| 1 |
<?php
|
| 2 |
// $Id$
|
| 3 |
|
| 4 |
/**
|
| 5 |
* @file
|
| 6 |
* This is an implementation of the Porter 2 Stemming algorithm from
|
| 7 |
* http://snowball.tartarus.org/algorithms/english/stemmer.html
|
| 8 |
* by Jennifer Hodgdon of Poplar ProductivityWare, www.poplarware.com
|
| 9 |
*/
|
| 10 |
|
| 11 |
/**
|
| 12 |
* Implementation of hook_search_preprocess().
|
| 13 |
*
|
| 14 |
* Stems the words in $text, using the Porter Stemmer 2 algorithm.
|
| 15 |
*/
|
| 16 |
function porterstemmer_search_preprocess(&$text) {
|
| 17 |
|
| 18 |
// Convert text to lower case, and replace special apostrophes with regular
|
| 19 |
// apostrophes
|
| 20 |
$text = drupal_strtolower(str_replace('’', "'", $text));
|
| 21 |
|
| 22 |
// Split into words
|
| 23 |
$words = preg_split('/(' . PORTERSTEMMER_BOUNDARY . '+)/', $text, -1, PREG_SPLIT_DELIM_CAPTURE);
|
| 24 |
|
| 25 |
if (!count( $words )) {
|
| 26 |
return $text;
|
| 27 |
}
|
| 28 |
|
| 29 |
// Process each word, skipping delimiters
|
| 30 |
$isword = !preg_match('/' . PORTERSTEMMER_BOUNDARY . '/', $words[0] );
|
| 31 |
foreach ($words as $k => $word) {
|
| 32 |
if ($isword) {
|
| 33 |
$words[$k] = porterstemmer_stem($word);
|
| 34 |
}
|
| 35 |
$isword = !$isword;
|
| 36 |
}
|
| 37 |
|
| 38 |
// Put it all back together
|
| 39 |
return implode('', $words);
|
| 40 |
}
|
| 41 |
|
| 42 |
/**
|
| 43 |
* Implementation of hook_help().
|
| 44 |
*/
|
| 45 |
function porterstemmer_help($section = 'admin/help#search') {
|
| 46 |
switch ($section) {
|
| 47 |
case 'admin/modules#description':
|
| 48 |
return t('Implements the Porter-Stemmer version 2 algorithm to improve English searching.');
|
| 49 |
}
|
| 50 |
}
|
| 51 |
|
| 52 |
/**
|
| 53 |
* Regular expression defining a vowel for Porter Stemmer purposes.
|
| 54 |
*/
|
| 55 |
define('PORTERSTEMMER_VOWEL', '[aeiouy]');
|
| 56 |
|
| 57 |
/**
|
| 58 |
* Regular expression defining not-a-vowel for Porter Stemmer purposes.
|
| 59 |
*/
|
| 60 |
define('PORTERSTEMMER_NOT_VOWEL', '[^aeiouy]');
|
| 61 |
|
| 62 |
/**
|
| 63 |
* Regular expression defining not-a-vowel and not w, x, Y,
|
| 64 |
* for Porter Stemmer purposes.
|
| 65 |
*/
|
| 66 |
define('PORTERSTEMMER_NOT_VOWEL_WXY', '[^aeiouywxY]');
|
| 67 |
|
| 68 |
/**
|
| 69 |
* Regular expression defining a double consonant for Porter Stemmer purposes.
|
| 70 |
*/
|
| 71 |
define('PORTERSTEMMER_DOUBLE', '(bb|dd|ff|gg|mm|nn|pp|rr|tt)');
|
| 72 |
|
| 73 |
/**
|
| 74 |
* Regular expression defining an li-ending for Porter Stemmer purposes.
|
| 75 |
*/
|
| 76 |
define('PORTERSTEMMER_LI_END', '[cdeghkmnrt]');
|
| 77 |
|
| 78 |
/**
|
| 79 |
* Regular expression defining a word boundary for Porter Stemmer purposes
|
| 80 |
* (anything not a letter or an apostrophe).
|
| 81 |
*/
|
| 82 |
define('PORTERSTEMMER_BOUNDARY', "[^a-zA-Z']+");
|
| 83 |
|
| 84 |
/**
|
| 85 |
* Minimum number of characters in a "word", for Drupal purposes.
|
| 86 |
* Note that the Porter Stemmer algorithm leaves alone words less than
|
| 87 |
* 2 characters, but Drupal Search doesn't like search terms less than
|
| 88 |
* 3 characters, so we establish a minimum stemming length of 3.
|
| 89 |
*/
|
| 90 |
define('PORTERSTEMMER_MIN_CHARS', 3);
|
| 91 |
|
| 92 |
/**
|
| 93 |
* Stems a word, using the Porter Stemmer 2 algorithm.
