http://snowball[dot]tartarus[dot]org/alg[dot][dot][dot]ian/stemmer[dot]html
Class:
Спойлер (Отобразить)
PHP:
скопировать код в буфер обмена
скопировать код в буфер обмена
- class Lingua_Stem_Ru
- {
- private $VERSION = "0.02";
- private $Stem_Caching = 0;
- private $VOWEL = '/аеиоуыэюя/u';
- private $PERFECTIVEGROUND = '/((ив|ивши|ившись|ыв|ывши|ывшись)|((?<=[ая])(в|вши|вшись)))$/u';
- private $REFLEXIVE = '/(с[яь])$/u';
- private $ADJECTIVE = '/(ее|ие|ые|ое|ими|ыми|ей|ий|ый|ой|ем|им|ым|ом|его|ого|еых|ую|юю|ая|яя|ою|ею)$/u';
- private $PARTICIPLE = '/((ивш|ывш|ующ)|((?<=[ая])(ем|нн|вш|ющ|щ)))$/u';
- private $VERB = '/((ила|ыла|ена|ейте|уйте|ите|или|ыли|ей|уй|ил|ыл|им|ым|ены|ить|ыть|ишь|ую|ю)|((?<=[ая])(ла|на|ете|йте|ли|й|л|ем|н|ло|но|ет|ют|ны|ть|ешь|нно)))$/u';
- private $NOUN = '/(а|ев|ов|ие|ье|е|иями|ями|ами|еи|ии|и|ией|ей|ой|ий|й|и|ы|ь|ию|ью|ю|ия|ья|я)$/u';
- private $RVRE = '/^(.*?[аеиоуыэюя])(.*)$/u';
- private $DERIVATIONAL = '/[^аеиоуыэюя][аеиоуыэюя]+[^аеиоуыэюя]+[аеиоуыэюя].*(?<=о)сть?$/u';
- private function s(&$s, $re, $to)
- {
- $orig = $s;
- return $orig !== $s;
- }
- private function m($s, $re)
- {
- }
- public function stem_word($word)
- {
- # Check against cache of stemmed words
- return $this->Stem_Cache[$word];
- }
- $stem = $word;
- do {
- $start = $p[1];
- $RV = $p[2];
- if (!$RV) break;
- # Step 1
- if (!$this->s($RV, $this->PERFECTIVEGROUND, '')) {
- $this->s($RV, $this->REFLEXIVE, '');
- if ($this->s($RV, $this->ADJECTIVE, '')) {
- $this->s($RV, $this->PARTICIPLE, '');
- } else {
- if (!$this->s($RV, $this->VERB, ''))
- $this->s($RV, $this->NOUN, '');
- }
- }
- # Step 2
- $this->s($RV, '/и$/', '');
- # Step 3
- if ($this->m($RV, $this->DERIVATIONAL))
- $this->s($RV, '/ость?$/', '');
- # Step 4
- if (!$this->s($RV, '/ь$/', '')) {
- $this->s($RV, '/ейше?/', '');
- $this->s($RV, '/нн$/', 'н');
- }
- $stem = $start.$RV;
- } while(false);
- if ($this->Stem_Caching) $this->Stem_Cache[$word] = $stem;
- return $stem;
- }
- public function stem_caching($parm_ref)
- {
- $caching_level = @$parm_ref['-level'];
- if ($caching_level) {
- if (!$this->m($caching_level, '/^[012]$/')) {
- die(__CLASS__ . "::stem_caching() - Legal values are '0','1' or '2'. '$caching_level' is not a legal value");
- }
- $this->Stem_Caching = $caching_level;
- }
- return $this->Stem_Caching;
- }
- function clear_stem_cache()
- {
- }
- }