EnglishInflector.php
581 lines
| 1 | <?php |
| 2 | |
| 3 | /* |
| 4 | * This file is part of the Symfony package. |
| 5 | * |
| 6 | * (c) Fabien Potencier <fabien@symfony.com> |
| 7 | * |
| 8 | * For the full copyright and license information, please view the LICENSE |
| 9 | * file that was distributed with this source code. |
| 10 | */ |
| 11 | |
| 12 | namespace Symfony\Component\String\Inflector; |
| 13 | |
| 14 | final class EnglishInflector implements InflectorInterface |
| 15 | { |
| 16 | /** |
| 17 | * Map English plural to singular suffixes. |
| 18 | * |
| 19 | * @see http://english-zone.com/spelling/plurals.html |
| 20 | */ |
| 21 | private const PLURAL_MAP = [ |
| 22 | // First entry: plural suffix, reversed |
| 23 | // Second entry: length of plural suffix |
| 24 | // Third entry: Whether the suffix may succeed a vowel |
| 25 | // Fourth entry: Whether the suffix may succeed a consonant |
| 26 | // Fifth entry: singular suffix, normal |
| 27 | |
| 28 | // bacteria (bacterium) |
| 29 | ['airetcab', 8, true, true, 'bacterium'], |
| 30 | |
| 31 | // corpora (corpus) |
| 32 | ['aroproc', 7, true, true, 'corpus'], |
| 33 | |
| 34 | // criteria (criterion) |
| 35 | ['airetirc', 8, true, true, 'criterion'], |
| 36 | |
| 37 | // curricula (curriculum) |
| 38 | ['alucirruc', 9, true, true, 'curriculum'], |
| 39 | |
| 40 | // genera (genus) |
| 41 | ['areneg', 6, true, true, 'genus'], |
| 42 | |
| 43 | // media (medium) |
| 44 | ['aidem', 5, true, true, 'medium'], |
| 45 | |
| 46 | // memoranda (memorandum) |
| 47 | ['adnaromem', 9, true, true, 'memorandum'], |
| 48 | |
| 49 | // phenomena (phenomenon) |
| 50 | ['anemonehp', 9, true, true, 'phenomenon'], |
| 51 | |
| 52 | // strata (stratum) |
| 53 | ['atarts', 6, true, true, 'stratum'], |
| 54 | |
| 55 | // nebulae (nebula) |
| 56 | ['ea', 2, true, true, 'a'], |
| 57 | |
| 58 | // services (service) |
| 59 | ['secivres', 8, true, true, 'service'], |
| 60 | |
| 61 | // mice (mouse), lice (louse) |
| 62 | ['eci', 3, false, true, 'ouse'], |
| 63 | |
| 64 | // geese (goose) |
| 65 | ['esee', 4, false, true, 'oose'], |
| 66 | |
| 67 | // fungi (fungus), alumni (alumnus), syllabi (syllabus), radii (radius) |
| 68 | ['i', 1, true, true, 'us'], |
| 69 | |
| 70 | // men (man), women (woman) |
| 71 | ['nem', 3, true, true, 'man'], |
| 72 | |
| 73 | // children (child) |
| 74 | ['nerdlihc', 8, true, true, 'child'], |
| 75 | |
| 76 | // oxen (ox) |
| 77 | ['nexo', 4, false, false, 'ox'], |
| 78 | |
| 79 | // indices (index), appendices (appendix), prices (price) |
| 80 | ['seci', 4, false, true, ['ex', 'ix', 'ice']], |
| 81 | |
| 82 | // codes (code) |
| 83 | ['sedoc', 5, false, true, 'code'], |
| 84 | |
| 85 | // selfies (selfie) |
| 86 | ['seifles', 7, true, true, 'selfie'], |
| 87 | |
| 88 | // zombies (zombie) |
| 89 | ['seibmoz', 7, true, true, 'zombie'], |
| 90 | |
| 91 | // movies (movie) |
| 92 | ['seivom', 6, true, true, 'movie'], |
| 93 | |
| 94 | // names (name) |
| 95 | ['seman', 5, true, false, 'name'], |
| 96 | |
| 97 | // conspectuses (conspectus), prospectuses (prospectus) |
