private function getCase($word, $case)
{
$info = $this->morphy->castFormByGramInfo($word, 'С', array($case, 'ЕД'), false);
if (isset($info[0])) {
return $this->mb_ucwords($info[0]['form']);
}
return $this->mb_ucwords($word);
}
static function instance(phpMorphy $morphy)
{
$key = $morphy->getEncoding();
if (!isset(self::$instances[$key])) {
$class = __CLASS__;
self::$instances[$key] = new $class($key);
}
return self::$instances[$key];
}
public static function forms($word)
{
$opts = array('storage' => PHPMORPHY_STORAGE_FILE, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$dir = MODPATH . 'phpMorphy/dicts';
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
$morphy = new phpMorphy($dict_bundle, $opts);
$base_form = $morphy->getPseudoRoot(mb_strtoupper($word));
if (reset($base_form)) {
return mb_strtolower(reset($base_form));
}
return $word;
}
/**
* Normalize
* @param \ZendSearch\Lucene\Analysis\Token $token
* @return null|\ZendSearch\Lucene\Analysis\Token
*/
public function normalize(Token $token)
{
$pseudo_root = $this->morphy->getPseudoRoot($this->string()->toUpper($token->getTermText()));
if ($pseudo_root === false) {
$new_str = $this->string()->toUpper($token->getTermText());
} else {
$new_str = $pseudo_root[0];
}
if (strlen($new_str) < 3) {
return null;
}
$new_token = new Token($new_str, $token->getStartOffset(), $token->getEndOffset());
$new_token->setPositionIncrement($token->getPositionIncrement());
return $new_token;
}
/**
* @param string $word
* @return bool|string
*/
public static function castChosenWordBy($word)
{
$cacheKey = __CLASS__ . __FUNCTION__ . $word;
$res = Yii::$app->cache->get($cacheKey);
if ($res === false) {
$phpMorphy = new \phpMorphy(Yii::getAlias('@madmin/phpmorphy-dicts'), 'ru_RU', ['storage' => PHPMORPHY_STORAGE_FILE]);
mb_internal_encoding('UTF-8');
$forms = $phpMorphy->getGramInfo(mb_strtoupper($word));
$forms = $forms[0];
foreach ($forms as $form) {
if (in_array('ИМ', $form['grammems'])) {
$rod = array_intersect($form['grammems'], ['МР', 'ЖР', 'СР', 'МР-ЖР']);
$rod = reset($rod);
$od = array_intersect($form['grammems'], ['ОД', 'НО']);
$od = reset($od);
break;
}
}
if (!empty($rod) && !empty($od)) {
$form = $phpMorphy->castFormByGramInfo(mb_strtoupper('выбранный'), 'ПРИЧАСТИЕ', [$rod, $od, 'ВН', 'ЕД', 'ПРШ', 'СТР'], true);
$res = mb_strtolower($form[0]);
}
Yii::$app->cache->set($cacheKey, $res);
}
return $res;
}
/**
* Returns word inflectional forms.
* @param string $content
* @return array
*/
public function inflectionalForms($content)
{
if (empty($content)) {
return null;
}
// optimization (Lazy loading)
$hash = md5($content);
if (isset(static::$content[$hash])) {
return static::$content[$hash];
}
$content = preg_replace(['/\\[.*\\]/isu', '/[^\\w\\x7F-\\xFF\\s]/isu', '/[\\«\\»\\d]+/iu'], "", trim(strip_tags($content)));
/**
* trim twice spaces
*/
$content = preg_replace('/ +/u', ' ', $content);
//preg_match_all('/[a-zA-Z]+/iu',mb_strtoupper($str, CHARSET),$words_latin);
//$words_latin = (is_array($words_latin) && count($words_latin) > 0) ? ' '.implode(' ', $words_latin[0]) : '';
$words = preg_split('/\\s|[,.:;!?"\'()]/', $content, -1, PREG_SPLIT_NO_EMPTY);
$bulkWords = [];
foreach ($words as $res_words) {
if (mb_strlen($res_words, 'utf-8') > 2) {
$bulkWords[] = mb_strtoupper($res_words, 'utf-8');
}
}
return static::$content[$hash] = static::$morphy->getAllForms($bulkWords);
//return $res.$words_latin;
}
protected function processing($words)
{
$morphy = new phpMorphy();
foreach ($words as &$word) {
if (preg_match('/[а-я]+/ui', $word)) {
$lang = 'ru';
} else {
$lang = 'en';
}
$_word = $morphy->get($lang)->getBaseForm($word);
if ($_word) {
if (is_array($_word)) {
$word = current($_word);
} else {
$word = $_word;
}
}
}
return $words;
}
/**
