function topWords()
{
global $contacts, $thresholdWords, $up;
logMsg('USER', "Finding top {$thresholdWords} most often used words for each contact..");
foreach (array_keys($contacts) as $email) {
echo ' <span>.</span>';
flush();
$language = $contacts[$email]['language'];
if ($language != "english" && file_exists($up . "/stopwords/{$language}.txt")) {
$extraStopWords = " | grep -v -w -f {$up}/stopwords/{$language}.txt";
}
$f = sanitizeFilename($email);
$filename = $up . '/content/' . $f . '.txt';
$filenameWords = $up . '/content/' . $f . '_words.txt';
$filenameWordsStem = $up . '/content/' . $f . '_words_stem.txt';
chdir($up);
if (file_exists($filename)) {
$cmd = 'cat ' . $filename . ' | tr "A-Z" "a-z" | tr -c "[:alpha:]" " " | tr " " "\\n" | sort | uniq -c | sort | grep -v -w -f ' . $up . '/stopwords/english.txt | grep -E [a-z]{3,} | tr -d " *[:digit:]*\\t" | tail -n ' . $thresholdWords * 4 . ' > ' . $filenameWords;
logMsg('DEBUG', "Running CMD: {$cmd}");
shell_exec($cmd);
#detect language
$language = new LangDetect($filenameWords, -1);
$lang = $language->Analyze();
$languages = array_keys($lang);
$contacts[$email]['language'] = $languages[0];
$language = $languages[0];
$score = array_shift($score = $lang);
array_shift($lang);
foreach ($lang as $l => $lscore) {
if ($lscore - $score > 7000) {
break;
}
if ($l != 'english') {
unset($language);
$language = $l;
break;
}
}
if ($language != 'english') {
logMsg('DEBUG', "Language for {$email} is " . $contacts[$email]['language'] . " (but removing also {$language} stopwords)");
} else {
logMsg('DEBUG', "Language for {$email} is " . $contacts[$email]['language']);
}
if ($language != 'english') {
$cmd = 'cat ' . $filenameWords . ' | tr "A-Z" "a-z" | tr -c "[:alpha:]" " " | tr " " "\\n" | sort | uniq -c | sort | grep -v -w -f ' . $up . '/stopwords/' . $contacts[$email]['language'] . '.txt | grep -E [a-z]{3,} | tr -d " *[:digit:]*\\t" | tail -n ' . $thresholdWords . ' > ' . $filenameWords;
logMsg('DEBUG', "Running CMD: {$cmd}");
shell_exec($cmd);
}
$contacts[$email]['words'] = array_reverse(array_trim(file($filenameWords)));
if ($language == 'english' || $language == 'swedish') {
$languageShort = substr($language, 0, 2);
$cmd = $up . '/cstlemma/bin/vc2008/cstlemma.exe -e1 -L -f ' . $up . '/cstlemma/flexrules_' . $languageShort . ' -t- -c"$B" -B"$w\\n" < ' . $filenameWords . ' > ' . $filenameWordsStem;
logMsg('DEBUG', "Running CMD: {$cmd}");
shell_exec($cmd);
$contacts[$email]['wordsStem'] = array_reverse(array_trim(file($filenameWordsStem)));
array_pop($contacts[$email]['wordsStem']);
}
}
}
logMsg('USER', "Done!");
}
function setLanguage($config)
{
$locales = array();
$lang_cookie = '';
# Получаем локаль из куков
if (isset($_COOKIE['lang'])) {
$lang_cookie = $_COOKIE['lang'];
}
#Запись выбранного языка в куки
if (isset($_GET['lang'])) {
$lang_cookie = $_GET['lang'];
SetCookie("lang", $lang_cookie, time() + 1000 * 24 * 60 * 60);
}
if (isset($config['config']['locales'])) {
foreach ($config['config']['locales'] as $k => $v) {
if (stristr($k, 'name_') !== false) {
$key_name = str_replace('name_', '', $k);
$val_en = 'enable_' . $key_name;
$val_file = 'file_' . $key_name;
$locales[$v] = array('file' => $config['config']['locales'][$val_file], 'enable' => $config['config']['locales'][$val_en]);
}
}
}
# Базовый язык ru-RU (на случай если по умолчанию язык недоступен или не включен)
$defaultNameLocale = 'ru_RU';
$defaultFileLocale = 'original';
# Локализациия по умолчанию.
if (isset($config['config']['locales_options']['default']) and $config['config']['locales_options']['default'] != '') {
$temp_default = $config['config']['locales_options']['default'];
# Проверяем активированна ли локализация по умолчанию
if (isset($locales[$temp_default])) {
$defaultNameLocale = $temp_default;
$defaultFileLocale = $locales[$temp_default]['file'];
}
}
# Если включен автодетект
if (isset($config['config']['locales_options']['autoDetect']) and $config['config']['locales_options']['autoDetect'] == 1) {
$langDetect = new LangDetect();
$langs = array('ru_RU' => array('ru'), 'uk_UA' => array('uk'), 'en_GB' => array('en'));
$detectNameLocale = $langDetect->getBestMatch($defaultNameLocale, $langs);
# Проверяем активированна ли определенная локаль
if (isset($locales[$detectNameLocale]) and $locales[$detectNameLocale]['enable'] == 1) {
$nameLocale = $detectNameLocale;
$fileLoacale = $locales[$temp_default]['file'];
} else {
$nameLocale = $defaultNameLocale;
$fileLoacale = $defaultFileLocale;
}
} else {
$nameLocale = $defaultNameLocale;
$fileLoacale = $defaultFileLocale;
}
#Если язык из куков включен то используем его
if (isset($locales[$lang_cookie]) and $locales[$lang_cookie]['enable'] == 1) {
$nameLocale = $lang_cookie;
}
T_setlocale(LC_MESSAGES, $nameLocale);
T_bindtextdomain($fileLoacale, './data/locale');
T_bind_textdomain_codeset($fileLoacale, 'UTF-8');
T_textdomain($fileLoacale);
}