| 98 | ['sesutcep', 8, true, true, 'pectus'], |
| 99 | |
| 100 | // feet (foot) |
| 101 | ['teef', 4, true, true, 'foot'], |
| 102 | |
| 103 | // geese (goose) |
| 104 | ['eseeg', 5, true, true, 'goose'], |
| 105 | |
| 106 | // teeth (tooth) |
| 107 | ['hteet', 5, true, true, 'tooth'], |
| 108 | |
| 109 | // news (news) |
| 110 | ['swen', 4, true, true, 'news'], |
| 111 | |
| 112 | // series (series) |
| 113 | ['seires', 6, true, true, 'series'], |
| 114 | |
| 115 | // babies (baby) |
| 116 | ['sei', 3, false, true, 'y'], |
| 117 | |
| 118 | // accesses (access), addresses (address), kisses (kiss) |
| 119 | ['sess', 4, true, false, 'ss'], |
| 120 | |
| 121 | // statuses (status) |
| 122 | ['sesutats', 8, true, true, 'status'], |
| 123 | |
| 124 | // article (articles), ancle (ancles) |
| 125 | ['sel', 3, true, true, 'le'], |
| 126 | |
| 127 | // analyses (analysis), ellipses (ellipsis), fungi (fungus), |
| 128 | // neuroses (neurosis), theses (thesis), emphases (emphasis), |
| 129 | // oases (oasis), crises (crisis), houses (house), bases (base), |
| 130 | // atlases (atlas) |
| 131 | ['ses', 3, true, true, ['s', 'se', 'sis']], |
| 132 | |
| 133 | // objectives (objective), alternative (alternatives) |
| 134 | ['sevit', 5, true, true, 'tive'], |
| 135 | |
| 136 | // drives (drive) |
| 137 | ['sevird', 6, false, true, 'drive'], |
| 138 | |
| 139 | // lives (life), wives (wife) |
| 140 | ['sevi', 4, false, true, 'ife'], |
| 141 | |
| 142 | // moves (move) |
| 143 | ['sevom', 5, true, true, 'move'], |
| 144 | |
| 145 | // hooves (hoof), dwarves (dwarf), elves (elf), leaves (leaf), caves (cave), staves (staff) |
| 146 | ['sev', 3, true, true, ['f', 've', 'ff']], |
| 147 | |
| 148 | // axes (axis), axes (ax), axes (axe) |
| 149 | ['sexa', 4, false, false, ['ax', 'axe', 'axis']], |
| 150 | |
| 151 | // indexes (index), matrixes (matrix) |
| 152 | ['sex', 3, true, false, 'x'], |
| 153 | |
| 154 | // quizzes (quiz) |
| 155 | ['sezz', 4, true, false, 'z'], |
| 156 | |
| 157 | // bureaus (bureau) |
| 158 | ['suae', 4, false, true, 'eau'], |
| 159 | |
| 160 | // fees (fee), trees (tree), employees (employee) |
| 161 | ['see', 3, true, true, 'ee'], |
| 162 | |
| 163 | // edges (edge) |
| 164 | ['segd', 4, true, true, 'dge'], |
| 165 | |
| 166 | // roses (rose), garages (garage), cassettes (cassette), |
| 167 | // waltzes (waltz), heroes (hero), bushes (bush), arches (arch), |
| 168 | // shoes (shoe) |
| 169 | ['se', 2, true, true, ['', 'e']], |
| 170 | |
| 171 | // status (status) |
| 172 | ['sutats', 6, true, true, 'status'], |
| 173 | |
| 174 | // tags (tag) |
| 175 | ['s', 1, true, true, ''], |
| 176 | |
| 177 | // chateaux (chateau) |
| 178 | ['xuae', 4, false, true, 'eau'], |
| 179 | |
| 180 | // people (person) |
| 181 | ['elpoep', 6, true, true, 'person'], |
| 182 | ]; |
| 183 | |
| 184 | /** |
| 185 | * Map English singular to plural suffixes. |
| 186 | * |
| 187 | * @see http://english-zone.com/spelling/plurals.html |
| 188 | */ |
| 189 | private const SINGULAR_MAP = [ |
| 190 | // First entry: singular suffix, reversed |
