function bench_enchant($words)
{
// TODO: check return values!!!
echo "Bench Enchant: ";
$tag = 'ru_RU';
$r = enchant_broker_init();
if (!enchant_broker_dict_exists($r, $tag)) {
echo "{$tag} dict not supported by enchant\n";
return false;
}
$d = enchant_broker_request_dict($r, $tag);
$not_found = 0;
$b = microtime(true);
foreach ($words as $word) {
// if(false === enchant_dict_quick_check($d, $word/*, $sugg*/)) { // this cause segfault
if (false === enchant_dict_check($d, $word)) {
enchant_dict_suggest($d, $word);
$not_found++;
}
}
$e = microtime(true);
printf("time = %0.2f sec, words per second = %0.2f, not found = %d\n", $e - $b, count($words) / ($e - $b), $not_found);
enchant_broker_free_dict($d);
enchant_broker_free($r);
}
/**
* Spellchecks an array of words.
*
* @param String $lang Selected language code (like en_US or de_DE). Shortcodes like "en" and "de" work with enchant >= 1.4.1
* @param Array $words Array of words to check.
* @return Name/value object with arrays of suggestions.
*/
public function getSuggestions($lang, $words)
{
$suggestions = array();
$enchant = enchant_broker_init();
$config = $this->getConfig();
if (isset($config["enchant_dicts_path"])) {
enchant_broker_set_dict_path($enchant, ENCHANT_MYSPELL, $config["enchant_dicts_path"]);
enchant_broker_set_dict_path($enchant, ENCHANT_ISPELL, $config["enchant_dicts_path"]);
}
if (!enchant_broker_describe($enchant)) {
throw new Exception("Enchant spellchecker not find any backends.");
}
$lang = $this->normalizeLangCode($enchant, $lang);
if (enchant_broker_dict_exists($enchant, $lang)) {
$dict = enchant_broker_request_dict($enchant, $lang);
foreach ($words as $word) {
if (!enchant_dict_check($dict, $word)) {
$suggs = enchant_dict_suggest($dict, $word);
if (!is_array($suggs)) {
$suggs = array();
}
$suggestions[$word] = $suggs;
}
}
enchant_broker_free_dict($dict);
enchant_broker_free($enchant);
} else {
enchant_broker_free($enchant);
throw new Exception("Enchant spellchecker could not find dictionary for language: " . $lang);
}
return $suggestions;
}
/**
* Returns suggestions for a specific word.
*
* @param String $lang Selected language code (like en_US or de_DE). Shortcodes like "en" and "de" work with enchant >= 1.4.1
* @param String $word Specific word to get suggestions for.
* @return Array of suggestions for the specified word.
*/
function &getSuggestions($lang, $word)
{
$r = enchant_broker_init();
$suggs = array();
if (enchant_broker_dict_exists($r, $lang)) {
$d = enchant_broker_request_dict($r, $lang);
$suggs = enchant_dict_suggest($d, $word);
enchant_broker_free_dict($d);
} else {
}
enchant_broker_free($r);
return $suggs;
}
/**
* Returns suggestions for a specific word.
*
* @param String $lang Selected language code (like en_US or de_DE). Shortcodes like "en" and "de" work with enchant >= 1.4.1
* @param String $word Specific word to get suggestions for.
* @return Array of suggestions for the specified word.
*/
function &getSuggestions($lang, $word)
{
$r = enchant_broker_init();
$suggs = array();
if (enchant_broker_dict_exists($r, $lang)) {
$d = enchant_broker_request_dict($r, $lang);
$suggs = enchant_dict_suggest($d, $word);
enchant_broker_free_dict($d);
} else {
$this->throwError("Language not installed");
}
enchant_broker_free($r);
return $suggs;
}
/**
* Returns suggestions for a specific word.
*
* @param String $lang Selected language code (like en_US or de_DE). Shortcodes like "en" and "de" work with enchant >= 1.4.1
* @param String $word Specific word to get suggestions for.
* @return Array of suggestions for the specified word.