|
| 94 |
*
|
| 95 |
* @param $word
|
| 96 |
* Word to stem.
|
| 97 |
* @return
|
| 98 |
* Stemmed word
|
| 99 |
*/
|
| 100 |
function porterstemmer_stem($word) {
|
| 101 |
// Each of these helper functions returns TRUE if it is time to stop
|
| 102 |
// stemming and return. If everything is fine, they modify params by
|
| 103 |
// reference, as necessary, for the next function.
|
| 104 |
|
| 105 |
$r1 = 0; // position of R1 region in original word
|
| 106 |
$r2 = 0; // position of R1 region in original word
|
| 107 |
porterstemmer_prestemming($word, $r1, $r2) OR
|
| 108 |
porterstemmer_exception1($word) OR
|
| 109 |
porterstemmer_step0($word) OR
|
| 110 |
porterstemmer_step1a($word) OR
|
| 111 |
porterstemmer_exception2($word) OR
|
| 112 |
porterstemmer_step1b($word, $r1) OR
|
| 113 |
porterstemmer_step1c($word) OR
|
| 114 |
porterstemmer_step2($word, $r1) OR
|
| 115 |
porterstemmer_step3($word, $r1, $r2) OR
|
| 116 |
porterstemmer_step4($word, $r2) OR
|
| 117 |
porterstemmer_step5($word, $r1, $r2);
|
| 118 |
|
| 119 |
porterstemmer_poststemming( $word );
|
| 120 |
return $word;
|
| 121 |
}
|
| 122 |
|
| 123 |
/**
|
| 124 |
* Returns TRUE if word is too short to continue stemming.
|
| 125 |
*/
|
| 126 |
function porterstemmer_too_short($word) {
|
| 127 |
if ( drupal_strlen($word) < PORTERSTEMMER_MIN_CHARS ) {
|
| 128 |
return TRUE;
|
| 129 |
}
|
| 130 |
}
|
| 131 |
|
| 132 |
/**
|
| 133 |
* Replaces word and calculates return value for steps.
|
| 134 |
*
|
| 135 |
* If $tmp is long enough, replaces $word with $tmp and returns FALSE
|
| 136 |
* to continue stemming process. If $tmp is too short, no replacement
|
| 137 |
* and returns TRUE to end stemming process.
|
| 138 |
*/
|
| 139 |
function porterstemmer_step_ending(&$word, $tmp) {
|
| 140 |
if ( drupal_strlen($tmp) < PORTERSTEMMER_MIN_CHARS ) {
|
| 141 |
return TRUE;
|
| 142 |
}
|
| 143 |
|
| 144 |
$word = $tmp;
|
| 145 |
return FALSE;
|
| 146 |
}
|
| 147 |
|
| 148 |
/**
|
| 149 |
* Replaces one word ending with another, if tests pass.
|
| 150 |
*
|
| 151 |
* The return value is TRUE of the ending is present at the end
|
| 152 |
* of the word, and FALSE if the ending is not present. The found
|
| 153 |
* word ending is also replaced with the given replacement, only if
|
| 154 |
* the additional regular expression (if present) matches and if the
|
| 155 |
* word is at least the given length.
|
| 156 |
*
|
| 157 |
* @param $word
|
| 158 |
* Word to performm search/replace on.
|
| 159 |
* @param $oldend
|
| 160 |
* Ending to check for.
|
| 161 |
* @param $newend
|
| 162 |
* Replacement ending.
|
| 163 |
* @param $didit
|
| 164 |
* Set to TRUE in the case that a replacement is done; left alone
|
| 165 |
* otherwise.
|
| 166 |
* @param $other
|
| 167 |
* Extra regular expression; must match to allow ending replacement.
|
| 168 |
* @param $minlen
|
| 169 |
* Minimum word length required to allow ending replacement. For
|
| 170 |
* instance, to see if a particular ending is in the R1 region,
|
| 171 |
* pass in $r1 + length of ending as the minimum word length.
|
| 172 |
* @return
|
| 173 |
* TRUE if ending was at the end of the word, FALSE if not.
|
| 174 |
*/
|
| 175 |
function porterstemmer_suffix(&$word, $oldend, $newend, &$didit, $other = NULL, $minlen = 1) {
|
| 176 |
// Check to see if the ending is there
|
| 177 |
$end_regexp = '/' . $oldend . '$/';
|
| 178 |
if (!preg_match( $end_regexp, $word )) {
|
| 179 |
// ending isn't even there
|
| 180 |
return FALSE;
|
| 181 |
}
|
| 182 |
|
| 183 |
// Does word match other regular expression?
