1<?php 2 /** 3 * Copyright (c) 2005 Richard Heyes (http://www.phpguru.org/) 4 * 5 * All rights reserved. 6 * 7 * This script is free software. 8 */ 9 10 /** 11 * PHP5 Implementation of the Porter Stemmer algorithm. Certain elements 12 * were borrowed from the (broken) implementation by Jon Abernathy. 13 * 14 * Usage: 15 * 16 * $stem = PorterStemmer::Stem($word); 17 * 18 * How easy is that? 19 */ 20 21 // Porter Stemmer 22class PorterStemmer 23{ 24 /** 25 * Regex for matching a consonant 26 * @var string 27 */ 28 private static $regex_consonant = '(?:[bcdfghjklmnpqrstvwxz]|(?<=[aeiou])y|^y)'; 29 30 31 /** 32 * Regex for matching a vowel 33 * @var string 34 */ 35 private static $regex_vowel = '(?:[aeiou]|(?<![aeiou])y)'; 36 37 38 /** 39 * Stems a word. Simple huh? 40 * 41 * @param string $word Word to stem 42 * @return string Stemmed word 43 */ 44 public static function Stem($word) 45 { 46 static $cache = []; 47 48 if (strlen($word) <= 2) { 49 return $word; 50 } 51 52 if (isset($cache[$word])) { 53 return $cache[$word]; 54 } 55 56 $initial = $word; 57 $word = self::step1ab($word); 58 $word = self::step1c($word); 59 $word = self::step2($word); 60 $word = self::step3($word); 61 $word = self::step4($word); 62 $word = self::step5($word); 63 64 $cache[$initial] = $word; 65 66 return $word; 67 } 68 69 70 /** 71 * Step 1 72 */ 73 private static function step1ab($word) 74 { 75 // Part a 76 if (substr($word, -1) == 's') { 77 self::replace($word, 'sses', 'ss') 78 or self::replace($word, 'ies', 'i') 79 or self::replace($word, 'ss', 'ss') 80 or self::replace($word, 's', ''); 81 } 82 83 // Part b 84 if (substr($word, -2, 1) != 'e' or ! self::replace($word, 'eed', 'ee', 0)) { // First rule 85 $v = self::$regex_vowel; 86 87 // ing and ed 88 if (preg_match("#$v+#", substr($word, 0, -3)) && self::replace($word, 'ing', '') 89 or preg_match("#$v+#", substr($word, 0, -2)) && self::replace($word, 'ed', '')) { // Note use of && and OR, for precedence reasons 90 // If one of above two test successful 91 if (! self::replace($word, 'at', 'ate') 92 and ! self::replace($word, 'bl', 'ble') 93 and ! self::replace($word, 'iz', 'ize')) { 94 // Double consonant ending 95 if (self::doubleConsonant($word) 96 and substr($word, -2) != 'll' 97 and substr($word, -2) != 'ss' 98 and substr($word, -2) != 'zz') { 99 $word = substr($word, 0, -1); 100 } elseif (self::m($word) == 1 and self::cvc($word)) { 101 $word .= 'e'; 102 } 103 } 104 } 105 } 106 107 return $word; 108 } 109 110 111 /** 112 * Step 1c 113 * 114 * @param string $word Word to stem 115 */ 116 private static function step1c($word) 117 { 118 $v = self::$regex_vowel; 119 120 if (substr($word, -1) == 'y' && preg_match("#$v+#", substr($word, 0, -1))) { 121 self::replace($word, 'y', 'i'); 122 } 123 124 return $word; 125 } 126 127 128 /** 129 * Step 2 130 * 131 * @param string $word Word to stem 132 */ 133 private static function step2($word) 134 { 135 switch (substr($word, -2, 1)) { 136 case 'a': 137 self::replace($word, 'ational', 'ate', 0) 138 or self::replace($word, 'tional', 'tion', 0); 139 break; 140 141 case 'c': 142 self::replace($word, 'enci', 'ence', 0) 143 or self::replace($word, 'anci', 'ance', 0); 144 break; 145 146 case 'e': 147 self::replace($word, 'izer', 'ize', 0); 148 break; 149 150 case 'g': 151 self::replace($word, 'logi', 'log', 0); 152 break; 153 154 case 'l': 155 self::replace($word, 'entli', 'ent', 0) 156 or self::replace($word, 'ousli', 'ous', 0) 157 or self::replace($word, 'alli', 'al', 0) 158 or self::replace($word, 'bli', 'ble', 0) 159 or self::replace($word, 'eli', 'e', 0); 160 break; 161 162 case 'o': 163 self::replace($word, 'ization', 'ize', 0) 164 or self::replace($word, 'ation', 'ate', 0) 165 or self::replace($word, 'ator', 'ate', 0); 166 break; 167 168 case 's': 169 self::replace($word, 'iveness', 'ive', 0) 170 or self::replace($word, 'fulness', 'ful', 0) 171 or self::replace($word, 'ousness', 'ous', 0) 172 or self::replace($word, 'alism', 'al', 0); 173 break; 174 175 case 't': 176 self::replace($word, 'biliti', 'ble', 0) 177 or self::replace($word, 'aliti', 'al', 0) 178 or self::replace($word, 'iviti', 'ive', 0); 179 break; 180 } 181 182 return $word; 183 } 184 185 186 /** 187 * Step 3 188 * 189 * @param string $word String to stem 190 */ 191 private static function step3($word) 192 { 193 switch (substr($word, -2, 1)) { 194 case 'a': 195 self::replace($word, 'ical', 'ic', 0); 196 break; 197 198 case 's': 199 self::replace($word, 'ness', '', 0); 200 break; 201 202 case 't': 203 self::replace($word, 'icate', 'ic', 0) 204 or self::replace($word, 'iciti', 'ic', 0); 205 break; 206 207 case 'u': 208 self::replace($word, 'ful', '', 0); 