/**
* DOCUMENT ME
* @param mixed $arguments
* @param mixed $options
*/
protected function execute($arguments = array(), $options = array())
{
// initialize the database connection
$databaseManager = new sfDatabaseManager($this->configuration);
$connection = $databaseManager->getDatabase($options['connection'] ? $options['connection'] : null)->getConnection();
// PDO connection not so useful, get the doctrine one
$conn = Doctrine_Manager::connection();
$context = sfContext::createInstance($this->configuration);
if (isset($options['table'])) {
$indexes = array($options['table']);
} else {
$indexes = sfConfig::get('app_aToolkit_indexes', array());
}
$count = 0;
foreach ($indexes as $index) {
$table = Doctrine::getTable($index);
if ($options['verbose']) {
echo "Optimizing {$index}\n";
}
aZendSearch::optimizeLuceneIndex($table);
}
if ($options['verbose']) {
echo "Success!\n";
}
}
protected function execute($arguments = array(), $options = array())
{
// Memory usage is a bit high here because we look at every page, and the Rackspace Cloud
// environment has a very low default memory limit for their ersatz "cron jobs."
// TODO: prioritize a low-memory solution for rebuild-search-index, which will be
// necessary for large sites anyway
ini_set('memory_limit', '256M');
// initialize the database connection
$databaseManager = new sfDatabaseManager($this->configuration);
$connection = $databaseManager->getDatabase($options['connection'] ? $options['connection'] : null)->getConnection();
// Initialize the context, which loading use of helpers, notably url_for
// First set config vars so that reasonable siteless-but-rooted URLs can be generated
// TODO: think about ways to make this work for people who like frontend_dev.php etc., although
// we're doing rather well with an index.php that suits each environment
sfConfig::set('sf_no_script_name', true);
$_SERVER['PHP_SELF'] = '';
$_SERVER['SCRIPT_NAME'] = '';
$context = sfContext::createInstance($this->configuration);
if (isset($options['table'])) {
$indexes = array($options['table']);
} else {
$indexes = sfConfig::get('app_aToolkit_indexes', array());
}
$count = 0;
foreach ($indexes as $index) {
$table = Doctrine::getTable($index);
if ($index === 'aPage') {
aZendSearch::purgeLuceneIndex($table);
// We're about to request updates of all page/culture combinations. Don't
// add that to an existing workload which could result in a huge pileup of
// repeat requests if someone starts interrupting this task and trying again, etc.
$this->query('DELETE FROM a_lucene_update');
$pages = Doctrine::getTable('aPage')->createQuery('p')->innerJoin('p.Areas a')->execute(array(), Doctrine::HYDRATE_ARRAY);
foreach ($pages as $page) {
$cultures = array();
foreach ($page['Areas'] as $area) {
$cultures[$area['culture']] = true;
}
$cultures = array_keys($cultures);
foreach ($cultures as $culture) {
$this->query('INSERT INTO a_lucene_update (page_id, culture) VALUES (:page_id, :culture)', array('page_id' => $page['id'], 'culture' => $culture));
}
}
while (true) {
$result = $this->query('SELECT COUNT(id) AS total FROM a_lucene_update');
$count = $result[0]['total'];
if ($count == 0) {
break;
}
$this->logSection('toolkit', "{$count} pages remain to be indexed, starting another update pass...");
$this->update();
}
} else {
// We don't have a deferred update feature for other tables,
// so we'll have to get them done in the memory available
$table->rebuildLuceneIndex();
}
$this->logSection('toolkit', sprintf('Index for "%s" rebuilt', $index));
}
}
/**
* Executes feedback action
*
*/
public function executeFeedback(sfRequest $request)
{
$section = $request->getParameter('section', false);
$this->form = new aFeedbackForm($section);
$this->feedbackSubmittedBy = false;
$this->failed = false;
sfContext::getInstance()->getConfiguration()->loadHelpers(array('Tag', 'Url'));
if ($request->isMethod('post')) {
$this->form->bind($request->getParameter('feedback'), $request->getFiles('feedback'));