*/
function &getSuggestions($lang, $word)
{
$r = enchant_broker_init();
if (enchant_broker_dict_exists($r, $lang)) {
$d = enchant_broker_request_dict($r, $lang);
$suggs = enchant_dict_suggest($d, $word);
// enchant_dict_suggest() sometimes returns NULL
if (!is_array($suggs)) {
$suggs = array();
}
enchant_broker_free_dict($d);
} else {
$suggs = array();
}
enchant_broker_free($r);
return $suggs;
}
/**
* @param array|Word[] $words
* @param array|string[] $locales
*
* @return SpellResult
*/
public function check($words, array $locales)
{
$misspelledWords = array();
$enchantResource = enchant_broker_init();
/*$bprovides = enchant_broker_describe($r);
echo "Current broker provides the following backend(s):\n";
print_r($bprovides);*/
/*$dicts = enchant_broker_list_dicts($r);
print_r($dicts);*/
$dictionaries = array();
foreach ($locales as $locale) {
if (!enchant_broker_dict_exists($enchantResource, $locale)) {
// TODO handle and log error
continue;
}
$dictionaries[$locale] = enchant_broker_request_dict($enchantResource, $locale);
}
//$dprovides = enchant_dict_describe($dictionary);
//echo "dictionary $tag provides:\n";
foreach ($words as $word) {
$checked = false;
$suggests = array();
foreach ($dictionaries as $locale => $dictionary) {
$suggests[$locale] = array();
$checked = $checked || enchant_dict_quick_check($dictionary, $word->getWord(), $suggests[$locale]);
}
$word->setChecked($checked);
if (!$word->isChecked()) {
$word->setSuggests($suggests);
$misspelledWords[] = $word;
}
}
foreach ($dictionaries as $dictionary) {
enchant_broker_free_dict($dictionary);
}
enchant_broker_free($enchantResource);
$spellResult = new SpellResult();
$spellResult->setCountOfWords(count($words));
$spellResult->setMisspelledWords($misspelledWords);
return $spellResult;
}
$shortest = $lev;
}
}
$keywords_suggest[] = '<b class="search-word-suggest">' . $closest . '</b>';
$keywords_suggest_plain[] = $closest;
} else {
$keywords_suggest[] = '<b class="search-word-correct">' . $word . '</b>';
$keywords_suggest_plain[] = $word;
}
$word = strtok(" \t\n");
}
$keywords_suggest_plain_str = implode(' ', $keywords_suggest_plain);
$search_result_info .= '<a class="search-suggestion-link" href="./index.php?keywords=' . urlencode($keywords_suggest_plain_str) . '&search=Search">';
$search_result_info .= implode(' ', $keywords_suggest);
$search_result_info .= '</a>?</div>';
enchant_broker_free_dict($dict);
}
}
if (isset($biblio_list) && isset($sysconf['enable_xml_result']) && $sysconf['enable_xml_result']) {
$search_result_info .= '<div><a href="index.php?resultXML=true&' . $_SERVER['QUERY_STRING'] . '" class="xmlResultLink" target="_blank" title="View Result in XML Format" style="clear: both;">XML Result</a></div>';
}
}
// check if we are on xml resultset mode
if ((isset($_GET['rss']) || isset($_GET['resultXML'])) && $sysconf['enable_xml_result']) {
// get document list but don't output the result
$biblio_list->getDocumentList(false);
if ($biblio_list->num_rows > 0) {
// send http header
header('Content-Type: text/xml');
echo '<?xml version="1.0" encoding="UTF-8" ?>' . "\n";
if (isset($_GET['rss'])) {
/**
* Return a collection of suggestion corresponding a query
*
* @param string $query
* @return ArrayCollection An array collection of SearchEngineSuggestion
*/
private function getSuggestions($query, SearchEngineOptions $options)
{
// First we split the query into simple words
$words = explode(" ", $this->cleanupQuery(mb_strtolower($query)));
$tmpWords = [];
foreach ($words as $word) {
if (trim($word) === '') {
continue;
}
$tmpWords[] = $word;
}
$words = array_unique($tmpWords);
$altVersions = [];
foreach ($words as $word) {
$altVersions[$word] = [$word];
}
// As we got words, we look for alternate word for each of them
if (function_exists('enchant_broker_init') && $options->getLocale()) {
$broker = enchant_broker_init();