|
| 184 |
if ($other && !preg_match($other, $word)) {
|
| 185 |
// no match, so just return without replacing
|
| 186 |
return TRUE;
|
| 187 |
}
|
| 188 |
|
| 189 |
// Is word long enough?
|
| 190 |
if (drupal_strlen($word) < $minlen) {
|
| 191 |
// too short, so just return without replacing
|
| 192 |
return TRUE;
|
| 193 |
}
|
| 194 |
|
| 195 |
// Replace word ending
|
| 196 |
$word = preg_replace($end_regexp, $newend, $word);
|
| 197 |
$didit = TRUE;
|
| 198 |
return TRUE;
|
| 199 |
}
|
| 200 |
|
| 201 |
/**
|
| 202 |
* Checks to see if a word is considered "short" in Porter Stemmer 2.
|
| 203 |
*
|
| 204 |
* A word is "short" if region R1 doesn't exist, and if it ends in a
|
| 205 |
* short syllable. A short syllable is consonant, followed by vowel,
|
| 206 |
* followed by consonant not w, x, Y; or else vowel starting a word,
|
| 207 |
* followed by a non-vowel.
|
| 208 |
*
|
| 209 |
* @param $word
|
| 210 |
* Word to check.
|
| 211 |
* @param $r1
|
| 212 |
* Start position of R1 region in word.
|
| 213 |
* @return
|
| 214 |
* TRUE if the word is short, false if not.
|
| 215 |
*/
|
| 216 |
function porterstemmer_short_word( $word, $r1 ) {
|
| 217 |
|
| 218 |
if (drupal_strlen($word) > $r1) {
|
| 219 |
// R1 region exists, so this is not a short word
|
| 220 |
return FALSE;
|
| 221 |
}
|
| 222 |
|
| 223 |
// Does it end in one type of short syllable?
|
| 224 |
if (preg_match('/^' . PORTERSTEMMER_VOWEL . PORTERSTEMMER_NOT_VOWEL . '$/',
|
| 225 |
$word)) {
|
| 226 |
return TRUE;
|
| 227 |
}
|
| 228 |
|
| 229 |
// Does it end in the other type of short syllable?
|
| 230 |
if (preg_match('/' . PORTERSTEMMER_NOT_VOWEL . PORTERSTEMMER_VOWEL .
|
| 231 |
PORTERSTEMMER_NOT_VOWEL_WXY . '$/', $word)) {
|
| 232 |
return TRUE;
|
| 233 |
}
|
| 234 |
|
| 235 |
return FALSE;
|
| 236 |
}
|
| 237 |
|
| 238 |
/**
|
| 239 |
* Pre-processes a word for the Porter Stemmer 2 algorithm.
|
| 240 |
*
|
| 241 |
* Checks for too-short words, removes initial apostrophes, sets y to
|
| 242 |
* Y (so as not to be considered a vowel) if y is at start of word or
|
| 243 |
* after a vowel. Then calculates the position of the R1 and R2
|
| 244 |
* regions in the word.
|
| 245 |
*
|
| 246 |
* @param $word
|
| 247 |
* Word to stem, modified in place if successful.
|
| 248 |
* @param $r1
|
| 249 |
* Returns the start position of the "R1" region in the word.
|
| 250 |
* @param $r2
|
| 251 |
* Returns the start position of the "R2" region in the word.
|
| 252 |
* @return
|
| 253 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 254 |
*/
|
| 255 |
function porterstemmer_prestemming(&$word, &$r1, &$r2) {
|
| 256 |
if (porterstemmer_too_short($word)) {
|
| 257 |
return TRUE;
|
| 258 |
}
|
| 259 |
|
| 260 |
$tmp = $word;
|
| 261 |
|
| 262 |
// Remove initial apostrophes
|
| 263 |
$tmp = preg_replace("/^'+/", '', $tmp);
|
| 264 |
if (porterstemmer_too_short($tmp)) {
|
| 265 |
return TRUE;
|
| 266 |
}
|
| 267 |
|
| 268 |
// y -> Y if we should treat it as consonant
|
| 269 |
$tmp = preg_replace('/^y/', 'Y', $tmp);
|
| 270 |
$count = 1;
|
| 271 |
while ( $count ) {
|
| 272 |
// Do this replacement one by one, to avoid unlikely yyyy issues
|
| 273 |
$tmp = preg_replace('/(' . PORTERSTEMMER_VOWEL . ')y/', '$1Y',
|
| 274 |
$tmp, 1, $count);
|
| 275 |
}
|
| 276 |
|
| 277 |
// This y/Y step should not have changed the word length
|
| 278 |
$word = $tmp;
|
| 279 |
|
| 280 |
// Find R1 and R2. R1 is the region after the first non-vowel
|
| 281 |
// following a vowel. R2 is the region after the first non-vowel
|
| 282 |
// following a vowel in R1.