| 191 | // Second entry: length of singular suffix |
| 192 | // Third entry: Whether the suffix may succeed a vowel |
| 193 | // Fourth entry: Whether the suffix may succeed a consonant |
| 194 | // Fifth entry: plural suffix, normal |
| 195 | |
| 196 | // axes (axis) |
| 197 | ['sixa', 4, false, false, 'axes'], |
| 198 | |
| 199 | // criterion (criteria) |
| 200 | ['airetirc', 8, false, false, 'criterion'], |
| 201 | |
| 202 | // nebulae (nebula) |
| 203 | ['aluben', 6, false, false, 'nebulae'], |
| 204 | |
| 205 | // children (child) |
| 206 | ['dlihc', 5, true, true, 'children'], |
| 207 | |
| 208 | // prices (price) |
| 209 | ['eci', 3, false, true, 'ices'], |
| 210 | |
| 211 | // services (service) |
| 212 | ['ecivres', 7, true, true, 'services'], |
| 213 | |
| 214 | // lives (life), wives (wife) |
| 215 | ['efi', 3, false, true, 'ives'], |
| 216 | |
| 217 | // selfies (selfie) |
| 218 | ['eifles', 6, true, true, 'selfies'], |
| 219 | |
| 220 | // movies (movie) |
| 221 | ['eivom', 5, true, true, 'movies'], |
| 222 | |
| 223 | // lice (louse) |
| 224 | ['esuol', 5, false, true, 'lice'], |
| 225 | |
| 226 | // mice (mouse) |
| 227 | ['esuom', 5, false, true, 'mice'], |
| 228 | |
| 229 | // geese (goose) |
| 230 | ['esoo', 4, false, true, 'eese'], |
| 231 | |
| 232 | // houses (house), bases (base) |
| 233 | ['es', 2, true, true, 'ses'], |
| 234 | |
| 235 | // geese (goose) |
| 236 | ['esoog', 5, true, true, 'geese'], |
| 237 | |
| 238 | // caves (cave) |
| 239 | ['ev', 2, true, true, 'ves'], |
| 240 | |
| 241 | // drives (drive) |
| 242 | ['evird', 5, false, true, 'drives'], |
| 243 | |
| 244 | // objectives (objective), alternative (alternatives) |
| 245 | ['evit', 4, true, true, 'tives'], |
| 246 | |
| 247 | // moves (move) |
| 248 | ['evom', 4, true, true, 'moves'], |
| 249 | |
| 250 | // staves (staff) |
| 251 | ['ffats', 5, true, true, 'staves'], |
| 252 | |
| 253 | // hooves (hoof), dwarves (dwarf), elves (elf), leaves (leaf) |
| 254 | ['ff', 2, true, true, 'ffs'], |
| 255 | |
| 256 | // hooves (hoof), dwarves (dwarf), elves (elf), leaves (leaf) |
| 257 | ['f', 1, true, true, ['fs', 'ves']], |
| 258 | |
| 259 | // arches (arch) |
| 260 | ['hc', 2, true, true, 'ches'], |
| 261 | |
| 262 | // bushes (bush) |
| 263 | ['hs', 2, true, true, 'shes'], |
| 264 | |
| 265 | // teeth (tooth) |
| 266 | ['htoot', 5, true, true, 'teeth'], |
| 267 | |
| 268 | // albums (album) |
| 269 | ['mubla', 5, true, true, 'albums'], |
| 270 | |
| 271 | // bacteria (bacterium), curricula (curriculum), media (medium), memoranda (memorandum), phenomena (phenomenon), strata (stratum) |
| 272 | ['mu', 2, true, true, 'a'], |
| 273 | |
| 274 | // men (man), women (woman) |
| 275 | ['nam', 3, true, true, 'men'], |
| 276 | |
| 277 | // people (person) |
| 278 | ['nosrep', 6, true, true, ['persons', 'people']], |
| 279 | |
| 280 | // criteria (criterion) |
| 281 | ['noiretirc', 9, true, true, 'criteria'], |
| 282 | |
| 283 | // phenomena (phenomenon) |
| 284 | ['nonemonehp', 10, true, true, 'phenomena'], |
| 285 | |
| 286 | // echoes (echo) |
