The class takes the HTML markup of a page, exctracts the content and writes
it to the database (search index). It also provides a method to query the
seach index, returning the matching entries.
Usage:
Search::indexPage($objPage->row());
$result = Search::searchFor('keyword');
while ($result->next())
{
echo $result->url;
}
/**
* Prepare the output
*
* @return \PageModel
*
* @internal
*/
protected function prepare()
{
// Check the search index (see #3761)
\Search::removeEntry(\Environment::get('relativeRequest'));
// Find the matching root page
$objRootPage = $this->getRootPageFromUrl();
// Forward if the language should be but is not set (see #4028)
if (\Config::get('addLanguageToUrl')) {
// Get the request string without the script name
$strRequest = \Environment::get('relativeRequest');
// Only redirect if there is no language fragment (see #4669)
if ($strRequest != '' && !preg_match('@^[a-z]{2}(\\-[A-Z]{2})?/@', $strRequest)) {
// Handle language fragments without trailing slash (see #7666)
if (preg_match('@^[a-z]{2}(\\-[A-Z]{2})?$@', $strRequest)) {
$this->redirect(\Environment::get('request') . '/', 301);
} else {
if ($strRequest == \Environment::get('request')) {
$strRequest = $objRootPage->language . '/' . $strRequest;
} else {
$strRequest = \Environment::get('script') . '/' . $objRootPage->language . '/' . $strRequest;
}
$this->redirect($strRequest, 301);
}
}
}
// Look for a 404 page
$obj404 = \PageModel::find404ByPid($objRootPage->id);
// Die if there is no page at all
if (null === $obj404) {
throw new PageNotFoundException('Page not found');
}
// Forward to another page
if ($obj404->autoforward && $obj404->jumpTo) {
$objNextPage = \PageModel::findPublishedById($obj404->jumpTo);
if (null === $objNextPage) {
$this->log('Forward page ID "' . $obj404->jumpTo . '" does not exist', __METHOD__, TL_ERROR);
throw new ForwardPageNotFoundException('Forward page not found');
}
$this->redirect($this->generateFrontendUrl($objNextPage->row(), null, $objRootPage->language), $obj404->redirect == 'temporary' ? 302 : 301);
}
return $obj404;
}
/**
* Generate the module
*/
protected function compile()
{
// Mark the x and y parameter as used (see #4277)
if (isset($_GET['x'])) {
\Input::get('x');
\Input::get('y');
}
// Trigger the search module from a custom form
if (!isset($_GET['keywords']) && \Input::post('FORM_SUBMIT') == 'tl_search') {
$_GET['keywords'] = \Input::post('keywords');
$_GET['query_type'] = \Input::post('query_type');
$_GET['per_page'] = \Input::post('per_page');
}
$blnFuzzy = $this->fuzzy;
$strQueryType = \Input::get('query_type') ?: $this->queryType;
$strKeywords = trim(\Input::get('keywords'));
$this->Template->uniqueId = $this->id;
$this->Template->queryType = $strQueryType;
$this->Template->keyword = \StringUtil::specialchars($strKeywords);
$this->Template->keywordLabel = $GLOBALS['TL_LANG']['MSC']['keywords'];
$this->Template->optionsLabel = $GLOBALS['TL_LANG']['MSC']['options'];
$this->Template->search = \StringUtil::specialchars($GLOBALS['TL_LANG']['MSC']['searchLabel']);
$this->Template->matchAll = \StringUtil::specialchars($GLOBALS['TL_LANG']['MSC']['matchAll']);
$this->Template->matchAny = \StringUtil::specialchars($GLOBALS['TL_LANG']['MSC']['matchAny']);
$this->Template->action = ampersand(\Environment::get('indexFreeRequest'));
$this->Template->advanced = $this->searchType == 'advanced';
// Redirect page