// $this->form->bind(array_merge($request->getParameter('feedback'), array('captcha' => $request->getParameter('captcha'))), $request->getFiles('feedback'));
if ($this->form->isValid()) {
$feedback = $this->form->getValues();
$feedback['browser'] = $_SERVER['HTTP_USER_AGENT'];
try {
aZendSearch::registerZend();
$mail = new Zend_Mail();
$mail->setBodyText($this->getPartial('feedbackEmailText', array('feedback' => $feedback)))->setFrom($feedback['email'], $feedback['name'])->addTo(sfConfig::get('app_aFeedback_email_auto'))->setSubject($this->form->getValue('subject', 'New aBugReport submission'));
if ($screenshot = $this->form->getValue('screenshot')) {
$mail->createAttachment(file_get_contents($screenshot->getTempName()), $screenshot->getType());
}
$mail->send();
// A new form for a new submission
$this->form = new aFeedbackForm();
} catch (Exception $e) {
$this->logMessage('Request email failed: ' . $e->getMessage(), 'err');
$this->failed = true;
return 'Success';
}
$this->getUser()->setFlash('reportSubmittedBy', $feedback['name']);
$this->redirect($feedback['section']);
}
}
}
public function executeSearch(sfWebRequest $request)
{
// create the array of pages matching the query
$q = $request->getParameter('q');
if ($request->hasParameter('x')) {
// We like to use input type="image" for presentation reasons, but it generates
// ugly x and y parameters with click coordinates. Get rid of those and come back.
return $this->redirect(sfContext::getInstance()->getController()->genUrl('a/search', true) . '?' . http_build_query(array("q" => $q)));
}
$key = strtolower(trim($q));
$key = preg_replace('/\\s+/', ' ', $key);
$replacements = sfConfig::get('app_a_search_refinements', array());
if (isset($replacements[$key])) {
$q = $replacements[$key];
}
$values = aZendSearch::searchLuceneWithValues(Doctrine::getTable('aPage'), $q, aTools::getUserCulture());
$nvalues = array();
foreach ($values as $value) {
// doesn't implement isset
if (strlen($value->info)) {
$info = unserialize($value->info);
if (!aPageTable::checkPrivilege('view', $info)) {
continue;
}
}
$nvalue = $value;
if (substr($nvalue->slug, 0, 1) === '@') {
// Virtual page slug is a named Symfony route, it wants search results to go there
$nvalue->url = $this->getController()->genUrl($nvalue->slug, true);
} else {
$slash = strpos($nvalue->slug, '/');
if ($slash === false) {
// A virtual page (such as global) taht isn't the least bit interested in
// being part of search results
continue;
}
if ($slash > 0) {
// A virtual page slug which is a valid Symfony route, such as foo/bar?id=55
$nvalue->url = $this->getController()->genUrl($nvalue->slug, true);
} else {
// A normal CMS page
$nvalue->url = aTools::urlForPage($nvalue->slug);
}
}
$nvalue->class = 'aPage';
$nvalues[] = $nvalue;
}
$values = $nvalues;
if ($this->searchAddResults($values, $q)) {
usort($values, "aActions::compareScores");
}
$this->pager = new aArrayPager(null, sfConfig::get('app_a_search_results_per_page', 10));
$this->pager->setResultArray($values);
$this->pager->setPage($request->getParameter('page', 1));
$this->pager->init();
$this->pagerUrl = "a/search?" . http_build_query(array("q" => $q));
// setTitle takes care of escaping things
$this->getResponse()->setTitle(aTools::getOptionI18n('title_prefix') . 'Search for ' . $q . aTools::getOptionI18n('title_suffix'));
$this->results = $this->pager->getResults();
}
public function updateLuceneIndex()
{
$title = $this->getTitle();
$summary = $this->getSearchSummary();
$text = $this->getSearchText();
$slug = $this->getSlug();
aZendSearch::updateLuceneIndex($this, array('text' => $text), $this->getCulture(), array('title' => $title, 'summary' => $summary, 'slug' => $slug, 'view_is_secure' => $this->getViewIsSecure()));
}
public static function registerZend()
{
if (self::$zendLoaded) {
return;
}
// Zend 1.8.0 and thereafter
include_once 'Zend/Loader/Autoloader.php';
$loader = Zend_Loader_Autoloader::getInstance();
// NOT the default autoloader, Symfony's is the default.