| 287 | ['ohce', 4, true, true, 'echoes'], |
| 288 | |
| 289 | // heroes (hero) |
| 290 | ['oreh', 4, true, true, 'heroes'], |
| 291 | |
| 292 | // atlases (atlas) |
| 293 | ['salta', 5, true, true, 'atlases'], |
| 294 | |
| 295 | // aliases (alias) |
| 296 | ['saila', 5, true, true, 'aliases'], |
| 297 | |
| 298 | // irises (iris) |
| 299 | ['siri', 4, true, true, 'irises'], |
| 300 | |
| 301 | // analyses (analysis), ellipses (ellipsis), neuroses (neurosis) |
| 302 | // theses (thesis), emphases (emphasis), oases (oasis), |
| 303 | // crises (crisis) |
| 304 | ['sis', 3, true, true, 'ses'], |
| 305 | |
| 306 | // accesses (access), addresses (address), kisses (kiss) |
| 307 | ['ss', 2, true, false, 'sses'], |
| 308 | |
| 309 | // syllabi (syllabus) |
| 310 | ['suballys', 8, true, true, 'syllabi'], |
| 311 | |
| 312 | // buses (bus) |
| 313 | ['sub', 3, true, true, 'buses'], |
| 314 | |
| 315 | // circuses (circus) |
| 316 | ['suc', 3, true, true, 'cuses'], |
| 317 | |
| 318 | // hippocampi (hippocampus) |
| 319 | ['supmacoppih', 11, false, false, 'hippocampi'], |
| 320 | |
| 321 | // campuses (campus) |
| 322 | ['sup', 3, true, true, 'puses'], |
| 323 | |
| 324 | // status (status) |
| 325 | ['sutats', 6, true, true, ['status', 'statuses']], |
| 326 | |
| 327 | // conspectuses (conspectus), prospectuses (prospectus) |
| 328 | ['sutcep', 6, true, true, 'pectuses'], |
| 329 | |
| 330 | // fungi (fungus), alumni (alumnus), syllabi (syllabus), radii (radius) |
| 331 | ['su', 2, true, true, 'i'], |
| 332 | |
| 333 | // news (news) |
| 334 | ['swen', 4, true, true, 'news'], |
| 335 | |
| 336 | // feet (foot) |
| 337 | ['toof', 4, true, true, 'feet'], |
| 338 | |
| 339 | // chateaux (chateau), bureaus (bureau) |
| 340 | ['uae', 3, false, true, ['eaus', 'eaux']], |
| 341 | |
| 342 | // oxen (ox) |
| 343 | ['xo', 2, false, false, 'oxen'], |
| 344 | |
| 345 | // hoaxes (hoax) |
| 346 | ['xaoh', 4, true, false, 'hoaxes'], |
| 347 | |
| 348 | // indices (index) |
| 349 | ['xedni', 5, false, true, ['indicies', 'indexes']], |
| 350 | |
| 351 | // fax (faxes, faxxes) |
| 352 | ['xaf', 3, true, true, ['faxes', 'faxxes']], |
| 353 | |
| 354 | // boxes (box) |
| 355 | ['xo', 2, false, true, 'oxes'], |
| 356 | |
| 357 | // indexes (index), matrixes (matrix), appendices (appendix) |
| 358 | ['x', 1, true, false, ['ces', 'xes']], |
| 359 | |
| 360 | // babies (baby) |
| 361 | ['y', 1, false, true, 'ies'], |
| 362 | |
| 363 | // quizzes (quiz) |
| 364 | ['ziuq', 4, true, false, 'quizzes'], |
| 365 | |
| 366 | // waltzes (waltz) |
| 367 | ['z', 1, true, true, 'zes'], |
| 368 | ]; |
| 369 | |
| 370 | /** |
| 371 | * A list of words which should not be inflected, reversed. |
| 372 | */ |
| 373 | private const UNINFLECTED = [ |
| 374 | '', |
| 375 | |
| 376 | // data |
| 377 | 'atad', |
| 378 | |
| 379 | // deer |
| 380 | 'reed', |
| 381 | |
| 382 | // equipment |
| 383 | 'tnempiuqe', |
| 384 | |
| 385 | // feedback |
| 386 | 'kcabdeef', |
| 387 | |
| 388 | // fish |
| 389 | 'hsif', |
| 390 | |
| 391 | // health |
| 392 | 'htlaeh', |
| 393 | |