* Возвращает все словоформы слов поискового запроса
*/
function Words2AllForms($text)
{
require_once $GLOBALS['_PATH']['PATH_INC'] . 'phpMorphy/src/common.php';
$opts = array('storage' => PHPMORPHY_STORAGE_MEM, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$dir = $GLOBALS['_PATH']['PATH_INC'] . 'phpMorphy/dicts';
// Создаем объект словаря
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
$morphy = new phpMorphy($dict_bundle, $opts);
// $codepage = $morphy->getCodepage();
setlocale(LC_CTYPE, array('ru_RU.CP1251', 'Russian_Russia.1251'));
$words = preg_split('#\\s|[,.:;!?"\'()]#', $text, -1, PREG_SPLIT_NO_EMPTY);
$bulk_words = array();
foreach ($words as $v) {
if (strlen($v) > 3) {
$v = iconv("UTF-8", "windows-1251", $v);
$bulk_words[] = strtoupper($v);
}
}
return $morphy->getAllForms($bulk_words);
}
function Words2BaseForm($text)
{
global $_COMMON_SITE_CONF;
static $dict_bundle, $morphy;
require_once $GLOBALS['_PATH']['PATH_INC'] . 'phpMorphy/src/common.php';
if (!$dict_bundle) {
$encoding = $_COMMON_SITE_CONF['encodings'][$_COMMON_SITE_CONF['site_encoding']];
$dir = $GLOBALS['_PATH']['PATH_INC'] . 'phpMorphy/dicts/';
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
}
if (!$morphy) {
$opts = array('storage' => PHPMORPHY_STORAGE_MEM, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$morphy = new phpMorphy($dict_bundle, $opts);
}
setlocale(LC_CTYPE, array('ru_RU.CP1251', 'rus_RUS.CP1251', 'rus_RUS.CP1251', 'Russian_Russia.1251'));
$words = preg_replace('#\\[.*\\]#isU', '', $text);
$words = preg_split('#\\s|[,.:;«»!?"\'()]#', $words, -1, PREG_SPLIT_NO_EMPTY);
$bulk_words = array();
foreach ($words as $v) {
if (strlen($v) > 3) {
$bulk_words[] = strtoupper($v);
}
}
$base_form = $morphy->getBaseForm($bulk_words);
$fullList = array();
if (is_array($base_form) && count($base_form)) {
foreach ($base_form as $k => $v) {
if (is_array($v)) {
foreach ($v as $v1) {
if (strlen($v1) > 3) {
$fullList[$v1] = 1;
}
}
}
}
}
$words = join(' ', array_keys($fullList));
return $words;
}
$words = preg_split('#\\s|[,.:;!?"\'()]#', $text, -1, PREG_SPLIT_NO_EMPTY);
$bulk_words = array();
foreach ($words as $v) {
if (strlen($v) > 3) {
$bulk_words[] = strtoupper($v);
}
}
return $morphy->getAllForms($bulk_words);
}
/*
*
* @param string $text
* @return string
*/
function Words2BaseForm($text)
{
require_once $GLOBALS['PATH_sys'] . 'phpmorphy/src/common.php';
// set some options
$opts = array('storage' => PHPMORPHY_STORAGE_MEM, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$dir = $GLOBALS['PATH_sys'] . 'phpmorphy/dicts';
// Create descriptor for dictionary located in $dir directory with russian language
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
// Create phpMorphy instance
$morphy = new phpMorphy($dict_bundle, $opts);
// All words in dictionary in UPPER CASE, so don`t forget set proper locale
// Supported dicts and locales:
// *------------------------------*
// | Dict. language | Locale name |
// |------------------------------|
// | Russian | cp1251 |
// |------------------------------|
// | English | cp1250 |
// |------------------------------|
// | German | cp1252 |
// *------------------------------*
// $codepage = $morphy->getCodepage();
//setlocale(LC_CTYPE, array('ru_RU.CP1251', 'Russian_Russia.1251'));
$words = preg_replace('#\\[.*\\]#isU', '', $text);
$words = preg_split('#\\s|[,.:;!?"\'()]#', $words, -1, PREG_SPLIT_NO_EMPTY);
$bulk_words = array();
foreach ($words as $v) {
if (strlen($v) > 3) {
$bulk_words[] = mb_strtoupper($v, "UTF-8");
static function create(phpMorphy $morphy)
{
$locale = $GLOBALS['__phpmorphy_strtolower']($morphy->getLocale());
if (!isset(self::$included[$locale])) {
$file_name = PHPMORPHY_DIR . "/langs_stuff/{$locale}.php";
