| 1 |
<?php |
| 2 |
|
| 3 |
namespace FluentCommunity\Framework\Support; |
| 4 |
|
| 5 |
class Pluralizer |
| 6 |
{ |
| 7 |
/** |
| 8 |
* Plural word form rules. |
| 9 |
* |
| 10 |
* @var array<string,string> |
| 11 |
*/ |
| 12 |
public static array $plural = [ |
| 13 |
'/(quiz)$/i' => '$1zes', |
| 14 |
'/^(ox)$/i' => '$1en', |
| 15 |
'/([ml])ouse$/i' => '$1ice', |
| 16 |
'/(matr|vert|ind)(?:ix|ex)$/i' => '$1ices', |
| 17 |
'/(stoma|epo|monar|matriar|patriar|oligar|eunu)ch$/i' => '$1chs', |
| 18 |
'/(x|ch|ss|sh)$/i' => '$1es', |
| 19 |
'/([^aeiouy]|qu)y$/i' => '$1ies', |
| 20 |
'/(hive)$/i' => '$1s', |
| 21 |
'/(?:([^f])fe|([lr])f)$/i' => '$1$2ves', |
| 22 |
'/(shea|lea|loa|thie)f$/i' => '$1ves', |
| 23 |
'/sis$/i' => 'ses', |
| 24 |
'/([ti])um$/i' => '$1a', |
| 25 |
'/(torped|embarg|tomat|potat|ech|her|vet)o$/i' => '$1oes', |
| 26 |
'/(bu)s$/i' => '$1ses', |
| 27 |
'/(alias)$/i' => '$1es', |
| 28 |
'/(fung)us$/i' => '$1i', |
| 29 |
'/(ax|test)is$/i' => '$1es', |
| 30 |
'/(us)$/i' => '$1es', |
| 31 |
'/s$/i' => 's', |
| 32 |
'/$/' => 's', |
| 33 |
]; |
| 34 |
|
| 35 |
/** |
| 36 |
* Singular word form rules. |
| 37 |
* |
| 38 |
* @var array<string,string> |
| 39 |
*/ |
| 40 |
public static array $singular = [ |
| 41 |
'/(quiz)zes$/i' => '$1', |
| 42 |
'/(matr)ices$/i' => '$1ix', |
| 43 |
'/(vert|vort|ind)ices$/i' => '$1ex', |
| 44 |
'/^(ox)en$/i' => '$1', |
| 45 |
'/(alias)es$/i' => '$1', |
| 46 |
'/(octop|vir|fung)i$/i' => '$1us', |
| 47 |
'/(cris|ax|test)es$/i' => '$1is', |
| 48 |
'/(shoe)s$/i' => '$1', |
| 49 |
'/(o)es$/i' => '$1', |
| 50 |
'/(bus)es$/i' => '$1', |
| 51 |
'/([ml])ice$/i' => '$1ouse', |
| 52 |
'/(x|ch|ss|sh)es$/i' => '$1', |
| 53 |
'/(m)ovies$/i' => '$1ovie', |
| 54 |
'/(s)eries$/i' => '$1eries', |
| 55 |
'/([^aeiouy]|qu)ies$/i' => '$1y', |
| 56 |
'/([lr])ves$/i' => '$1f', |
| 57 |
'/(tive)s$/i' => '$1', |
| 58 |
'/(hive)s$/i' => '$1', |
| 59 |
'/(li|wi|kni)ves$/i' => '$1fe', |
| 60 |
'/(shea|loa|lea|thie)ves$/i' => '$1f', |
| 61 |
'/(^analy)ses$/i' => '$1sis', |
| 62 |
'/((a)naly|(b)a|(d)iagno|(p)arenthe|(p)rogno|(s)ynop|(t)he)ses$/i' => '$1$2sis', |
| 63 |
'/([ti])a$/i' => '$1um', |
| 64 |
'/(n)ews$/i' => '$1ews', |
| 65 |
'/(h|bl)ouses$/i' => '$1ouse', |
| 66 |
'/(corpse)s$/i' => '$1', |
| 67 |
'/(gallows|headquarters)$/i' => '$1', |
| 68 |
'/(us)es$/i' => '$1', |
| 69 |
'/(us|ss)$/i' => '$1', |
| 70 |
'/s$/i' => '', |
| 71 |
]; |
| 72 |
|
| 73 |
/** |
| 74 |
* Irregular word forms. |
| 75 |
* |
| 76 |
* @var array<string,string> Singular to plural mappings |
| 77 |
*/ |
| 78 |
public static array $irregular = [ |
| 79 |
'child' => 'children', |
| 80 |
'corpus' => 'corpora', |
| 81 |
'criterion' => 'criteria', |
| 82 |
'foot' => 'feet', |
| 83 |
'freshman' => 'freshmen', |
| 84 |
'goose' => 'geese', |
| 85 |
'genus' => 'genera', |
| 86 |
'human' => 'humans', |
| 87 |
'man' => 'men', |
| 88 |
'woman' => 'women', |
| 89 |
'mouse' => 'mice', |
| 90 |
'louse' => 'lice', |
| 91 |
'person' => 'people', |
| 92 |