if (enchant_broker_dict_exists($broker, $options->getLocale())) {
$dictionnary = enchant_broker_request_dict($broker, $options->getLocale());
foreach ($words as $word) {
if (enchant_dict_check($dictionnary, $word) == false) {
$suggs = array_merge(enchant_dict_suggest($dictionnary, $word));
}
$altVersions[$word] = array_unique($suggs);
}
enchant_broker_free_dict($dictionnary);
}
enchant_broker_free($broker);
}
/**
* @todo enhance the trigramm query, as it could be sent in one batch
*/
foreach ($altVersions as $word => $versions) {
$altVersions[$word] = array_unique(array_merge($versions, $this->get_sugg_trigrams($word, $options)));
}
// We now build an array of all possibilities based on the original query
$queries = [$query];
foreach ($altVersions as $word => $versions) {
$tmp_queries = [];
foreach ($versions as $version) {
foreach ($queries as $alt_query) {
$tmp_queries[] = $alt_query;
$tmp_queries[] = str_replace($word, $version, $alt_query);
}
$tmp_queries[] = str_replace($word, $version, $query);
}
$queries = array_unique(array_merge($queries, $tmp_queries));
}
$suggestions = [];
$max_results = 0;
foreach ($queries as $alt_query) {
$results = $this->sphinx->Query($alt_query, $this->getQueryIndex($alt_query, $options));
if ($results !== false && isset($results['total_found'])) {
if ($results['total_found'] > 0) {
$max_results = max($max_results, (int) $results['total_found']);
$suggestions[] = new SearchEngineSuggestion($query, $alt_query, (int) $results['total_found']);
}
}
}
usort($suggestions, ['self', 'suggestionsHitSorter']);
$tmpSuggestions = new ArrayCollection();
foreach ($suggestions as $key => $suggestion) {
if ($suggestion->getHits() < $max_results / 100) {
continue;
}
$tmpSuggestions->add($suggestion);
}
return $tmpSuggestions;
}
/**
* Wiki-specific search suggestions using enchant library.
* Use SphinxSearch_setup.php to create the dictionary
*/
function suggestWithEnchant() {
if (!function_exists('enchant_broker_init')) {
return;
}
$broker = enchant_broker_init();
enchant_broker_set_dict_path($broker, ENCHANT_MYSPELL, dirname( __FILE__ ));
if ( enchant_broker_dict_exists( $broker, 'sphinx' ) ) {
$dict = enchant_broker_request_dict( $broker, 'sphinx' );
$suggestion_found = false;
$full_suggestion = '';
foreach ( $this->mTerms as $word ) {
if ( !enchant_dict_check($dict, $word) ) {
$suggestions = enchant_dict_suggest($dict, $word);
while ( count( $suggestions ) ) {
$candidate = array_shift( $suggestions );
if ( strtolower($candidate) != strtolower($word) ) {
$word = $candidate;
$suggestion_found = true;
break;
}
}
}
$full_suggestion .= $word . ' ';
}
enchant_broker_free_dict( $dict );
if ($suggestion_found) {
$this->mSuggestion = trim( $full_suggestion );
}
}
enchant_broker_free( $broker );
}
/**
* @fn convert
* @memberof Converter
* @brief Public function to convert encoding of text contents.
* @param $text Source text content to convert.
* @param $options Options array
* Available Options -
* -# output (Output encoding - default is myanmar3)
* -# input_font (Input encoding)
* -# encoding (encoding type - ascii or utf8 - default is utf8 or none)
* -# spelling_check (Enable spelling checking when converting ascii fonts.)
* -# text-only (Source content is plain text or other format such as html, php etc. - default is false)
* -# en_zwsp (enable or disable adding Zero-Width-Space in converted result.)
* -# exceptions (commas seperated list of words or phrase to ignore from conversion process.)
* -# suggestion (enable to use user suggested words list.)
*
* @returns string $this->text Return converted content.
*/
public function convert($text, $options = array())
{
foreach ($options as $option_name => $option_value) {
${$option_name} = $option_value;
//extract $options and set $option_name as variable name and $option_value as variable value.