|
| 283 |
$max = drupal_strlen($word);
|
| 284 |
$r1 = $max;
|
| 285 |
$r2 = $max;
|
| 286 |
$matches = array();
|
| 287 |
$rdef = '/^' . PORTERSTEMMER_NOT_VOWEL . '*' .
|
| 288 |
PORTERSTEMMER_VOWEL . '+(' . PORTERSTEMMER_NOT_VOWEL . ')/';
|
| 289 |
|
| 290 |
// Exceptions to R1: If word begins with 'gener', 'commun', or 'arsen',
|
| 291 |
// R1 is the remainder of the word.
|
| 292 |
if ( preg_match( '/^(gener|commun|arsen)/', $word, $matches )) {
|
| 293 |
$r1 = drupal_strlen( $matches[1] );
|
| 294 |
}
|
| 295 |
elseif (preg_match( $rdef, $word, $matches, PREG_OFFSET_CAPTURE)) {
|
| 296 |
$r1 = $matches[1][1] + 1;
|
| 297 |
};
|
| 298 |
$R1 = drupal_substr($word, $r1);
|
| 299 |
if ($R1 && preg_match( $rdef, $R1, $matches, PREG_OFFSET_CAPTURE)) {
|
| 300 |
$r2 = $r1 + $matches[1][1] + 1;
|
| 301 |
};
|
| 302 |
|
| 303 |
return FALSE;
|
| 304 |
}
|
| 305 |
|
| 306 |
/**
|
| 307 |
* Turn Y back into y to undo pre-processing.
|
| 308 |
*/
|
| 309 |
function porterstemmer_poststemming(&$word) {
|
| 310 |
$word = str_replace('Y', 'y', $word);
|
| 311 |
}
|
| 312 |
|
| 313 |
/**
|
| 314 |
* Step 0 of the algorithm: remove possessive endings.
|
| 315 |
*
|
| 316 |
* @param $word
|
| 317 |
* Word to stem, modified in place if successful.
|
| 318 |
* @return
|
| 319 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 320 |
*/
|
| 321 |
function porterstemmer_step0(&$word) {
|
| 322 |
$tmp = $word;
|
| 323 |
$didit = FALSE;
|
| 324 |
porterstemmer_suffix($tmp, "'s'", '', $didit) OR
|
| 325 |
porterstemmer_suffix($tmp, "'s", '', $didit) OR
|
| 326 |
porterstemmer_suffix($tmp, "'", '', $didit);
|
| 327 |
|
| 328 |
if (porterstemmer_too_short($tmp)) {
|
| 329 |
return TRUE;
|
| 330 |
}
|
| 331 |
|
| 332 |
$word = $tmp;
|
| 333 |
return FALSE;
|
| 334 |
}
|
| 335 |
|
| 336 |
|
| 337 |
/**
|
| 338 |
* Step 1a of algorithm: plurals, etc.
|
| 339 |
*
|
| 340 |
* @param $word
|
| 341 |
* Word to stem, modified in place if successful.
|
| 342 |
* @return
|
| 343 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 344 |
*/
|
| 345 |
function porterstemmer_step1a(&$word) {
|
| 346 |
$tmp = $word;
|
| 347 |
$didit = FALSE;
|
| 348 |
|
| 349 |
$done = porterstemmer_suffix($tmp, 'sses', 'ss', $didit);
|
| 350 |
|
| 351 |
// ies/ied endings -- have different replacements depending on
|
| 352 |
// if there is more than one letter preceeding. So make sure to
|
| 353 |
// test/replace for both conditions.
|
| 354 |
|
| 355 |
if ( !$done && porterstemmer_suffix($tmp, 'ies', 'ie', $didit, '/^.ies$/')) {
|
| 356 |
if ( !$didit ) {
|
| 357 |
porterstemmer_suffix($tmp, 'ies', 'i', $didit);
|
| 358 |
}
|
| 359 |
$done = TRUE;
|
| 360 |
}
|
| 361 |
|
| 362 |
if ( !$done && porterstemmer_suffix($tmp, 'ied', 'ie', $didit, '/^.ied$/')) {
|
| 363 |
if ( !$didit ) {
|
| 364 |
porterstemmer_suffix($tmp, 'ied', 'i', $didit);
|
| 365 |
}
|
| 366 |
$done = TRUE;
|
| 367 |
}
|
| 368 |
|
| 369 |
if ( !$done ) {
|
| 370 |
porterstemmer_suffix($tmp, 'ss', 'ss', $didit) OR
|
| 371 |
porterstemmer_suffix($tmp, 'us', 'us', $didit) OR
|
| 372 |
// only delete s at end of word if there is at least one vowel that
|
| 373 |
// is not immediately before the s
|
| 374 |
porterstemmer_suffix($tmp, 's', '', $didit,
|
| 375 |
'/' . PORTERSTEMMER_VOWEL . '.+s$/');
|
| 376 |
}
|
| 377 |
|
| 378 |
if (porterstemmer_too_short($tmp)) {
|
| 379 |
return TRUE;
|
| 380 |
}
|
| 381 |
|
| 382 |
$word = $tmp;
|
| 383 |
return FALSE;
|
| 384 |
}
|
| 385 |
|
| 386 |
|
| 387 |
/**
|
| 388 |
* Step 1b of algorithm: eed, eedly, ed, edly, ing, ingly
|
| 389 |
*
|
| 390 |
* @param $word
|
| 391 |
* Word to stem, modified in place if successful.