'tooth' => 'teeth', |
| 93 |
'ox' => 'oxen', |
| 94 |
'wolf' => 'wolves', |
| 95 |
'calf' => 'calves', |
| 96 |
'leaf' => 'leaves', |
| 97 |
'loaf' => 'loaves', |
| 98 |
'life' => 'lives', |
| 99 |
'wife' => 'wives', |
| 100 |
'knife' => 'knives', |
| 101 |
'self' => 'selves', |
| 102 |
'thief' => 'thieves', |
| 103 |
'datum' => 'data', |
| 104 |
'bacterium' => 'bacteria', |
| 105 |
'medium' => 'media', |
| 106 |
'analysis' => 'analyses', |
| 107 |
'diagnosis' => 'diagnoses', |
| 108 |
'parenthesis' => 'parentheses', |
| 109 |
'hypothesis' => 'hypotheses', |
| 110 |
'thesis' => 'theses', |
| 111 |
'crisis' => 'crises', |
| 112 |
'oasis' => 'oases', |
| 113 |
'phenomenon' => 'phenomena', |
| 114 |
'nucleus' => 'nuclei', |
| 115 |
'radius' => 'radii', |
| 116 |
'stimulus' => 'stimuli', |
| 117 |
'syllabus' => 'syllabi', |
| 118 |
'index' => 'indices', |
| 119 |
'appendix' => 'appendices', |
| 120 |
'matrix' => 'matrices', |
| 121 |
'axis' => 'axes', |
| 122 |
'alumnus' => 'alumni', |
| 123 |
'formula' => 'formulae', |
| 124 |
'move' => 'moves', |
| 125 |
'tax' => 'taxes', |
| 126 |
'tech' => 'techs', |
| 127 |
'cactus' => 'cacti', |
| 128 |
'fungus' => 'fungi', |
| 129 |
'focus' => 'foci', |
| 130 |
'ellipsis' => 'ellipses', |
| 131 |
'basis' => 'bases', |
| 132 |
'vertebra' => 'vertebrae', |
| 133 |
'stratum' => 'strata', |
| 134 |
'hero' => 'heroes', |
| 135 |
'echo' => 'echoes', |
| 136 |
'potato' => 'potatoes', |
| 137 |
'tomato' => 'tomatoes', |
| 138 |
]; |
| 139 |
|
| 140 |
/** |
| 141 |
* Uncountable word forms. |
| 142 |
* |
| 143 |
* @var string[] |
| 144 |
*/ |
| 145 |
public static array $uncountable = [ |
| 146 |
'advice', |
| 147 |
'air', |
| 148 |
'alcohol', |
| 149 |
'art', |
| 150 |
'bison', |
| 151 |
'bread', |
| 152 |
'butter', |
| 153 |
'cheese', |
| 154 |
'chassis', |
| 155 |
'clothing', |
| 156 |
'commerce', |
| 157 |
'compensation', |
| 158 |
'coreopsis', |
| 159 |
'data', |
| 160 |
'deer', |
| 161 |
'dust', |
| 162 |
'education', |
| 163 |
'electricity', |
| 164 |
'equipment', |
| 165 |
'evidence', |
| 166 |
'fashion', |
| 167 |
'feedback', |
| 168 |
'fish', |
| 169 |
'flour', |
| 170 |
'food', |
| 171 |
'furniture', |
| 172 |
'garbage', |
| 173 |
'gold', |
| 174 |
'grass', |
| 175 |
'happiness', |
| 176 |
'homework', |
| 177 |
'honesty', |
| 178 |
'information', |
| 179 |
'jewelry', |
| 180 |
'knowledge', |
| 181 |
'luggage', |
| 182 |
'mathematics', |
| 183 |
'meat', |
| 184 |
'money', |
| 185 |
'moose', |
| 186 |
'music', |
| 187 |
'news', |
| 188 |
'nutrition', |
| 189 |
'offspring', |
| 190 |
'oil', |
| 191 |
'oxygen', |
| 192 |
'patience', |
| 193 |
'permission', |
| 194 |
'plankton', |
| 195 |
'poetry', |
| 196 |
'police', |
| 197 |
'progress', |
| 198 |
'rain', |
| 199 |
'rice', |
| 200 |
'salt', |
| 201 |
'series', |
| 202 |
'sheep', |
| 203 |
'shopping', |
| 204 |
'software', |
| 205 |
'species', |
| 206 |
'sugar', |
| 207 |
'swine', |
| 208 |
'tea', |
| 209 |
'traffic', |
| 210 |
'transportation', |