| 394 | // history |
| 395 | 'yrotsih', |
| 396 | |
| 397 | // info |
| 398 | 'ofni', |
| 399 | |
| 400 | // information |
| 401 | 'noitamrofni', |
| 402 | |
| 403 | // money |
| 404 | 'yenom', |
| 405 | |
| 406 | // moose |
| 407 | 'esoom', |
| 408 | |
| 409 | // series |
| 410 | 'seires', |
| 411 | |
| 412 | // sheep |
| 413 | 'peehs', |
| 414 | |
| 415 | // species |
| 416 | 'seiceps', |
| 417 | |
| 418 | // traffic |
| 419 | 'ciffart', |
| 420 | |
| 421 | // aircraft |
| 422 | 'tfarcria', |
| 423 | |
| 424 | // hardware |
| 425 | 'erawdrah', |
| 426 | ]; |
| 427 | |
| 428 | public function singularize(string $plural): array |
| 429 | { |
| 430 | $pluralRev = strrev($plural); |
| 431 | $lowerPluralRev = strtolower($pluralRev); |
| 432 | $pluralLength = \strlen($lowerPluralRev); |
| 433 | |
| 434 | // Check if the word is one which is not inflected, return early if so |
| 435 | if (\in_array($lowerPluralRev, self::UNINFLECTED, true)) { |
| 436 | return [$plural]; |
| 437 | } |
| 438 | |
| 439 | // The outer loop iterates over the entries of the plural table |
| 440 | // The inner loop $j iterates over the characters of the plural suffix |
| 441 | // in the plural table to compare them with the characters of the actual |
| 442 | // given plural suffix |
| 443 | foreach (self::PLURAL_MAP as $map) { |
| 444 | $suffix = $map[0]; |
| 445 | $suffixLength = $map[1]; |
| 446 | $j = 0; |
| 447 | |
| 448 | // Compare characters in the plural table and of the suffix of the |
| 449 | // given plural one by one |
| 450 | while ($suffix[$j] === $lowerPluralRev[$j]) { |
| 451 | // Let $j point to the next character |
| 452 | ++$j; |
| 453 | |
| 454 | // Successfully compared the last character |
| 455 | // Add an entry with the singular suffix to the singular array |
| 456 | if ($j === $suffixLength) { |
| 457 | // Is there any character preceding the suffix in the plural string? |
| 458 | if ($j < $pluralLength) { |
| 459 | $nextIsVowel = str_contains('aeiou', $lowerPluralRev[$j]); |
| 460 | |
| 461 | if (!$map[2] && $nextIsVowel) { |
| 462 | // suffix may not succeed a vowel but next char is one |
| 463 | break; |
| 464 | } |
| 465 | |
| 466 | if (!$map[3] && !$nextIsVowel) { |
| 467 | // suffix may not succeed a consonant but next char is one |
| 468 | break; |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | $newBase = substr($plural, 0, $pluralLength - $suffixLength); |
| 473 | $newSuffix = $map[4]; |
| 474 | |
| 475 | // Check whether the first character in the plural suffix |
| 476 | // is uppercased. If yes, uppercase the first character in |
| 477 | // the singular suffix too |
| 478 | $firstUpper = ctype_upper($pluralRev[$j - 1]); |
| 479 | |
| 480 | if (\is_array($newSuffix)) { |
| 481 | $singulars = []; |
| 482 | |
| 483 | foreach ($newSuffix as $newSuffixEntry) { |
| 484 | $singulars[] = $newBase.($firstUpper ? ucfirst($newSuffixEntry) : $newSuffixEntry); |
| 485 | } |
| 486 | |
| 487 | return $singulars; |
| 488 | } |
| 489 | |
| 490 | return [$newBase.($firstUpper ? ucfirst($newSuffix) : $newSuffix)]; |
| 491 | } |