|
| 392 |
* @param $r1
|
| 393 |
* Position of start of R1 region in word.
|
| 394 |
* @return
|
| 395 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 396 |
*/
|
| 397 |
function porterstemmer_step1b(&$word, $r1) {
|
| 398 |
$tmp = $word;
|
| 399 |
$didit = FALSE;
|
| 400 |
|
| 401 |
// Replace these endings if in R1 region
|
| 402 |
$done = ( porterstemmer_suffix($tmp, 'eedly', 'ee', $didit, NULL, $r1 + 5) OR
|
| 403 |
porterstemmer_suffix($tmp, 'eed', 'ee', $didit, NULL, $r1 + 3));
|
| 404 |
|
| 405 |
// Delete these endings if there's a vowel before the ending
|
| 406 |
$didit = FALSE;
|
| 407 |
if ( !$done ) {
|
| 408 |
porterstemmer_suffix($tmp, 'edly', '', $didit,
|
| 409 |
'/' . PORTERSTEMMER_VOWEL . '.*edly$/' ) OR
|
| 410 |
porterstemmer_suffix($tmp, 'ed', '', $didit,
|
| 411 |
'/' . PORTERSTEMMER_VOWEL . '.*ed$/' ) OR
|
| 412 |
porterstemmer_suffix($tmp, 'ingly', '', $didit,
|
| 413 |
'/' . PORTERSTEMMER_VOWEL . '.*ingly$/' ) OR
|
| 414 |
porterstemmer_suffix($tmp, 'ing', '', $didit,
|
| 415 |
'/' . PORTERSTEMMER_VOWEL . '.*ing$/' );
|
| 416 |
}
|
| 417 |
|
| 418 |
// If we did one of these replacements, post-process...
|
| 419 |
if ( $didit ) {
|
| 420 |
$done = porterstemmer_suffix($tmp, 'at', 'ate', $didit) OR
|
| 421 |
porterstemmer_suffix($tmp, 'bl', 'ble', $didit) OR
|
| 422 |
porterstemmer_suffix($tmp, 'iz', 'ize', $didit);
|
| 423 |
if (!$done &&
|
| 424 |
preg_match('/' . PORTERSTEMMER_DOUBLE . '$/', $tmp)) {
|
| 425 |
// drop last letter if it's a double-letter ending
|
| 426 |
$tmp = drupal_substr($tmp, 0, -1);
|
| 427 |
$done = TRUE;
|
| 428 |
}
|
| 429 |
if ( !$done && porterstemmer_short_word($tmp, $r1)) {
|
| 430 |
$tmp = $tmp . 'e';
|
| 431 |
}
|
| 432 |
}
|
| 433 |
|
| 434 |
return porterstemmer_step_ending($word, $tmp);
|
| 435 |
}
|
| 436 |
|
| 437 |
/**
|
| 438 |
* Step 1c of algorithm: y suffixes
|
| 439 |
*
|
| 440 |
* @param $word
|
| 441 |
* Word to stem, modified in place if successful.
|
| 442 |
* @return
|
| 443 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 444 |
*/
|
| 445 |
function porterstemmer_step1c(&$word) {
|
| 446 |
$tmp = $word;
|
| 447 |
$didit = FALSE;
|
| 448 |
|
| 449 |
// Replace y or Y by i if the letter before is not a vowel,
|
| 450 |
// and that non-vowel is not the beginning of the word.
|
| 451 |
|
| 452 |
$ytest = '/.' . PORTERSTEMMER_NOT_VOWEL . '[Yy]$/';
|
| 453 |
porterstemmer_suffix($tmp, 'Y', 'i', $didit, $ytest ) OR
|
| 454 |
porterstemmer_suffix($tmp, 'y', 'i', $didit, $ytest );
|
| 455 |
|
| 456 |
return porterstemmer_step_ending($word, $tmp);
|
| 457 |
}
|
| 458 |
|
| 459 |
/**
|
| 460 |
* Step 2 of algorithm: misc endings in region R1.
