$word = iconv('windows-1251', $morphy->getEncoding(), $word);
}
unset($word);
}
*/
try {
foreach ($words as $word) {
// by default, phpMorphy finds $word in dictionary and when nothig found, try to predict them
// you can change this behaviour, via second argument to getXXX or findWord methods
$base = $morphy->getBaseForm($word);
$all = $morphy->getAllForms($word);
$part_of_speech = $morphy->getPartOfSpeech($word);
// $base = $morphy->getBaseForm($word, phpMorphy::NORMAL); // normal behaviour
// $base = $morphy->getBaseForm($word, phpMorphy::IGNORE_PREDICT); // don`t use prediction
// $base = $morphy->getBaseForm($word, phpMorphy::ONLY_PREDICT); // always predict word
$is_predicted = $morphy->isLastPredicted();
// or $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_NONE
$is_predicted_by_db = $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_DB;
$is_predicted_by_suffix = $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_SUFFIX;
// this used for deep analysis
$collection = $morphy->findWord($word);
// or var_dump($morphy->getAllFormsWithGramInfo($word)); for debug
if (false === $collection) {
echo $word, " NOT FOUND\n";
continue;
} else {
}
echo $is_predicted ? '-' : '+', $word, "\n";
echo 'lemmas: ', implode(', ', $base), "\n";
echo 'all: ', implode(', ', $all), "\n";
echo 'poses: ', implode(', ', $part_of_speech), "\n";
protected function getLemmas(array $words)
{
require_once MODX_CORE_PATH . 'components/modsearch/external/phpmorphy/src/common.php';
// set some options
$opts = array('storage' => PHPMORPHY_STORAGE_FILE, 'predict_by_suffix' => true, 'predict_by_db' => true, 'graminfo_as_text' => true);
// Path to directory where dictionaries located
$dir = MODX_CORE_PATH . 'components/modsearch/external/phpmorphy/dicts';
$lang = 'ru_RU';
// Create phpMorphy instance
# try {
# $morphy = new phpMorphy($dir, $lang, $opts);
# // print $morphy->getEncoding();
# // exit;
# } catch(phpMorphy_Exception $e) {
# die('Error occured while creating phpMorphy instance: ' . PHP_EOL . $e);
# }
if (!($morphy = new phpMorphy($dir, $lang, $opts))) {
$this->modx->log(xPDO::LOG_LEVEL_ERROR, "[" . __CLASS__ . "] Не был получен объект phpMorphy");
return false;
}
$lemmas = array();
# print_r($words);
#
# exit;
// print $morphy->getEncoding();
// if(function_exists('iconv')) {
// foreach($words as &$word) {
// $word = iconv('windows-1251', $morphy->getEncoding(), $word);
// }
// unset($word);
// }
# try {
foreach ($words as $word) {
if (!$word) {
continue;
}
# $word = mb_strtoupper($word, 'UTF-8');
// by default, phpMorphy finds $word in dictionary and when nothig found, try to predict them
// you can change this behaviour, via second argument to getXXX or findWord methods
$base = $morphy->getBaseForm($word);
$all = $morphy->getAllForms($word);
$part_of_speech = $morphy->getPartOfSpeech($word);
// echo $morphy->getLocale();
// var_dump($morphy->getShmCache()->getFilesList());
// print_r($base);
// exit;
// $base = $morphy->getBaseForm($word, phpMorphy::NORMAL); // normal behaviour
// $base = $morphy->getBaseForm($word, phpMorphy::IGNORE_PREDICT); // don`t use prediction
// $base = $morphy->getBaseForm($word, phpMorphy::ONLY_PREDICT); // always predict word
$is_predicted = $morphy->isLastPredicted();
// or $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_NONE
$is_predicted_by_db = $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_DB;
$is_predicted_by_suffix = $morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_SUFFIX;
// this used for deep analysis
$collection = $morphy->findWord($word);
// or var_dump($morphy->getAllFormsWithGramInfo($word)); for debug
if (false === $collection) {
# echo $word, " NOT FOUND\n";
// Если слово не найдено, добавляем его в массив как есть
$lemmas[] = $word;
continue;
} else {
# print "\n<br />Найдено слово: ". $word;
}
foreach ($base as $lemma) {
$lemmas[] = $lemma;
}
# echo $is_predicted ? '-' : '+', $word, "\n";
# echo 'lemmas: ', implode(', ', $base), "\n";
# echo 'all: ', implode(', ', $all), "\n";
# echo 'poses: ', implode(', ', $part_of_speech), "\n";
#
# echo "\n";
// $collection collection of paradigm for given word
// TODO: $collection->getByPartOfSpeech(...);
# foreach($collection as $paradigm) {
// TODO: $paradigm->getBaseForm();
// TODO: $paradigm->getAllForms();
// TODO: $paradigm->hasGrammems(array('', ''));
// TODO: $paradigm->getWordFormsByGrammems(array('', ''));
// TODO: $paradigm->hasPartOfSpeech('');
// TODO: $paradigm->getWordFormsByPartOfSpeech('');
# echo "lemma: ", $paradigm[0]->getWord(), "\n";
# foreach($paradigm->getFoundWordForm() as $found_word_form) {
# echo
# $found_word_form->getWord(), ' ',
# $found_word_form->getPartOfSpeech(), ' ',
# '(', implode(', ', $found_word_form->getGrammems()), ')',
# "\n";
# }
# echo "\n";
#
# foreach($paradigm as $word_form) {
# // TODO: $word_form->getWord();
# // TODO: $word_form->getFormNo();
# // TODO: $word_form->getGrammems();
# // TODO: $word_form->getPartOfSpeech();
# // TODO: $word_form->hasGrammems(array('', ''));
# }
# }
# echo "--\n";
}
# } catch(phpMorphy_Exception $e) {
# die('Error occured while text processing: ' . $e->getMessage());
# }
$lemmas = $this->sanitizeWordsArray($lemmas);
return array_unique($lemmas);
}
function bench_morphy($words, $encoding, $dictDir, $lang, $storage, $useBulk, $usePrediction = true)
{
$opts = array('storage' => $storage, 'predict_by_suffix' => $usePrediction, 'predict_by_db' => false);
$bundle = new phpMorphy_FilesBundle($dictDir, $lang);
$morphy = new phpMorphy($bundle, $opts);
$unicode = phpMorphy_UnicodeHelper_UnicodeHelperAbstract::getHelperForEncoding($morphy->getEncoding());
echo "Bench phpMorphy[{$encoding}][{$storage}][" . ($useBulk ? 'BULK' : 'SINGLE') . "] : ";
convert_words($words, $encoding, MB_CASE_UPPER);
$predicted = 0;
$b = microtime(true);
if ($useBulk) {
$morphy->getBaseForm($words);
} else {
foreach ($words as $word) {
//$unicode->strrev($word); mb_strtoupper($word, 'utf-8');
//strtr($word, $replace);
//strrev($word);
//mb_strtolower($word, 'utf-8');
$lemma = $morphy->getBaseForm($word);
if ($morphy->isLastPredicted()) {
$predicted++;
}
}
}
$e = microtime(true);
printf("time = %0.2f sec, words per second = %0.2f, predicted = %d\n", $e - $b, count($words) / ($e - $b), $predicted);
}