| 492 | |
| 493 | // Suffix is longer than word |
| 494 | if ($j === $pluralLength) { |
| 495 | break; |
| 496 | } |
| 497 | } |
| 498 | } |
| 499 | |
| 500 | // Assume that plural and singular is identical |
| 501 | return [$plural]; |
| 502 | } |
| 503 | |
| 504 | public function pluralize(string $singular): array |
| 505 | { |
| 506 | $singularRev = strrev($singular); |
| 507 | $lowerSingularRev = strtolower($singularRev); |
| 508 | $singularLength = \strlen($lowerSingularRev); |
| 509 | |
| 510 | // Check if the word is one which is not inflected, return early if so |
| 511 | if (\in_array($lowerSingularRev, self::UNINFLECTED, true)) { |
| 512 | return [$singular]; |
| 513 | } |
| 514 | |
| 515 | // The outer loop iterates over the entries of the singular table |
| 516 | // The inner loop $j iterates over the characters of the singular suffix |
| 517 | // in the singular table to compare them with the characters of the actual |
| 518 | // given singular suffix |
| 519 | foreach (self::SINGULAR_MAP as $map) { |
| 520 | $suffix = $map[0]; |
| 521 | $suffixLength = $map[1]; |
| 522 | $j = 0; |
| 523 | |
| 524 | // Compare characters in the singular table and of the suffix of the |
| 525 | // given plural one by one |
| 526 | |
| 527 | while ($suffix[$j] === $lowerSingularRev[$j]) { |
| 528 | // Let $j point to the next character |
| 529 | ++$j; |
| 530 | |
| 531 | // Successfully compared the last character |
| 532 | // Add an entry with the plural suffix to the plural array |
| 533 | if ($j === $suffixLength) { |
| 534 | // Is there any character preceding the suffix in the plural string? |
| 535 | if ($j < $singularLength) { |
| 536 | $nextIsVowel = str_contains('aeiou', $lowerSingularRev[$j]); |
| 537 | |
| 538 | if (!$map[2] && $nextIsVowel) { |
| 539 | // suffix may not succeed a vowel but next char is one |
| 540 | break; |
| 541 | } |
| 542 | |
| 543 | if (!$map[3] && !$nextIsVowel) { |
| 544 | // suffix may not succeed a consonant but next char is one |
| 545 | break; |
| 546 | } |
| 547 | } |
| 548 | |
| 549 | $newBase = substr($singular, 0, $singularLength - $suffixLength); |
| 550 | $newSuffix = $map[4]; |
| 551 | |
| 552 | // Check whether the first character in the singular suffix |
| 553 | // is uppercased. If yes, uppercase the first character in |
| 554 | // the singular suffix too |
| 555 | $firstUpper = ctype_upper($singularRev[$j - 1]); |
| 556 | |
| 557 | if (\is_array($newSuffix)) { |
| 558 | $plurals = []; |
| 559 | |
| 560 | foreach ($newSuffix as $newSuffixEntry) { |
| 561 | $plurals[] = $newBase.($firstUpper ? ucfirst($newSuffixEntry) : $newSuffixEntry); |
| 562 | } |
| 563 | |
| 564 | return $plurals; |
| 565 | } |
| 566 | |
| 567 | return [$newBase.($firstUpper ? ucfirst($newSuffix) : $newSuffix)]; |
| 568 | } |
| 569 | |
| 570 | // Suffix is longer than word |
| 571 | if ($j === $singularLength) { |
| 572 | break; |
| 573 | } |
| 574 | } |
| 575 | } |
| 576 | |
| 577 | // Assume that plural is singular with a trailing `s` |
| 578 | return [$singular.'s']; |
| 579 | } |
| 580 | } |
| 581 |