|
| 461 |
*
|
| 462 |
* @param $word
|
| 463 |
* Word to stem, modified in place if successful.
|
| 464 |
* @param $r1
|
| 465 |
* Position of start of R1 region in word.
|
| 466 |
* @return
|
| 467 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 468 |
*/
|
| 469 |
function porterstemmer_step2(&$word, $r1) {
|
| 470 |
$tmp = $word;
|
| 471 |
$didit = FALSE;
|
| 472 |
|
| 473 |
// Search for the longest of these suffixes, and if found in R1, replace
|
| 474 |
porterstemmer_suffix($tmp, 'ational', 'ate', $didit, NULL, $r1 + 7) OR
|
| 475 |
porterstemmer_suffix($tmp, 'fulness', 'ful', $didit, NULL, $r1 + 7) OR
|
| 476 |
porterstemmer_suffix($tmp, 'iveness', 'ive', $didit, NULL, $r1 + 7) OR
|
| 477 |
porterstemmer_suffix($tmp, 'ization', 'ize', $didit, NULL, $r1 + 7) OR
|
| 478 |
porterstemmer_suffix($tmp, 'ousness', 'ous', $didit, NULL, $r1 + 7) OR
|
| 479 |
porterstemmer_suffix($tmp, 'biliti', 'ble', $didit, NULL, $r1 + 6) OR
|
| 480 |
porterstemmer_suffix($tmp, 'lessli', 'less', $didit, NULL, $r1 + 6) OR
|
| 481 |
porterstemmer_suffix($tmp, 'tional', 'tion', $didit, NULL, $r1 + 6) OR
|
| 482 |
porterstemmer_suffix($tmp, 'aliti', 'al', $didit, NULL, $r1 + 5) OR
|
| 483 |
porterstemmer_suffix($tmp, 'ation', 'ate', $didit, NULL, $r1 + 5) OR
|
| 484 |
porterstemmer_suffix($tmp, 'alism', 'al', $didit, NULL, $r1 + 5) OR
|
| 485 |
porterstemmer_suffix($tmp, 'entli', 'ent', $didit, NULL, $r1 + 5) OR
|
| 486 |
porterstemmer_suffix($tmp, 'fulli', 'ful', $didit, NULL, $r1 + 5) OR
|
| 487 |
porterstemmer_suffix($tmp, 'iviti', 'ive', $didit, NULL, $r1 + 5) OR
|
| 488 |
porterstemmer_suffix($tmp, 'ousli', 'ous', $didit, NULL, $r1 + 5) OR
|
| 489 |
porterstemmer_suffix($tmp, 'abli', 'able', $didit, NULL, $r1 + 4) OR
|
| 490 |
porterstemmer_suffix($tmp, 'alli', 'al', $didit, NULL, $r1 + 4) OR
|
| 491 |
porterstemmer_suffix($tmp, 'ator', 'ate', $didit, NULL, $r1 + 4) OR
|
| 492 |
porterstemmer_suffix($tmp, 'anci', 'ance', $didit, NULL, $r1 + 4) OR
|
| 493 |
porterstemmer_suffix($tmp, 'enci', 'ence', $didit, NULL, $r1 + 4) OR
|
| 494 |
porterstemmer_suffix($tmp, 'izer', 'ize', $didit, NULL, $r1 + 4) OR
|
| 495 |
porterstemmer_suffix($tmp, 'bli', 'ble', $didit, NULL, $r1 + 3) OR
|
| 496 |
// ogi is only replaced if preceeded by l
|
| 497 |
porterstemmer_suffix($tmp, 'ogi', 'og', $didit,
|
| 498 |
'/logi$/', $r1 + 3) OR
|
| 499 |
// li is only replaced if preceeded by a valid li-ending
|
| 500 |
porterstemmer_suffix($tmp, 'li', '', $didit,
|
| 501 |
'/' . PORTERSTEMMER_LI_END . 'li$/', $r1 + 2);
|
| 502 |
|
| 503 |
return porterstemmer_step_ending($word, $tmp);
|
| 504 |
}
|
| 505 |
|
| 506 |
/**
|
| 507 |
* Step 3 of algorithm: misc endings in region R1.
|
| 508 |
*
|
| 509 |
* @param $word
|
| 510 |
* Word to stem, modified in place if successful.
|
| 511 |
* @param $r1
|
| 512 |
* Position of start of R1 region in word.
|
| 513 |
* @param $r2
|
| 514 |
* Position of start of R2 region in word.