| 211 |
'water', |
| 212 |
'weather', |
| 213 |
'wildlife', |
| 214 |
'wood', |
| 215 |
'wool', |
| 216 |
'work', |
| 217 |
]; |
| 218 |
|
| 219 |
/** |
| 220 |
* Cache for plural inflections. |
| 221 |
* |
| 222 |
* @var array<string,string> |
| 223 |
*/ |
| 224 |
protected static array $pluralCache = []; |
| 225 |
|
| 226 |
/** |
| 227 |
* Cache for singular inflections. |
| 228 |
* |
| 229 |
* @var array<string,string> |
| 230 |
*/ |
| 231 |
protected static array $singularCache = []; |
| 232 |
|
| 233 |
/** |
| 234 |
* Convert a plural word to its singular form. |
| 235 |
* |
| 236 |
* @param string $value |
| 237 |
* @return string |
| 238 |
*/ |
| 239 |
public static function singular(string $value): string |
| 240 |
{ |
| 241 |
if (isset(static::$singularCache[$value])) { |
| 242 |
return static::$singularCache[$value]; |
| 243 |
} |
| 244 |
|
| 245 |
$result = static::inflect($value, static::$singular, static::$irregular); |
| 246 |
|
| 247 |
return static::$singularCache[$value] = $result ?: $value; |
| 248 |
} |
| 249 |
|
| 250 |
/** |
| 251 |
* Convert a singular word to its plural form. |
| 252 |
* |
| 253 |
* @param string $value |
| 254 |
* @param int $count Number of items; if 1, returns singular. |
| 255 |
* @return string |
| 256 |
*/ |
| 257 |
public static function plural(string $value, int $count = 2): string |
| 258 |
{ |
| 259 |
if ($count === 1) { |
| 260 |
return $value; |
| 261 |
} |
| 262 |
|
| 263 |
if (in_array($value, static::$irregular, true)) { |
| 264 |
return $value; |
| 265 |
} |
| 266 |
|
| 267 |
if (isset(static::$pluralCache[$value])) { |
| 268 |
return static::$pluralCache[$value]; |
| 269 |
} |
| 270 |
|
| 271 |
$irregular = array_flip(static::$irregular); |
| 272 |
$result = static::inflect($value, static::$plural, $irregular); |
| 273 |
|
| 274 |
return static::$pluralCache[$value] = $result; |
| 275 |
} |
| 276 |
|
| 277 |
/** |
| 278 |
* Perform inflection (singular or plural) on a word. |
| 279 |
* |
| 280 |
* @param string $value |
| 281 |
* @param array<string,string> $source |
| 282 |
* @param array<string,string> $irregulars |
| 283 |
* @return string|null |
| 284 |
*/ |
| 285 |
protected static function inflect(string $value, array $source, array $irregulars): ?string |
| 286 |
{ |
| 287 |
if (static::uncountable($value)) { |
| 288 |
return $value; |
| 289 |
} |
| 290 |
|
| 291 |
foreach ($irregulars as $irregular => $pattern) { |
| 292 |
if (preg_match($pattern = '/'.$pattern.'$/i', $value)) { |
| 293 |
$irregular = static::matchCase($irregular, $value); |
| 294 |
return preg_replace($pattern, $irregular, $value); |
| 295 |
} |
| 296 |
} |
| 297 |
|
| 298 |
foreach ($source as $pattern => $inflected) { |
| 299 |
if (preg_match($pattern, $value)) { |
| 300 |
$inflected = preg_replace($pattern, $inflected, $value); |
| 301 |
return static::matchCase($inflected, $value); |
| 302 |
} |
| 303 |
} |
| 304 |
|
| 305 |
return null; |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* Check if a word is uncountable. |