if ($this->jumpTo && ($objTarget = $this->objModel->getRelated('jumpTo')) instanceof PageModel) {
/** @var PageModel $objTarget */
$this->Template->action = $objTarget->getFrontendUrl();
}
$this->Template->pagination = '';
$this->Template->results = '';
// Execute the search if there are keywords
if ($strKeywords != '' && $strKeywords != '*' && !$this->jumpTo) {
// Reference page
if ($this->rootPage > 0) {
$intRootId = $this->rootPage;
$arrPages = $this->Database->getChildRecords($this->rootPage, 'tl_page');
array_unshift($arrPages, $this->rootPage);
} else {
/** @var PageModel $objPage */
global $objPage;
$intRootId = $objPage->rootId;
$arrPages = $this->Database->getChildRecords($objPage->rootId, 'tl_page');
}
// HOOK: add custom logic (see #5223)
if (isset($GLOBALS['TL_HOOKS']['customizeSearch']) && is_array($GLOBALS['TL_HOOKS']['customizeSearch'])) {
foreach ($GLOBALS['TL_HOOKS']['customizeSearch'] as $callback) {
$this->import($callback[0]);
$this->{$callback[0]}->{$callback[1]}($arrPages, $strKeywords, $strQueryType, $blnFuzzy);
}
}
// Return if there are no pages
if (!is_array($arrPages) || empty($arrPages)) {
return;
}
$strCachePath = str_replace(TL_ROOT . DIRECTORY_SEPARATOR, '', \System::getContainer()->getParameter('kernel.cache_dir'));
$arrResult = null;
$strChecksum = md5($strKeywords . $strQueryType . $intRootId . $blnFuzzy);
$query_starttime = microtime(true);
$strCacheFile = $strCachePath . '/contao/search/' . $strChecksum . '.json';
// Load the cached result
if (file_exists(TL_ROOT . '/' . $strCacheFile)) {
$objFile = new \File($strCacheFile);
if ($objFile->mtime > time() - 1800) {
$arrResult = json_decode($objFile->getContent(), true);
} else {
$objFile->delete();
}
}
// Cache the result
if ($arrResult === null) {
try {
$objSearch = \Search::searchFor($strKeywords, $strQueryType == 'or', $arrPages, 0, 0, $blnFuzzy);
$arrResult = $objSearch->fetchAllAssoc();
} catch (\Exception $e) {
$this->log('Website search failed: ' . $e->getMessage(), __METHOD__, TL_ERROR);
$arrResult = array();
}
\File::putContent($strCacheFile, json_encode($arrResult));
}
$query_endtime = microtime(true);
// Sort out protected pages
if (\Config::get('indexProtected') && !BE_USER_LOGGED_IN) {
$this->import('FrontendUser', 'User');
foreach ($arrResult as $k => $v) {
if ($v['protected']) {
if (!FE_USER_LOGGED_IN) {
unset($arrResult[$k]);
} else {
$groups = \StringUtil::deserialize($v['groups']);
if (!is_array($groups) || empty($groups) || !count(array_intersect($groups, $this->User->groups))) {
unset($arrResult[$k]);
}
}
}
}
$arrResult = array_values($arrResult);
}
$count = count($arrResult);
$this->Template->count = $count;
$this->Template->page = null;
$this->Template->keywords = $strKeywords;
// No results
if ($count < 1) {
$this->Template->header = sprintf($GLOBALS['TL_LANG']['MSC']['sEmpty'], $strKeywords);
$this->Template->duration = substr($query_endtime - $query_starttime, 0, 6) . ' ' . $GLOBALS['TL_LANG']['MSC']['seconds'];
return;
}
$from = 1;
$to = $count;
// Pagination
if ($this->perPage > 0) {
$id = 'page_s' . $this->id;
$page = \Input::get($id) !== null ? \Input::get($id) : 1;
$per_page = \Input::get('per_page') ?: $this->perPage;
// Do not index or cache the page if the page number is outside the range
if ($page < 1 || $page > max(ceil($count / $per_page), 1)) {
throw new PageNotFoundException('Page not found: ' . \Environment::get('uri'));