|
| 515 |
* @return
|
| 516 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 517 |
*/
|
| 518 |
function porterstemmer_step3(&$word, $r1, $r2) {
|
| 519 |
$tmp = $word;
|
| 520 |
$didit = FALSE;
|
| 521 |
|
| 522 |
porterstemmer_suffix($tmp, 'ational', 'ate', $didit, NULL, $r1 + 7) OR
|
| 523 |
porterstemmer_suffix($tmp, 'tional', 'tion', $didit, NULL, $r1 + 6) OR
|
| 524 |
porterstemmer_suffix($tmp, 'alize', 'al', $didit, NULL, $r1 + 5) OR
|
| 525 |
porterstemmer_suffix($tmp, 'ative', '', $didit, NULL, $r2 + 5) OR
|
| 526 |
porterstemmer_suffix($tmp, 'icate', 'ic', $didit, NULL, $r1 + 5) OR
|
| 527 |
porterstemmer_suffix($tmp, 'iciti', 'ic', $didit, NULL, $r1 + 5) OR
|
| 528 |
porterstemmer_suffix($tmp, 'ical', 'ic', $didit, NULL, $r1 + 4) OR
|
| 529 |
porterstemmer_suffix($tmp, 'ness', '', $didit, NULL, $r1 + 4) OR
|
| 530 |
porterstemmer_suffix($tmp, 'ful', '', $didit, NULL, $r1 + 3);
|
| 531 |
|
| 532 |
return porterstemmer_step_ending($word, $tmp);
|
| 533 |
}
|
| 534 |
|
| 535 |
/**
|
| 536 |
* Step 4 of algorithm: misc endings in region R2.
|
| 537 |
*
|
| 538 |
* @param $word
|
| 539 |
* Word to stem, modified in place if successful.
|
| 540 |
* @param $r2
|
| 541 |
* Position of start of R2 region in word.
|
| 542 |
* @return
|
| 543 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 544 |
*/
|
| 545 |
function porterstemmer_step4(&$word, $r2) {
|
| 546 |
$tmp = $word;
|
| 547 |
$didit = FALSE;
|
| 548 |
|
| 549 |
porterstemmer_suffix($tmp, 'ement', '', $didit, NULL, $r2 + 5) OR
|
| 550 |
porterstemmer_suffix($tmp, 'able', '', $didit, NULL, $r2 + 4) OR
|
| 551 |
porterstemmer_suffix($tmp, 'ance', '', $didit, NULL, $r2 + 4) OR
|
| 552 |
porterstemmer_suffix($tmp, 'ence', '', $didit, NULL, $r2 + 4) OR
|
| 553 |
porterstemmer_suffix($tmp, 'ible', '', $didit, NULL, $r2 + 4) OR
|
| 554 |
porterstemmer_suffix($tmp, 'ment', '', $didit, NULL, $r2 + 4) OR
|
| 555 |
porterstemmer_suffix($tmp, 'ant', '', $didit, NULL, $r2 + 3) OR
|
| 556 |
porterstemmer_suffix($tmp, 'ate', '', $didit, NULL, $r2 + 3) OR
|
| 557 |
porterstemmer_suffix($tmp, 'ent', '', $didit, NULL, $r2 + 3) OR
|
| 558 |
porterstemmer_suffix($tmp, 'ion', '', $didit, '/[st]ion$/', $r2 + 3) OR
|
| 559 |
porterstemmer_suffix($tmp, 'ism', '', $didit, NULL, $r2 + 3) OR
|
| 560 |
porterstemmer_suffix($tmp, 'iti', '', $didit, NULL, $r2 + 3) OR
|
| 561 |
porterstemmer_suffix($tmp, 'ive', '', $didit, NULL, $r2 + 3) OR
|
| 562 |
porterstemmer_suffix($tmp, 'ize', '', $didit, NULL, $r2 + 3) OR
|
| 563 |
porterstemmer_suffix($tmp, 'ous', '', $didit, NULL, $r2 + 3) OR
|
| 564 |
porterstemmer_suffix($tmp, 'al', '', $didit, NULL, $r2 + 2) OR
|
| 565 |
porterstemmer_suffix($tmp, 'er', '', $didit, NULL, $r2 + 2) OR
|
| 566 |
porterstemmer_suffix($tmp, 'ic', '', $didit, NULL, $r2 + 2);
|
| 567 |
|
| 568 |
return porterstemmer_step_ending($word, $tmp);
|
| 569 |
}
|
| 570 |
|
| 571 |
/**
|
| 572 |
* Step 5 of algorithm: e, l endings in region R1/R2.
|
| 573 |
*
|
| 574 |
* @param $word
|
| 575 |
* Word to stem, modified in place if successful.
|
| 576 |
* @param $r1
|
| 577 |
* Position of start of R1 region in word.