}
if ($input_font === '' || $input_font == 'auto') {
/**
* @var $input
*/
$input = $this->enc_test($text);
} else {
/**
* @var $input
*/
$input = $input_font;
}
/**
* @var $ext_dir
*/
$ext_dir = 'ext/';
$rules_dir = 'rules/' . $output;
/**
* @var $ext_file
*/
if (isset($encoding) && $encoding != '') {
$ext_file = $input . '-rules-' . $encoding . '.php';
} else {
$ext_file = $input . '-rules.php';
}
if (file_exists($this->ROOT_DIR . $ext_dir . $rules_dir . '/' . $ext_file)) {
require $this->ROOT_DIR . $ext_dir . $rules_dir . '/' . $ext_file;
include $this->ROOT_DIR . $ext_dir . $rules_dir . '/correction.php';
include $this->ROOT_DIR . $ext_dir . $rules_dir . '/zwsp.php';
} else {
$correction = array();
$final_text = $this->ROOT_DIR . $ext_dir . $rules_dir . '/' . $ext_file . " Fatal Error: Your converting rules file cannot be found!\nFile is missing or never existed. Try another choice or contact developer!";
//return ;
}
if (!isset($order)) {
/**
* @var array $order Character ordering rules array.
*/
$order = array();
//if $order is not defined, set it to empty array.
}
/**
* @var bool $en_zwsp enable/disable adding zwsp.
*/
if ($en_zwsp == true) {
/**
* @var $final_regex_array
*/
$final_regex_array = array_merge_recursive($order, $correction, $zwsp);
//merge all defined regular expression arrays.
} else {
/**
* @var $final_regex_array
*/
$final_regex_array = array_merge_recursive($order, $correction);
//merge all defined regular expression arrays without zwsp array.
}
if ($input === $output) {
foreach ($final_regex_array as $key => $value) {
/**
* @var $final_text
*/
$final_text = preg_replace('/' . $key . '/u', $value, $text);
}
} else {
if (isset($conv_rules)) {
if ($encoding == 'ascii') {
/**
* @fn trim_value
* @brief
* @param $value
* @returns
*/
function trim_value(&$value)
{
/**
* @var $value
*/
$value = trim($value);
}
/**
* @fn ucwords_value
* @brief
* @param $value
* @returns
*/
function ucwords_value(&$value)
{
/**
* @var $value
*/
$value = ucwords($value);
}
/**
* @fn space_on_short_words
* @brief
* @param $value
* @returns
*/
function space_on_short_words(&$value)
{
/**
* @var $value
*/
$value = preg_replace('/^[\\d\\w]{1,3}$/u', ' $0 ', $value);
}
if (isset($spelling_check) && false !== $spelling_check) {
/**
* @var $stripped_text
*/
$stripped_text = strip_tags($text);
/**
* @var $paragraph
*/
$paragraph = preg_split("/[\\s,]+/s", $stripped_text);
array_walk($paragraph, 'trim_value');
/**
* @var $words_array
*/
$words_array = array_unique($paragraph);
array_multisort($words_array);
if (function_exists('enchant_broker_init')) {
/**
* @var $tag
*/
$tag = 'en_US';
/**
* @var $r
*/
$r = enchant_broker_init();
/**
* @var $bprovides
*/
$bprovides = enchant_broker_describe($r);
/**
* @var $dicts
*/
$dicts = enchant_broker_list_dicts($r);
if (enchant_broker_dict_exists($r, $tag)) {
/**
* @var $d
*/
$d = enchant_broker_request_dict($r, $tag);
/**
* @var $dprovides
*/
$dprovides = enchant_dict_describe($d);
foreach ($words_array as $word) {
if (!empty($word)) {
/**
* @var $wordcorrect
*/
$wordcorrect = enchant_dict_check($d, $word);
if ($wordcorrect) {
$word = preg_replace('/^\\d+$/u', '"$0"', $word);
$word = preg_replace('/^[\\d\\w]{1,3}$/u', ' $0 ', $word);
$english_words_array[$word] = $word;
}
}
}
enchant_broker_free_dict($d);
}
enchant_broker_free($r);
} else {
include $this->ROOT_DIR . $ext_dir . 'dic/dictionary_array.php';
array_walk($words_array, 'space_on_short_words');
foreach ($words_array as $word) {
if (!empty($word)) {
/**
* @var $plural_ies
*/
$plural_ies = preg_match('/(\\w+)(ies)|(\\w+)(s)/', $word, $plural_match_ies);
if (!empty($plural_match_ies)) {
array_walk($plural_match_ies, 'space_on_short_words');