$class = "phpMorphy_GrammemsProvider_{$locale}";
if (is_readable($file_name)) {
require $file_name;
if (!class_exists($class)) {
throw new phpMorphy_Exception("Class '{$class}' not found in '{$file_name}' file");
}
self::$included[$locale] = call_user_func(array($class, 'instance'), $morphy);
} else {
self::$included[$locale] = new phpMorphy_GrammemsProvider_Empty($morphy);
}
}
return self::$included[$locale];
}
protected function getMorphy($txt = null)
{
require_once $_SERVER['DOCUMENT_ROOT'] . '/plugins/phpmorphy/src/common.php';
$opts = array('storage' => PHPMORPHY_STORAGE_FILE, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
// Path to directory where dictionaries located
$dir = $_SERVER['DOCUMENT_ROOT'] . '/plugins/phpmorphy/dicts/';
// Create descriptor for dictionary located in $dir directory with russian language
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
// Create phpMorphy instance
try {
$morphy = new phpMorphy($dict_bundle, $opts);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . $e->getMessage());
}
$search = $_GET['q'];
$search = mysql_real_escape_string($search);
$search = mb_substr($search, 0, 30, 'UTF-8');
$search = explode(" ", $search);
$q = '';
for ($i = 0; $i < count($search); $i++) {
$ass = $search[$i];
$ass = mb_strtoupper($ass, 'utf-8');
$pseudo_root = $morphy->getPseudoRoot($ass);
if (false === $pseudo_root) {
$rep = $ass;
} else {
$rep = $pseudo_root[0];
}
if ($txt == "text") {
$q = $q . "A.text LIKE '%" . functions::q($rep) . "%' and ";
} else {
$q = $q . "A.title LIKE '%" . functions::q($rep) . "%' and ";
}
}
$searchwords = substr($q, 0, strlen($q) - 4);
return $searchwords;
}
function Words2BaseForm($text)
{
require_once dirname(__FILE__) . '/../sys/phpmorphy/src/common.php';
// set some options
$opts = array('storage' => PHPMORPHY_STORAGE_MEM, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$dir = dirname(__FILE__) . '/../sys/phpmorphy/dicts';
// Create descriptor for dictionary located in $dir directory with russian language
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
// Create phpMorphy instance
$morphy = new phpMorphy($dict_bundle, $opts);
$words = preg_replace('#\\[.*\\]#isU', '', $text);
$words = preg_split('#\\s|[,.:;!?"\'()]#', $words, -1, PREG_SPLIT_NO_EMPTY);
$bulk_words = array();
foreach ($words as $v) {
if (strlen($v) > 2) {
$bulk_words[] = strtoupper($v);
}
}
$base_form = $morphy->getBaseForm($bulk_words);
$fullList = array();
if (is_array($base_form) && count($base_form)) {
foreach ($base_form as $k => $v) {
if (is_array($v)) {
foreach ($v as $v1) {
if (strlen($v1) > 2) {
$fullList[$v1] = 1;
}
}
}
}
}
$words = join(' ', array_keys($fullList));
return $words;
}
function bench_morphy_dict($words, $encoding, $dictDir, $lang, $storage)
{
$opts = array('storage' => $storage, 'predict_by_suffix' => false, 'predict_by_db' => false);
$bundle = new phpMorphy_FilesBundle($dictDir, $lang);
$morphy = new phpMorphy($bundle, $opts);
echo "Bench phpMorphy - Dict[{$encoding}][{$storage}]: ";
convert_words($words, $encoding, MB_CASE_UPPER);
$fsa = $morphy->getCommonMorphier()->getFinder()->getFsa();
$root = $fsa->getRootTrans();
$predicted = 0;
$b = microtime(true);
foreach ($words as $word) {
$result = $fsa->walk($root, $word, true);
}
$e = microtime(true);
printf("time = %0.2f sec, words per second = %0.2f, predicted = %d\n", $e - $b, count($words) / ($e - $b), $predicted);
}
<?php
session_start();
error_reporting(E_ALL | E_STRICT);
//ini_set('default_charset','UTF-8');
$path = $_SESSION['path'];
require 'stemmer_utf.php';
require 'work_csv.php';
require 'readfile.php';
require_once dirname(__FILE__) . '/phpmorphy/src/common.php';
$opts = array('storage' => PHPMORPHY_STORAGE_FILE, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