}
$from = ($page - 1) * $per_page + 1;
$to = $from + $per_page > $count ? $count : $from + $per_page - 1;
// Pagination menu
if ($to < $count || $from > 1) {
$objPagination = new \Pagination($count, $per_page, \Config::get('maxPaginationLinks'), $id);
$this->Template->pagination = $objPagination->generate("\n ");
}
$this->Template->page = $page;
}
// Get the results
for ($i = $from - 1; $i < $to && $i < $count; $i++) {
/** @var FrontendTemplate|object $objTemplate */
$objTemplate = new \FrontendTemplate($this->searchTpl);
$objTemplate->url = $arrResult[$i]['url'];
$objTemplate->link = $arrResult[$i]['title'];
$objTemplate->href = $arrResult[$i]['url'];
$objTemplate->title = \StringUtil::specialchars($arrResult[$i]['title']);
$objTemplate->class = ($i == $from - 1 ? 'first ' : '') . ($i == $to - 1 || $i == $count - 1 ? 'last ' : '') . ($i % 2 == 0 ? 'even' : 'odd');
$objTemplate->relevance = sprintf($GLOBALS['TL_LANG']['MSC']['relevance'], number_format($arrResult[$i]['relevance'] / $arrResult[0]['relevance'] * 100, 2) . '%');
$objTemplate->filesize = $arrResult[$i]['filesize'];
$objTemplate->matches = $arrResult[$i]['matches'];
$arrContext = array();
$arrMatches = \StringUtil::trimsplit(',', $arrResult[$i]['matches']);
// Get the context
foreach ($arrMatches as $strWord) {
$arrChunks = array();
preg_match_all('/(^|\\b.{0,' . $this->contextLength . '}\\PL)' . str_replace('+', '\\+', $strWord) . '(\\PL.{0,' . $this->contextLength . '}\\b|$)/ui', $arrResult[$i]['text'], $arrChunks);
foreach ($arrChunks[0] as $strContext) {
$arrContext[] = ' ' . $strContext . ' ';
}
}
// Shorten the context and highlight all keywords
if (!empty($arrContext)) {
$objTemplate->context = trim(\StringUtil::substrHtml(implode('…', $arrContext), $this->totalLength));
$objTemplate->context = preg_replace('/(\\PL)(' . implode('|', $arrMatches) . ')(\\PL)/ui', '$1<mark class="highlight">$2</mark>$3', $objTemplate->context);
$objTemplate->hasContext = true;
}
$this->Template->results .= $objTemplate->parse();
}
$this->Template->header = vsprintf($GLOBALS['TL_LANG']['MSC']['sResults'], array($from, $to, $count, $strKeywords));
$this->Template->duration = substr($query_endtime - $query_starttime, 0, 6) . ' ' . $GLOBALS['TL_LANG']['MSC']['seconds'];
}
}
/**
* Index a page if applicable
*
* @param Response $objResponse
*/
public static function indexPageIfApplicable(Response $objResponse)
{
global $objPage;
if ($objPage === null) {
return;
}
// Index page if searching is allowed and there is no back end user
if (\Config::get('enableSearch') && $objPage->type == 'regular' && !BE_USER_LOGGED_IN && !$objPage->noSearch) {
// Index protected pages if enabled
if (\Config::get('indexProtected') || !FE_USER_LOGGED_IN && !$objPage->protected) {
$blnIndex = true;
// Do not index the page if certain parameters are set
foreach (array_keys($_GET) as $key) {
if (in_array($key, $GLOBALS['TL_NOINDEX_KEYS']) || strncmp($key, 'page_', 5) === 0) {
$blnIndex = false;
break;
}
}
if ($blnIndex) {
$arrData = array('url' => \Environment::get('base') . \Environment::get('relativeRequest'), 'content' => $objResponse->getContent(), 'title' => $objPage->pageTitle ?: $objPage->title, 'protected' => $objPage->protected ? '1' : '', 'groups' => $objPage->groups, 'pid' => $objPage->id, 'language' => $objPage->language);
\Search::indexPage($arrData);
}
}
}
}