209 break; 210 211 case 'v': 212 self::replace($word, 'ative', '', 0); 213 break; 214 215 case 'z': 216 self::replace($word, 'alize', 'al', 0); 217 break; 218 } 219 220 return $word; 221 } 222 223 224 /** 225 * Step 4 226 * 227 * @param string $word Word to stem 228 */ 229 private static function step4($word) 230 { 231 switch (substr($word, -2, 1)) { 232 case 'a': 233 self::replace($word, 'al', '', 1); 234 break; 235 236 case 'c': 237 self::replace($word, 'ance', '', 1) 238 or self::replace($word, 'ence', '', 1); 239 break; 240 241 case 'e': 242 self::replace($word, 'er', '', 1); 243 break; 244 245 case 'i': 246 self::replace($word, 'ic', '', 1); 247 break; 248 249 case 'l': 250 self::replace($word, 'able', '', 1) 251 or self::replace($word, 'ible', '', 1); 252 break; 253 254 case 'n': 255 self::replace($word, 'ant', '', 1) 256 or self::replace($word, 'ement', '', 1) 257 or self::replace($word, 'ment', '', 1) 258 or self::replace($word, 'ent', '', 1); 259 break; 260 261 case 'o': 262 if (substr($word, -4) == 'tion' or substr($word, -4) == 'sion') { 263 self::replace($word, 'ion', '', 1); 264 } else { 265 self::replace($word, 'ou', '', 1); 266 } 267 break; 268 269 case 's': 270 self::replace($word, 'ism', '', 1); 271 break; 272 273 case 't': 274 self::replace($word, 'ate', '', 1) 275 or self::replace($word, 'iti', '', 1); 276 break; 277 278 case 'u': 279 self::replace($word, 'ous', '', 1); 280 break; 281 282 case 'v': 283 self::replace($word, 'ive', '', 1); 284 break; 285 286 case 'z': 287 self::replace($word, 'ize', '', 1); 288 break; 289 } 290 291 return $word; 292 } 293 294 295 /** 296 * Step 5 297 * 298 * @param string $word Word to stem 299 */ 300 private static function step5($word) 301 { 302 // Part a 303 if (substr($word, -1) == 'e') { 304 if (self::m(substr($word, 0, -1)) > 1) { 305 self::replace($word, 'e', ''); 306 } elseif (self::m(substr($word, 0, -1)) == 1) { 307 if (! self::cvc(substr($word, 0, -1))) { 308 self::replace($word, 'e', ''); 309 } 310 } 311 } 312 313 // Part b 314 if (self::m($word) > 1 and self::doubleConsonant($word) and substr($word, -1) == 'l') { 315 $word = substr($word, 0, -1); 316 } 317 318 return $word; 319 } 320 321 322 /** 323 * Replaces the first string with the second, at the end of the string. If third 324 * arg is given, then the preceding string must match that m count at least. 325 * 326 * @param string $str String to check 327 * @param string $check Ending to check for 328 * @param string $repl Replacement string 329 * @param int $m Optional minimum number of m() to meet 330 * @return bool Whether the $check string was at the end 331 * of the $str string. True does not necessarily mean 332 * that it was replaced. 333 */ 334 private static function replace(&$str, $check, $repl, $m = null) 335 { 336 $len = 0 - strlen($check); 337 338 if (substr($str, $len) == $check) { 339 $substr = substr($str, 0, $len); 340 if (is_null($m) or self::m($substr) > $m) { 341 $str = $substr . $repl; 342 } 343 344 return true; 345 } 346 347 return false; 348 } 349 350 351 /** 352 * What, you mean it's not obvious from the name? 353 * 354 * m() measures the number of consonant sequences in $str. if c is 355 * a consonant sequence and v a vowel sequence, and <..> indicates arbitrary 356 * presence, 357 * 358 * <c><v> gives 0 359 * <c>vc<v> gives 1 360 * <c>vcvc<v> gives 2 361 * <c>vcvcvc<v> gives 3 362 * 363 * @param string $str The string to return the m count for 364 * @return int The m count 365 */ 366 private static function m($str) 367 { 368 $c = self::$regex_consonant; 369 $v = self::$regex_vowel; 370 371 $str = preg_replace("#^$c+#", '', $str); 372 $str = preg_replace("#$v+$#", '', $str); 373 374 preg_match_all("#($v+$c+)#", $str, $matches); 375 376 return count($matches[1]); 377 } 378 379 380 /** 381 * Returns true/false as to whether the given string contains two 382 * of the same consonant next to each other at the end of the string. 383 * 384 * @param string $str String to check 385 * @return bool Result 386 */ 387 private static function doubleConsonant($str) 388 { 389 $c = self::$regex_consonant; 390 391 return preg_match("#$c{2}$#", $str, $matches) and $matches[0]{0} == $matches[0]{1}; 392 } 393 394 395 /** 396 * Checks for ending CVC sequence where second C is not W, X or Y 397 * 398 * @param string $str String to check 399 * @return bool Result 400 */ 401 private static function cvc($str) 402 { 403 $c = self::$regex_consonant; 404 $v = self::$regex_vowel; 405 406 return preg_match("#($c$v$c)$#", $str, $matches) 407 and strlen($matches[1]) == 3 408 and $matches[1]{2} != 'w' 409 and $matches[1]{2} != 'x' 410 and $matches[1]{2} != 'y'; 411 } 412} 413