function stem_word($word, $type) { global $debug, $stem_words, $stem_dir, $min_word_length, $common; //if ($debug == '2') echo "\r\n\r\n<br /> unstemmed: $word<br />\r\n"; // no stemming for too short words or words containing some special characters if (strlen($word) < $min_word_length || preg_match("/[\\*\\!:]|[0-9]/si", $word)) { return $word; } if ($stem_words == 'bg') { require_once "{$stem_dir}/bg_stem.php"; $word1 = bg_stemmer::stem($word); } if ($stem_words == 'cz') { require_once "{$stem_dir}/cz_stem.php"; $word1 = cz_stemmer::stem($word); } if ($stem_words == 'de') { require_once "{$stem_dir}/de_stem.php"; $word1 = de_stemmer::stem($word); } if ($stem_words == 'el') { require_once "{$stem_dir}/el_stem.php"; $stemmer = new el_stemmer(); $word1 = $stemmer->stem($word); } if ($stem_words == 'en') { require_once "{$stem_dir}/en_stem.php"; // JFIELD jfield // make all special characters ascii (for english only) // NOTE: to kill a word completely, return void setlocale(LC_CTYPE, 'en_GB'); $word = iconv('UTF-8', 'ASCII//TRANSLIT', $word); // does most of the magic $word = preg_replace("/[^\\w\\s]/", "", $word); // clean up a few weird things, // like umlauts becoming double quotes (!?) // END JFIELD end jfield $word1 = en_stemmer::stem($word); if ($word1 == "informal") { echo "<h1>'{$word}' - '{$word1}'</h1>"; exit; } } if ($stem_words == 'es') { require_once "{$stem_dir}/es_stem.php"; $word1 = es_stemmer::stem($word); } if ($stem_words == 'fi') { require_once "{$stem_dir}/fi_stem.php"; $word1 = fi_stemmer::stem($word); } if ($stem_words == 'fr') { require_once "{$stem_dir}/fr_stem.php"; $word1 = fr_stemmer::stem($word); } if ($stem_words == 'hu') { require_once "{$stem_dir}/hu_stem.php"; $word1 = hu_stemmer::stem($word); } if ($stem_words == 'nl') { require_once "{$stem_dir}/nl_stem.php"; $word1 = nl_stemmer::stem($word); } if ($stem_words == 'it') { require_once "{$stem_dir}/it_stem.php"; $stemmer = new it_stemmer(); $word1 = $stemmer->stem($word); } if ($stem_words == 'pt') { require_once "{$stem_dir}/pt_stem.php"; $word1 = pt_stemmer::stem($word); } if ($stem_words == 'ru') { require_once "{$stem_dir}/ru_stem.php"; $word1 = ru_stemmer::stem($word); } if ($stem_words == 'se') { require_once "{$stem_dir}/se_stem.php"; $word1 = se_stemmer::stem($word); } // Hopefully the stemmed word did not become too short // and the stemming algorithm did not create a common word // JFIELD doesn't think we should undo stemming for common words // because that's f*****g stupid - instead discard the whole word if (strlen($word1) < $min_word_length || $common[$word1]) { return; } //if ($debug == '2') echo "\r\n\r\n<br /> stemmed: $word<br />\r\n"; return $word1; }
function stem($word) { //$word = lower_case($word); // � and � should be treated as a vowel followed by a consonant $word = str_replace('�', 'a~', $word); $word = str_replace('�', '~o', $word); $len = strlen($word); if ($len <= 2) { return $word; } $r1 = $r2 = $rv = $len; //R1 is the region after the first non-vowel following a vowel, or is the null region at the end of the word if there is no such non-vowel. for ($i = 0; $i < $len - 1 && $r1 == $len; $i++) { if (pt_stemmer::is_vowel($word[$i]) && !pt_stemmer::is_vowel($word[$i + 1])) { $r1 = $i + 2; } } //R2 is the region after the first non-vowel following a vowel in R1, or is the null region at the end of the word if there is no such non-vowel. for ($i = $r1; $i < $len - 1 && $r2 == $len; $i++) { if (pt_stemmer::is_vowel($word[$i]) && !pt_stemmer::is_vowel($word[$i + 1])) { $r2 = $i + 2; } } if ($len > 3) { if (!pt_stemmer::is_vowel($word[1])) { // If the second letter is a consonant, RV is the region after the next following vowel $rv = pt_stemmer::getNextVowelPos($word, 2) + 1; } elseif (pt_stemmer::is_vowel($word[0]) && pt_stemmer::is_vowel($word[1])) { // or if the first two letters are vowels, RV is the region after the next consonant $rv = pt_stemmer::getNextConsonantPos($word, 2) + 1; } else { //otherwise (consonant-vowel case) RV is the region after the third letter. But RV is the end of the word if these positions cannot