if ($plural_match_ies[2] == ' ies ') {
/**
* @var $singular
*/
$singular = $plural_match_ies[1] . 'y';
} elseif ($plural_match_ies[4] == ' s ') {
/**
* @var $singular
*/
$singular = $plural_match_ies[3];
}
if (in_array($singular, $dictionary) || in_array(strtolower($singular), $dictionary)) {
$plural_array[$plural_match_ies[0]] = $plural_match_ies[0];
}
}
if (in_array($word, $dictionary) || in_array(strtolower($word), $dictionary)) {
$english_words_array[$word] = $word;
}
}
}
}
/**
* @var $english_words
*/
$english_words = array();
if (isset($english_words_array) && !empty($english_words_array)) {
/**
* @var $english_words
*/
$english_words = $english_words_array;
}
if (isset($plural_array) && !empty($plural_array)) {
/**
* @var $english_words
*/
$english_words = array_merge($english_words, $plural_array);
}
/**
* @var $english_words
*/
$english_words = array_unique($english_words);
}
/**
* @var $generated_array
*/
$generated_array = array();
if (true !== $text_only) {
preg_match_all('/<(.*)>/uU', $text, $html_tags);
foreach ($html_tags[0] as $html_tag) {
if (!empty($html_tag)) {
$generated_array[$html_tag] = $html_tag;
}
}
preg_match_all('/<(style|script)(.*)<\\/(style|script)>/uUs', $text, $script_tags);
foreach ($script_tags[0] as $script_tag) {
if (!empty($script_tag)) {
$generated_array[$script_tag] = $script_tag;
}
}
}
/**
* @var $user_content_array
*/
$user_content_array = array();
if (isset($exceptions)) {
/**
* @var $exceps_array
*/
$exceps_array = explode(',', $exceptions);
if (!empty($exceps_array)) {
foreach ($exceps_array as $ignore_list) {
if (!empty($ignore_list) && strlen($ignore_list) > 4) {
$generated_array[$ignore_list] = $ignore_list;
$user_content_array[] = $ignore_list;
}
}
}
}
if (isset($suggested) && true === $suggested) {
if (file_exists($this->ROOT_DIR . $ext_dir . 'dic/userdic.dic')) {
/**
* @var $user_dic
*/
$user_dic = file($this->ROOT_DIR . $ext_dir . 'dic/userdic.dic', FILE_SKIP_EMPTY_LINES);
array_walk($user_dic, 'trim_value');
foreach ($user_dic as $user_word) {
if (!empty($user_word)) {
$generated_array[$user_word] = $user_word;
$user_content_array[] = $user_word;
}
}
}
}
if (!empty($user_content_array)) {
/**
* @var $user_content
*/
$user_content = "";
array_walk($user_content_array, 'trim_value');
asort($user_content_array);
/**
* @var $user_content_array
*/
$user_content_array = array_unique($user_content_array);
foreach ($user_content_array as $phrase) {
$user_content .= "{$phrase}\n";
}
/**
* @var $userdic_file
*/
$userdic_file = $this->ROOT_DIR . $ext_dir . 'dic/userdic.dic';
/**
* @var $uaf
*/
$uaf = fopen($userdic_file, 'w') or die("File is not writable or directory does not exist.");
fwrite($uaf, $user_content);
fclose($uaf);
}
/**
* @var $conv_array
*/
$conv_array = $conv_rules;
if (!empty($generated_array)) {
/**
* @var $conv_array
*/
$conv_array = array_merge($generated_array, $conv_array);
}
if (isset($english_words) && !empty($english_words)) {
/**
* @var $conv_array
*/
$conv_array = array_merge($english_words, $conv_array);
}
/**
* @var $final_text
*/
$final_text = strtr($text, $conv_array);
} else {
/**
* @var $final_text
*/
$final_text = strtr($text, $conv_rules);
}
foreach ($final_regex_array as $key => $value) {
$reg_count = 0;
/**
* @var $final_text
*/
$final_text = preg_replace('/' . $key . '/us', $value, $final_text, -1, $reg_count);
// print($reg_count.'<br>');
}
} else {
foreach ($final_regex_array as $key => $value) {
$reg_count = 0;
/**
* @var $final_text
*/
$final_text = preg_replace('/' . $key . '/us', $value, $text, -1, $reg_count);
// print($reg_count.'<br>');
}
}
}
/**
* @var $text
*/
$this->text = $final_text;
return $this->text;
}