/**
* Creates a new Pspell spell checker
*
* @param string $language the language used by this spell checker. This
* should be a two-letter ISO 639 language code
* followed by an optional two digit ISO 3166
* country code separated by a dash or underscore.
* For example, 'en', 'en-CA' and 'en_CA' are
* valid languages.
* @param string $personal_wordlist optional. The filename of the personal
* wordlist for this spell checker. If not
* specified, no personal wordlist is
* used. The personal wordlist may contain
* spellings for words that are correct
* but are not in the regular dictionary.
*
* @throws NateGoSearchException if the Pspell extension is not available.
* @throws NateGoSearchtException if a dictionary in the specified language
* could not be loaded.
*/
public function __construct($language, $path_to_data = '', $repl_pairs = '', $personal_wordlist = '')
{
if (!extension_loaded('pspell')) {
throw new NateGoSearchException('The Pspell PHP extension is ' . 'required for NateGoSearchPSpellSpellChecker.');
}
$config = pspell_config_create($language, '', '', 'utf-8');
pspell_config_mode($config, PSPELL_FAST);
if ($path_to_data != '') {
pspell_config_data_dir($config, $path_to_data);
pspell_config_dict_dir($config, $path_to_data);
}
if ($repl_pairs != '') {
pspell_config_repl($config, $repl_pairs);
}
if ($personal_wordlist != '') {
pspell_config_personal($config, $personal_wordlist);
if (file_exists($personal_wordlist) && fileowner($personal_wordlist) == posix_getuid()) {
// update permissions (-rw-rw----)
chmod($personal_wordlist, 0666);
}
$this->personal_wordlist = $personal_wordlist;
}
$this->dictionary = pspell_new_config($config);
if ($this->dictionary === false) {
throw new NateGoSearchException(sprintf("Could not create Pspell dictionary with language '%s'.", $this->language));
}
$this->loadBlacklistedSuggestions();
}
private function loadReplacements()
{
$path = $this->language_path . $this->language . ".repl";
if (file_exists($path)) {
//echo file_get_contents($path);
$this->pspell_config = pspell_config_create($this->language);
pspell_config_repl($this->pspell_config, $path);
$this->pspell = pspell_new_config($this->pspell_config);
}
}
/**
* Initialise our aSpell calling environment.
*
* @return array A tuple of environmental details (dictionary list,aspell call command,temporary file name,language being used)
*/
function aspell_init()
{
// Find the language
if (!isset($_REQUEST['dictionary']) || strlen(trim($_REQUEST['dictionary'])) < 1) {
$lang = function_exists('do_lang') ? do_lang('dictionary') : 'en_GB';
// Default to UK English (as per ocPortal)
} else {
$lang = $_REQUEST['dictionary'];
}
$aspellcommand = mixed();
$force_shell = false;
if (!function_exists('pspell_check') || $force_shell) {
if (str_replace(array('on', 'true', 'yes'), array('1', '1', '1'), strtolower(ini_get('safe_mode'))) == '1') {
exit('Spell Checker does not work with safe mode systems that do not have direct pspell support into PHP');
}
if (strpos(@ini_get('disable_functions'), 'shell_exec') !== false) {
exit('Spell Checker does not work on systems with shell_exec disabled that do not have direct pspell support into PHP');
}
// Our temporary spell check file
$temptext = tempnam(str_replace(array('on', 'true', 'yes'), array('1', '1', '1'), strtolower(ini_get('safe_mode'))) == '1' || @strval(ini_get('open_basedir')) != '' && preg_match('#(^|:|;)/tmp($|:|;|/)#', ini_get('open_basedir')) == 0 ? get_custom_file_base() . '/safe_mode_temp/' : '/tmp/', 'spell_');
if ($temptext === false) {
$temptext = tempnam(get_custom_file_base() . '/safe_mode_temp/', 'spell_');
}
// Find aspell
$aspell = 'aspell';
$aspell_args = '-a --lang=' . _filter_naughty_harsh($lang);
if (DIRECTORY_SEPARATOR == '\\') {
// See if there is a local install of aspell here
if (file_exists(dirname(__FILE__) . '\\aspell\\bin\\aspell.exe')) {
$aspell = dirname(__FILE__) . '\\aspell\\bin\\aspell.exe';
if (file_exists(dirname(__FILE__) . '\\aspell\\bin\\aspell_wrap.exe')) {
$aspell = dirname(__FILE__) . '\\aspell\\bin\\aspell_wrap.exe ' . dirname(__FILE__) . '\\aspell\\bin\\';
}
//$dic_dir=wrap_exec($aspell.' config dict-dir');
//$dicfil=preg_replace('/^.*\/lib\/(aspell\S*)\n.*/s','$1',$dic_dir);
//$aspell_args.=' --dict-dir='.$dicfil;
} else {
$aspell = 'C:\\Progra~1\\Aspell\\bin\\aspell.exe';
}
if (!file_exists($aspell)) {
exit('ASpell not installed in default locations.');
}
$aspell_version = wrap_exec($aspell . ' version');
} else {
// See if there is a local install of aspell here
if (file_exists(dirname(__FILE__) . '/aspell/bin/aspell')) {
putenv('PATH=' . dirname(__FILE__) . '/aspell/bin:' . getenv('PATH'));
putenv('LD_LIBRARY_PATH=' . dirname(__FILE__) . '/aspell/lib:' . getenv('LD_LIBRARY_PATH'));
//$dic_dir=wrap_exec($aspell.' config dict-dir');
//$dicfil=dirname(__FILE__).'/aspell/lib/'.preg_replace('/^.*\/lib\/(aspell\S*)\n.*/s','$1',$dic_dir);
//$aspell_args.=' --dict-dir='.$dicfil.' --add-filter-path='.$dicfil;
}
$aspell_version = wrap_exec($aspell . ' version');
}
if ($aspell_version === false) {
exit('ASpell would not execute. It is most likely not installed, or a security measure is in place, or file permissions are not correctly set. If on Windows, you may need to give windows\\system32\\cmd.exe execute permissions to the web user.');
}
// Old aspell doesn't know about encoding, which means that unicode will be broke, but we should at least let it try.