$dir = dirname(__FILE__) . '/phpmorphy/dicts';
// Create descriptor for dictionary located in $dir directory with russian language
$dict_bundle = new phpMorphy_FilesBundle($dir, 'rus');
try {
$morphy = new phpMorphy($dict_bundle, $opts);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . $e->getMessage());
}
//$csv = new CSV($path);
// $csv_lines = $csv->getCSV($path);
$handle = fopen($path, "r");
//$row = 1;
while (($data = fgetcsv($handle, 1000, ",")) !== FALSE) {
$insertValues = array();
foreach ($data as $v) {
$insertValues[] = addslashes(trim($v));
$values = implode(',', $insertValues);
$word = $morphy->lemmatize($values);
//print_r($morphy->getBaseForm($value));
//$z=$morphy->getBaseForm($values);
}
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);
}
<?php
error_reporting(E_ALL | E_STRICT);
// first we include phpmorphy library
require_once dirname(__FILE__) . '/../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 = dirname(__FILE__) . '/../dicts';
$lang = 'ru_RU';
// Create phpMorphy instance
try {
$morphy = new phpMorphy($dir, $lang, $opts);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . PHP_EOL . $e);
}
// All words in dictionary in UPPER CASE, so don`t forget set proper locale via setlocale(...) call
// $morphy->getEncoding() returns dictionary encoding
$words = array('КРАКОЗЯБЛИКИ', 'СТАЛИ', 'ВИНА', 'И', 'ДУХИ', 'abc');
/*
if(function_exists('iconv')) {
foreach($words as &$word) {
$word = iconv('windows-1251', $morphy->getEncoding(), $word);
}
unset($word);
}
*/
try {
foreach ($words as $word) {
// by default, phpMorphy finds $word in dictionary and when nothig found, try to predict them
// you can change this behaviour, via second argument to getXXX or findWord methods
#!/usr/bin/env php
<?php
set_include_path(__DIR__ . '/../src/' . PATH_SEPARATOR . get_include_path());
require 'phpMorphy.php';
define('WORD_NOT_FOUND', 1);
if ($argc < 2) {
die("Usage {$argv['0']} WORD [LANG] [ENCODING]" . PHP_EOL);
}
$word = $argv[1];
$lang = $argc > 2 ? $argv[2] : 'ru_RU';
$dir = __DIR__ . '/../dicts/';
$dir .= $argc > 3 ? "/{$argv[3]}" : 'utf-8';
$opts = array('storage' => PHPMORPHY_STORAGE_FILE, 'predict_by_suffix' => true, 'predict_by_db' => true);
$morphy = new phpMorphy($dir, $lang, $opts);
$encoding = $morphy->getEncoding();
$formatter = new phpMorphy_Paradigm_Formatter();
$word = iconv('utf-8', $encoding, $word);
$word = mb_strtoupper($word, $encoding);
$result = $morphy->findWord($word);
$predict_text = 'DICT';
if ($morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_DB) {
$predict_text = 'PREDICT_BY_DB';
} else {
if ($morphy->getLastPredictionType() == phpMorphy::PREDICT_BY_SUFFIX) {
$predict_text = 'PREDICT_BY_SUFFIX';
}
}
echo "Paradigms for {$word}({$predict_text}):" . PHP_EOL;
if (false === $result) {
echo 'NOT FOUND' . PHP_EOL;
exit(WORD_NOT_FOUND);
public function getSearchString($string)
{
$query_string = '';
$s_words = array();
$materials = new Model_Material('groups');
//подключаем библиотеки
require_once DOCROOT . 'application/classes/PHPMorphy/src/common.php';
$dir = DOCROOT . 'application/classes/PHPMorphy/dicts/';
$lang = 'ru_RU';
$opts = array('storage' => PHPMORPHY_STORAGE_FILE);
try {
$morphy = new phpMorphy($dir, $lang, $opts);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . $e->getMessage());
}
$request = null;
$errors = null;
if (!empty($string)) {
// Очищаем от html-тегов и прочего
$search = trim(addslashes(strip_tags($string)));
$request = $search;
}
if (!empty($search)) {
// Обрабатываем данные как и в прошлом контроллере
if (mb_strlen($search, "UTF-8") > 1) {
preg_match_all('/([a-zа-яё0-9]+)/ui', mb_strtoupper($search, "UTF-8"), $search_words);
$words = $morphy->getPseudoRoot($search_words[1]);
foreach ($words as $k => $w) {