// Thanks to Guglielmo Celata
// $loader->setFallbackAutoloader(true);
$loader->suppressNotFoundWarnings(false);
// Before Zend 1.8.0
// require_once 'Zend/Loader.php';
// Zend_Loader::registerAutoload();
self::$zendLoaded = true;
}
/**
* DOCUMENT ME
*/
public function updateLuceneIndex()
{
aZendSearch::updateLuceneIndex(array('object' => $this, 'indexed' => array('type' => $this->getType(), 'title' => $this->getTitle(), 'description' => $this->getDescription(), 'credit' => $this->getCredit(), 'categories' => implode(", ", $this->getCategoryNames()), 'tags' => implode(", ", $this->getTags()))));
}
public static function filterForEngine($options)
{
// This method filters the virtual pages, tags and categories associated with a particular engine based on
// specified criteria such as tag, category, publication date, etc.
// Strategy: do Lucene queries and direct SQL queries that will get us all the info about relevant categories,
// tags and virtual pages. Then turn that into a select distinct query for each of those things. The resulting
// information is sufficient to populate the filters sidebar with options that are still relevant given the
// other criteria in effect, and also to fetch the result pages (you'll want to do that with a LIMIT and an IN
// query looking at the first n IDs returned by this method).
// The options array looks like this. Note that all of these are optional and if each is unspecified or empty
// no restriction is made on that particular basis. 'categoryIds' is used to limit to the categories associated
// with the engine page, while 'categorySlug' is used to limit to a category specified by the user as a
// filter. The 'q' option is Lucene search.
// array(
// 'q' => 'gromit',
// 'categoryIds' => array(1, 3, 5),
// 'categorySlug' => 'cheese',
// 'tag' => 'wensleydale',
// 'slugStem' => '@a_event_search_redirect',
// 'year' => 2010, # Optional, if present only 2010 is shown
// 'month' => 12, # Optional, if present only Dec. 2010 is shown
// 'day' => 15, # Optional, if present only Dec. 15th 2010 is shown
// 'byEventDateRange' => true, # For events only, joins with a_blog_item to get the range
// 'byPublishedAt' => true, # For blog posts or pages
// The returned value looks like this:
// array(
// 'categoriesInfo' => array('slug' => 'cheese', 'name' => 'Cheese'),
// 'tagNames' => array('wensleydale'),
// 'pageIds' => array(10, 15, 20, 25)
$alphaSort = isset($options['alphaSort']) && $options['alphaSort'];
if (isset($options['q']) && strlen($options['q'])) {
$q = $options['q'];
$key = strtolower(trim($q));
$key = preg_replace('/\\s+/', ' ', $key);
$replacements = sfConfig::get('app_a_search_refinements', array());
if (isset($replacements[$key])) {
$q = $replacements[$key];
}
if (isset($options['slugStem'])) {
$q = "({$q}) AND slug:" . $options['slugStem'];
}
try {
$values = aZendSearch::searchLuceneWithValues(Doctrine::getTable('aPage'), $q, aTools::getUserCulture());
} catch (Exception $e) {
// Lucene search error. TODO: display it nicely if they are always safe things to display. For now: just don't crash
$values = array();
}
$now = date('YmdHis');
$pageIds = array();
foreach ($values as $value) {
// Regardless of the above if it ain't published yet we can't see it.