$a_ver = array();
preg_match('/really [aA]spell ([0-9]+)\\.([0-9]+)(?:\\.([0-9]+))?/i', $aspell_version, $a_ver);
if (!array_key_exists(1, $a_ver)) {
$a_ver[1] = '1';
}
if (!array_key_exists(2, $a_ver)) {
$a_ver[2] = '0';
}
if (!array_key_exists(3, $a_ver)) {
$a_ver[3] = '0';
}
$a_ver = array('major' => (int) $a_ver[1], 'minor' => (int) $a_ver[2], 'release' => (int) $a_ver[3]);
if ($a_ver['major'] >= 0 && $a_ver['minor'] >= 60) {
$aspell_args .= ' -H --encoding=utf-8';
} elseif (preg_match('/--encoding/', wrap_exec($aspell . ' 2>&1')) != 0) {
$aspell_args .= ' --mode=none --add-filter=sgml --encoding=utf-8';
} else {
$aspell_args .= ' --mode=none --add-filter=sgml';
}
$aspelldictionaries = $aspell . ' dump dicts';
$aspellcommand = $aspell . ' ' . $aspell_args . ' < ' . $temptext;
} else {
//list($lang,$spelling)=explode('_',$lang);
$spelling = '';
$temptext = NULL;
$aspelldictionaries = NULL;
}
// Personal dictionaries
global $SITE_INFO;
if (!isset($SITE_INFO)) {
require_once '../../../../info.php';
}
$cookie_member_id = $SITE_INFO['user_cookie'];
$p_dicts_name = array_key_exists($cookie_member_id, $_COOKIE) ? _filter_naughty_harsh($_COOKIE[$cookie_member_id]) : 'guest';
$p_dict_path = get_custom_file_base() . '/data_custom/spelling/personal_dicts' . DIRECTORY_SEPARATOR . $p_dicts_name;
if (!file_exists($p_dict_path)) {
mkdir($p_dict_path, 02770);
}
if (is_null($temptext)) {
list($lang_stub, ) = explode('_', $lang);
$charset = str_replace('ISO-', 'iso', str_replace('iso-', 'iso', do_lang('charset')));
if (DIRECTORY_SEPARATOR == '\\') {
$aspellcommand = @pspell_new_personal($p_dict_path . '/' . $lang_stub . '.pws', $lang, $spelling, '', $charset);
if ($aspellcommand === false) {
$aspellcommand = pspell_new_personal($p_dict_path . '/' . $lang_stub . '.pws', $lang, $spelling, '', $charset);
}
} else {
$aspellconfig = @pspell_config_create($lang, $spelling, '', $charset);
if ($aspellconfig === false) {
$aspellconfig = pspell_config_create('en', $spelling, '', $charset);
}
pspell_config_personal($aspellconfig, $p_dict_path . '/' . $lang_stub . '.pws');
pspell_config_repl($aspellconfig, $p_dict_path . '/' . $lang_stub . '.prepl');
$aspellcommand = @pspell_new_config($aspellconfig);
if ($aspellcommand === false && $lang != 'en') {
$aspellconfig = pspell_config_create('en', $spelling, '', $charset);
pspell_config_personal($aspellconfig, $p_dict_path . '/' . $lang_stub . '.pws');
pspell_config_repl($aspellconfig, $p_dict_path . '/' . $lang_stub . '.prepl');
$aspellcommand = pspell_new_config($aspellconfig);
}
}
if (is_null($aspellcommand)) {
exit;
}
}
return array($aspelldictionaries, $aspellcommand, $temptext, $lang);
}
function pspell_check($text, $lang = false)
{
global $charset;
if ($lang) {
$lang = $GLOBALS['LANG'];
}
$words = preg_split('/[\\W]+?/', $text);
$misspelled = $return = array();
$pspell_config = pspell_config_create($lang, "", "", $charset, PSPELL_NORMAL | PSPELL_RUN_TOGETHER);
//pspell_config_runtogether($pspell_config, true);
if (PSPELL_PWL) {
pspell_config_personal($pspell_config, PSPELL_PWL);
}
if (PSPELL_REPL) {
pspell_config_repl($pspell_config, PSPELL_REPL);
}
$pspell = pspell_new_config($pspell_config);
foreach ($words as $value) {
// SplitPagename $value
if (!pspell_check($pspell, $value)) {
$misspelled[] = $value;
}
}
foreach ($misspelled as $value) {
$return[$value] = pspell_suggest($pspell, $value);
}
return $return;
}