if (!$w) {
$w[0] = $k;
}
if (mb_strlen($w[0], "UTF-8") >= 2) {
$s_words[] = $w[0];
}
// else
// {
// // $s_words[] = $w[0];
// }
}
}
return $s_words;
}
}
<?php
error_reporting(E_ALL | E_STRICT);
// first we include phpmorphy library
require_once __DIR__ . '/../vendor/autoload.php';
// set some options
$opts = array('storage' => phpMorphy::STORAGE_FILE, 'with_gramtab' => false, 'predict_by_suffix' => true, 'predict_by_db' => true);
// Create phpMorphy instance
try {
$morphy = new phpMorphy(null, 'ru_RU', $opts);
//$morphy = new phpMorphy(new phpMorphy_FilesBundle(phpMorphy::getDefaultDictsDir(), 'ru_RU'), $opts);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . $e->getMessage());
}
// All words in dictionary in UPPER CASE, so don`t forget set proper locale
// Supported dicts and locales:
// *------------------------------*
// | Dict. language | Locale name |
// |------------------------------|
// | Russian | cp1251 |
// |------------------------------|
// | English | cp1250 |
// |------------------------------|
// | German | cp1252 |
// *------------------------------*
// $codepage = $morphy->getCodepage();
// setlocale(LC_CTYPE, array('ru_RU.CP1251', 'Russian_Russia.1251'));
// Hint: in this example words $word_one, $word_two are in russian language(cp1251 encoding)
$word_one = 'КОТ';
$word_two = 'СОБАКА';
echo "Testing single mode...\n";
<?php
// Подключите файл common.php. phpmorphy-0.3.2 - для версии 0.3.2,
// если используется иная версия исправьте код.
require_once 'c:/server/bin/PHP/lib/phpmorphy-0.3.7/src/common.php';
//global $morphy;
// Укажите путь к каталогу со словарями
$dir = 'c:/server/bin/PHP/lib/phpmorphy-0.3.7/dicts';
// Укажите, для какого языка будем использовать словарь.
// Язык указывается как ISO3166 код страны и ISO639 код языка,
// разделенные символом подчеркивания (ru_RU, uk_UA, en_EN, de_DE и т.п.)
$lang = 'ru_RU';
// Укажите опции
// Список поддерживаемых опций см. ниже
$opts = array('storage' => PHPMORPHY_STORAGE_MEM, 'predict_by_suffix' => TRUE);
// создаем экземпляр класса phpMorphy
// обратите внимание: все функции phpMorphy являются throwable т.е.
// могут возбуждать исключения типа phpMorphy_Exception (конструктор тоже)
try {
$morphy = new phpMorphy($dir, $lang, $opts);
print_r($morphy->lemmatize('яблоко', phpMorphy::NORMAL));
//exit(0);
} catch (phpMorphy_Exception $e) {
die('Error occured while creating phpMorphy instance: ' . $e->getMessage());
}
// далее под $morphy мы подразумеваем экземпляр класса phpMorphy
#!/usr/bin/env php
<?php
require_once __DIR__ . '/init.php';
$user_dict_xml = PHPMORPHY_DIR . '/../bin/user_dict.xml';
$outXmlFilePath = PHPMORPHY_DIR . '/../bin/out.xml';
$morphy_dict_dir = phpMorphy::getDefaultDictsDir();
$morphy_dict_lang = 'ru_RU';
$verbose = true;
/**
*
* @param string $dir
* @param string $lang
* @return phpMorphy_MorphyInterface
*/
function createMorphy($dir, $lang)
{
$opts = array('storage' => phpMorphy::STORAGE_MEM, 'predict_by_suffix' => true, 'predict_by_db' => true);
return new phpMorphy($dir, $lang, $opts);
}
$morphy = createMorphy($morphy_dict_dir, $morphy_dict_lang);
$encoding_converter = new phpMorphy_UserDict_EncodingConverter($morphy->getEncoding(), $morphy->isInUpperCase() ? MB_CASE_UPPER : MB_CASE_LOWER, 'utf-8', MB_CASE_UPPER);
//$errors_handler = new phpMorphy_UserDict_Log_ErrorsHandlerException();
$errors_handler = new phpMorphy_UserDict_Log_ErrorsHandlerPass();
$observer = new phpMorphy_UserDict_Log_CLI($verbose, $errors_handler, $encoding_converter);
$b = microtime(true);
if (1) {
phpMorphy_UserDict_XmlDiff_Generator::convertFromXmlToXml($user_dict_xml, $outXmlFilePath, $morphy, $observer, $encoding_converter);
}
$e = microtime(true);
printf("Time taken: %0.2f\n", $e - $b);
