function rebuild_search_indexes()
{
global $success_msg;
global $error_msg;
global $warning_msg;
global $all_settings;
$index_folder = get_setting('search_indexes_folder', $all_settings);
try {
$index = new Zend_Search_Lucene($index_folder, true);
setlocale(LC_CTYPE, 'en_US');
foreach (get_all_html_files(dirname(__FILE__)) as $html_file => $html_url) {
if (can_index_html_file($html_file)) {
$file_content = file_get_contents($html_file);
$file_content = '<html>' . strstr($file_content, '<head');
$doc = Zend_Search_Lucene_Document_Html::loadHTML($file_content);
$doc->addField(Zend_Search_Lucene_Field::Text('url', $html_url, 'UTF-8'));
$index->addDocument($doc);
flush();
}
}
$broken_urls = array();
foreach (get_dynamic_urls(get_setting('search_dynamic_pages', $all_settings)) as $url) {
$headers = get_headers($url);
if (strrpos($headers[0], '200')) {
$content = file_get_contents($url);
$content = '<html>' . strstr($content, '<head');
$doc = Zend_Search_Lucene_Document_Html::loadHTML($content);
$doc->addField(Zend_Search_Lucene_Field::Text('url', $url, 'UTF-8'));
$index->addDocument($doc);
flush();
} else {
array_push($broken_urls, $url);
}
}
if (file_exists($index_folder)) {
if (count($broken_urls) > 0) {
$warning_msg = '<p>The website was successfully indexed, but the following URL\'s were skipped because they are broken:</p>';
$warning_msg .= '<ul class="disc">';
foreach ($broken_urls as $broken_url) {
$warning_msg .= '<li><a href="' . $broken_url . '">' . $broken_url . '</a></li>';
}
$warning_msg .= '</ul>';
$warning_msg .= '<p>Please remove them from the "List of dynamic pages" field.</p>';
} else {
$success_msg = 'The website was successfully indexed.';
}
} else {
$error_msg = 'An error occurred during the website indexing. The error message is: the folder that stores the website indexes couldn\'t be created';
}
} catch (Exception $e) {
$error_msg = 'An error occurred during the website indexing. The error message is: ' . $e->getMessage();
}
}
function can_index_website_file($html_file)
{
global $all_settings;
foreach (get_dynamic_urls(get_setting('search_exclude_from_indexing', $all_settings)) as $excluded_file) {
$excluded_file = rtrim(str_replace("\\", "/", $excluded_file), '/');
$pos = strpos($html_file, $excluded_file);
if ($pos && $pos >= 0) {
return false;
}
}
$file_content = file_get_contents($html_file);
$pos = strpos($file_content, '@skip-indexing');
if ($pos && $pos >= 0) {
return false;
}
return true;
}