|
| 578 |
* @param $r2
|
| 579 |
* Position of start of R2 region in word.
|
| 580 |
* @return
|
| 581 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 582 |
*/
|
| 583 |
function porterstemmer_step5(&$word, $r1, $r2) {
|
| 584 |
$tmp = $word;
|
| 585 |
$didit = FALSE;
|
| 586 |
$done = FALSE;
|
| 587 |
|
| 588 |
// Delete l at end of word if in R2 and preceded by another l
|
| 589 |
$done = porterstemmer_suffix( $tmp, 'll', 'l', $didit, NULL, $r2 + 1 );
|
| 590 |
|
| 591 |
// Delete e at end of word if in R2, or in R1 and not preceded by
|
| 592 |
// a short syllable
|
| 593 |
$len = drupal_strlen( $tmp );
|
| 594 |
if ( !$done && preg_match( '/e$/', $tmp ) &&
|
| 595 |
( $len > $r2 ||
|
| 596 |
( $len > $r1 &&
|
| 597 |
!preg_match( '/^' . PORTERSTEMMER_VOWEL . PORTERSTEMMER_NOT_VOWEL .
|
| 598 |
'e$/', $tmp ) &&
|
| 599 |
!preg_match( '/' . PORTERSTEMMER_NOT_VOWEL . PORTERSTEMMER_VOWEL .
|
| 600 |
PORTERSTEMMER_NOT_VOWEL_WXY . 'e$/', $tmp )))) {
|
| 601 |
$tmp = drupal_substr( $tmp, 0, -1 );
|
| 602 |
}
|
| 603 |
|
| 604 |
return porterstemmer_step_ending($word, $tmp);
|
| 605 |
}
|
| 606 |
|
| 607 |
/**
|
| 608 |
* Checks exceptions for Porter Stemmer.
|
| 609 |
*
|
| 610 |
* @param $word
|
| 611 |
* Word to stem, modified in place if successful.
|
| 612 |
* @return
|
| 613 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 614 |
*/
|
| 615 |
function porterstemmer_exception1(&$word) {
|
| 616 |
// Special cases for stemming. Don't add anything in this list that
|
| 617 |
// is shorter than the minimum allowed length!
|
| 618 |
$repl = array(
|
| 619 |
'skis' => 'ski',
|
| 620 |
'skies' => 'sky',
|
| 621 |
'dying' => 'die',
|
| 622 |
'lying' => 'lie',
|
| 623 |
'tying' => 'tie',
|
| 624 |
'idly' => 'idl',
|
| 625 |
'gently' => 'gentl',
|
| 626 |
'ugly' => 'ugli',
|
| 627 |
'early' => 'earli',
|
| 628 |
'only' => 'onli',
|
| 629 |
'singly' => 'singl',
|
| 630 |
'sky' => 'sky',
|
| 631 |
'news' => 'news',
|
| 632 |
'howe' => 'howe',
|
| 633 |
'atlas' => 'atlas',
|
| 634 |
'cosmos' => 'cosmos',
|
| 635 |
'bias' => 'bias',
|
| 636 |
'andes' => 'andes',
|
| 637 |
);
|
| 638 |
|
| 639 |
// If our word is in that list, we're done.
|
| 640 |
if ( isset( $repl[ $word ])) {
|
| 641 |
$word = $repl[ $word ];
|
| 642 |
return TRUE;
|
| 643 |
}
|
| 644 |
|
| 645 |
return FALSE;
|
| 646 |
}
|
| 647 |
|
| 648 |
/**
|
| 649 |
* Checks exceptions for Porter Stemmer after Step 1a.
|
| 650 |
*
|
| 651 |
* @param $word
|
| 652 |
* Word to stem, modified in place if successful.
|
| 653 |
* @return
|
| 654 |
* TRUE if it is time to stop stemming, FALSE to continue.
|
| 655 |
*/
|
| 656 |
function porterstemmer_exception2(&$word) {
|
| 657 |
// The following words are to be left invariant.
|
| 658 |
$repl = array(
|
| 659 |
'inning' => 1,
|
| 660 |
'outing' => 1,
|
| 661 |
'canning' => 1,
|
| 662 |
'herring' => 1,
|
| 663 |
'earring' => 1,
|
| 664 |
'proceed' => 1,
|
| 665 |
'exceed' => 1,
|
| 666 |
'succeed' => 1,
|
| 667 |
);
|
| 668 |
|
| 669 |
if ( isset( $repl[ $word ])) {
|
| 670 |
return TRUE;
|
| 671 |
}
|
| 672 |
|
| 673 |
return FALSE;
|
| 674 |
}
|
Parent Directory
| 
Revision Log
| 
Revision Graph