| 310 |
* |
| 311 |
* @param string $value |
| 312 |
* @return bool |
| 313 |
*/ |
| 314 |
protected static function uncountable(string $value): bool |
| 315 |
{ |
| 316 |
return in_array(strtolower($value), static::$uncountable, true); |
| 317 |
} |
| 318 |
|
| 319 |
/** |
| 320 |
* Match the case of a word to another word's case. |
| 321 |
* |
| 322 |
* @param string $value |
| 323 |
* @param string $comparison |
| 324 |
* @return string |
| 325 |
*/ |
| 326 |
protected static function matchCase(string $value, string $comparison): string |
| 327 |
{ |
| 328 |
$functions = ['mb_strtolower', 'mb_strtoupper', 'ucfirst', 'ucwords']; |
| 329 |
|
| 330 |
foreach ($functions as $function) { |
| 331 |
if (call_user_func($function, $comparison) === $comparison) { |
| 332 |
return call_user_func($function, $value); |
| 333 |
} |
| 334 |
} |
| 335 |
|
| 336 |
return $value; |
| 337 |
} |
| 338 |
|
| 339 |
/** |
| 340 |
* Check if a word is plural. |
| 341 |
* |
| 342 |
* @param string $word |
| 343 |
* @return bool |
| 344 |
*/ |
| 345 |
public static function isPlural(string $word): bool |
| 346 |
{ |
| 347 |
return static::singular($word) !== $word; |
| 348 |
} |
| 349 |
|
| 350 |
/** |
| 351 |
* Check if a word is singular. |
| 352 |
* |
| 353 |
* @param string $word |
| 354 |
* @return bool |
| 355 |
*/ |
| 356 |
public static function isSingular(string $word): bool |
| 357 |
{ |
| 358 |
return static::singular($word) === $word; |
| 359 |
} |
| 360 |
|
| 361 |
/** |
| 362 |
* Pluralize a word only if needed based on count. |
| 363 |
* |
| 364 |
* @param string $word |
| 365 |
* @param int $count |
| 366 |
* @return string |
| 367 |
*/ |
| 368 |
public static function pluralizeIfNeeded(string $word, int $count): string |
| 369 |
{ |
| 370 |
return $count === 1 ? $word : static::plural($word, $count); |
| 371 |
} |
| 372 |
|
| 373 |
/** |
| 374 |
* Add a new irregular singular/plural pair. |
| 375 |
* |
| 376 |
* @param string $singular |
| 377 |
* @param string $plural |
| 378 |
* @return void |
| 379 |
*/ |
| 380 |
public static function addIrregular(string $singular, string $plural): void |
| 381 |
{ |
| 382 |
static::$irregular[$singular] = $plural; |
| 383 |
static::clearCache(); |
| 384 |
} |
| 385 |
|
| 386 |
/** |
| 387 |
* Add a new uncountable word. |
| 388 |
* |
| 389 |
* @param string $word |
| 390 |
* @return void |
| 391 |
*/ |
| 392 |
public static function addUncountable(string $word): void |
| 393 |
{ |
| 394 |
$word = strtolower($word); |
| 395 |
if (!in_array($word, static::$uncountable, true)) { |
| 396 |
static::$uncountable[] = $word; |
| 397 |
static::clearCache(); |
| 398 |
} |
| 399 |
} |
| 400 |
|
| 401 |
/** |
| 402 |
* Clear the plural and singular caches. |
| 403 |
* |
| 404 |
* @return void |
| 405 |
*/ |
| 406 |
public static function clearCache(): void |
| 407 |
{ |
| 408 |
static::$pluralCache = []; |
| 409 |
static::$singularCache = []; |
| 410 |
} |
| 411 |
} |
| 412 |
|
