5); |
| 3734 |
} |
| 3735 |
|
| 3736 |
#[test] |
| 3737 |
fn test_gbk_single_byte_ff() { |
| 3738 |
let mut v = Vec::new(); |
| 3739 |
v.extend_from_slice(b"\xFF"); |
| 3740 |
for _ in 0..80 { |
| 3741 |
v.extend_from_slice(b"\xB5\xC4"); |
| 3742 |
} |
| 3743 |
check_bytes(&v, GBK); |
| 3744 |
} |
| 3745 |
|
| 3746 |
#[test] |
| 3747 |
fn test_gbk_single_byte_a0() { |
| 3748 |
let mut v = Vec::new(); |
| 3749 |
v.extend_from_slice(b"\xA0 "); |
| 3750 |
for _ in 0..80 { |
| 3751 |
v.extend_from_slice(b"\xB5\xC4"); |
| 3752 |
} |
| 3753 |
check_bytes(&v, GBK); |
| 3754 |
} |
| 3755 |
|
| 3756 |
#[test] |
| 3757 |
fn test_gbk_single_byte_fe() { |
| 3758 |
let mut v = Vec::new(); |
| 3759 |
v.extend_from_slice(b"\xFE "); |
| 3760 |
for _ in 0..80 { |
| 3761 |
v.extend_from_slice(b"\xB5\xC4"); |
| 3762 |
} |
| 3763 |
check_bytes(&v, GBK); |
| 3764 |
} |
| 3765 |
|
| 3766 |
#[test] |
| 3767 |
fn test_not_gbk_single_byte_fc() { |
| 3768 |
let mut v = Vec::new(); |
| 3769 |
v.extend_from_slice(b"\xFC "); |
| 3770 |
for _ in 0..80 { |
| 3771 |
v.extend_from_slice(b"\xB5\xC4"); |
| 3772 |
} |
| 3773 |
check_bytes(&v, ISO_8859_5); |
| 3774 |
} |
| 3775 |
} |