be found. $rv = 3; } } $r1_txt = substr($word, $r1); $r2_txt = substr($word, $r2); $rv_txt = substr($word, $rv); $word_orig = $word; // Step 1: Standard ending removal if (($suf = pt_stemmer::endsinArr($r2_txt, array('amentos', 'imentos', 'amento', 'imento', 'adoras', 'adores', 'a�o~es', 'ismos', 'istas', 'adora', 'a�a~o', 'antes', '�ncia', 'ezas', 'icos', 'icas', 'ismo', '�vel', '�vel', 'ista', 'osos', 'osas', 'ador', 'ante', 'eza', 'ico', 'ica', 'oso', 'osa'))) != '') { $word = substr($word, 0, -strlen($suf)); # rule1 } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('log�a', 'log�as'))) != '') { $word = substr($word, 0, -strlen($suf)) . 'log'; } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('uci�n', 'uciones'))) != '') { $word = substr($word, 0, -strlen($suf)) . 'u'; } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('�ncia', '�ncias'))) != '') { $word = substr($word, 0, -strlen($suf)) . 'ente'; } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('ativamente', 'ivamente', 'osamente', 'icamente', 'adamente'))) != '') { $word = substr($word, 0, -strlen($suf)); } elseif (($suf = pt_stemmer::endsinArr($r1_txt, array('amente'))) != '') { $word = substr($word, 0, -strlen($suf)); } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('antemente', 'avelmente', '�velmente', 'mente'))) != '') { $word = substr($word, 0, -strlen($suf)); } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('abilidade', 'abilidades', 'icidade', 'icidades', 'ividad', 'ividades', 'idade', 'idades'))) != '') { $word = substr($word, 0, -strlen($suf)); } elseif (($suf = pt_stemmer::endsinArr($r2_txt, array('ativa', 'ativo', 'ativas', 'ativos', 'iva', 'ivo', 'ivas', 'ivos'))) != '') { $word = substr($word, 0, -strlen($suf)); } elseif (($suf = pt_stemmer::endsinArr($rv_txt, array('eira', 'eiras'))) != '') { $word = substr($word, 0, -strlen($suf)) . 'ir'; } if ($word != $word_after0) { $r1_txt = substr($word, $r1); $r2_txt = substr($word, $r2); $rv_txt = substr($word, $rv); } $stem = $word; if ($stem == $word_orig) { // Do step 2 if no ending was removed by step 1: now remove verb suffixes if ($suf = pt_stemmer::endsinArr($rv_txt, array('ar�amos', 'er�amos', 'ir�amos', '�ssemos', '�ssemos', '�ssemos', 'ar�eis', 'er�eis', 'ir�eis', '�sseis', '�sseis', '�sseis', '�ramos', '�ramos', '�ramos', '�vamos', 'aremos', 'eremos', 'iremos', 'ariam', 'eriam', 'iriam', 'assem', 'essem', 'issem', 'ara~o', 'era~o', 'ira~o', 'arias', 'erias', 'irias', 'ardes', 'erdes', 'irdes', 'asses', 'esses', 'isses', 'astes', 'estes', 'istes', '�reis', 'areis', '�reis', 'ereis', '�reis', 'ireis', '�veis', '�amos', 'armos', 'ermos', 'irmos', 'aria', 'eria', 'iria', 'asse', 'esse', 'isse', 'aste', 'este', 'iste', 'arei', 'erei', 'irei', 'aram', 'eram', 'iram', 'avam', 'arem', 'erem', 'irem', 'ando', 'endo', 'indo', 'adas', 'idas', 'ar�s', 'aras', 'er�s', 'eras', 'ir�s', 'avas', 'ares', 'eres', 'ires', '�eis', 'ados', 'idos', '�mos', 'amos', 'emos', 'imos', 'iras', 'ada', 'ida', 'ar�', 'ara', 'er�', 'era', 'ir�', 'ava', 'iam', 'ado', 'ido', 'ias', 'ais', 'eis', 'ira', 'ia', 'ei', 'am', 'em', 'ar', 'er', 'ir', 'as', 'es', 'is', 'eu', 'iu', 'ou'))) { $word = substr($word, 0, -strlen($suf)); } if ($word != $word_after1) { $r1_txt = substr($word, $r1); $r2_txt = substr($word, $r2); $rv_txt = substr($word, $rv); } $stem = $word; } if ($stem != $word_orig) { // Step 3 if (pt_stemmer::endsin($rv_txt, 'ci')) { $word = substr($word, 0, -1); $r1_txt = substr($word, $r1); $r2_txt = substr($word, $r2); $rv_txt = substr($word, $rv); } } else { // Step 4 conditioned if (($suf = pt_stemmer::endsinArr($rv_txt, array('os', 'a', 'i', 'o', '�', '�', '�'))) != '') { $word = substr($word, 0, -strlen($suf)); $r1_txt = substr($word, $r1); $r2_txt = substr($word, $r2); $rv_txt = substr($word, $rv); } } // Always perform step 5 if (($suf = pt_stemmer::endsinArr($rv_txt, array(' cie', ' ci�', ' ci�', 'gue', 'gu�', 'gu�', 'e', '�', '�'))) != '') { $word = substr($word, 0, -strlen($suf)); } $word = str_replace('a~', '�', $word); $word = str_replace('~o', '�', $word); return pt_stemmer::removeAccent($word); }