// We filter on that in the Doctrine query too but take advantage of
// this chance to preempt a little work
if ($value->published_at > $now) {
continue;
}
// 1.5: the names under which we store columns in Zend Lucene have changed to
// avoid conflict with also indexing them
$info = unserialize($value->info_stored);
if (!aPageTable::checkPrivilege('view', $info)) {
continue;
}
$pageIds[] = $info['id'];
}
}
$mysql = new aMysql();
if (isset($options['slugStem'])) {
$params['slug_pattern'] = $options['slugStem'] . '%';
}
// Select the relevant virtual pages for this engine
$q = 'from a_page p ';
// If alpha sort is present we need title slots
if ($alphaSort) {
if (!isset($options['culture'])) {
$options['culture'] = aTools::getUserCulture();
}
$culture = $options['culture'];
$q .= "\n LEFT JOIN a_area a ON a.page_id = p.id AND a.name = 'title' AND a.culture = :culture\n LEFT JOIN a_area_version v ON v.area_id = a.id AND a.latest_version = v.version \n LEFT JOIN a_area_version_slot avs ON avs.area_version_id = v.id\n LEFT JOIN a_slot s ON s.id = avs.slot_id ";
$params['culture'] = $culture;
}
// Merge in categories. A left join unless we are restricted to certain categories
$hasCategoryIds = isset($options['categoryIds']) && count($options['categoryIds']);
$hasCategorySlug = isset($options['categorySlug']) && strlen($options['categorySlug']);
$restrictedByCategory = $hasCategoryIds || $hasCategorySlug;
if ($restrictedByCategory) {
$cjoin = 'inner join';
} else {
$cjoin = 'left join';
}
$q .= $cjoin . ' a_page_to_category ptc on ptc.page_id = p.id ' . $cjoin . ' a_category c on ptc.category_id = c.id ';
// The engine page is locked down to these categories. If none are specified it is not
// locked down by category
if ($hasCategoryIds) {
$q .= "and c.id in :category_ids ";
$params['category_ids'] = $options['categoryIds'];
}
// Bring in tags...
$hasTag = isset($options['tag']) && strlen($options['tag']);
if ($hasTag) {
$q .= 'inner join ';
} else {
$q .= 'left join ';
}
$q .= 'tagging ti on ti.taggable_id = p.id and ti.taggable_model = "aPage" left join tag t on ti.tag_id = t.id ';
// Get ready to filter posts or events chronologically
$year = sprintf("%04d", isset($options['year']) ? $options['year'] : 0);
$month = sprintf("%02d", isset($options['month']) ? $options['month'] : 0);
$day = sprintf("%02d", isset($options['day']) ? $options['day'] : 0);
$startYear = $year;
$endYear = $year;
if ($year > 0) {
if ($month == 0) {
// Do not mess up the two digit strings please
$startMonth = '01';
$startDay = '01';
$endMonth = '12';
$endDay = '31';
} else {
$startMonth = $month;
$endMonth = $month;
if ($day == 0) {
// Do not mess up the two digit strings please
$startDay = '01';
$endDay = '31';
} else {
$startDay = $day;
$endDay = $day;
}
}
} else {
// For posts "today and forward" is not a relevant concept (and a separate clause
// already makes sure we don't see unpublished stuff). For events we'll override
// the start date below
$startYear = '0000';
$startMonth = '01';
$startDay = '01';
$endYear = '9999';
$endMonth = '12';
$endDay = '31';
}
$events = isset($options['byEventDateRange']) && $options['byEventDateRange'];
if ($events && $startYear === '0000') {
list($startYear, $startMonth, $startDay) = preg_split('/-/', date('Y-m-d'));
}
if ($events) {
// The event's start and end dates are part of the blog item table
$q .= ' inner join a_blog_item bi on bi.page_id = p.id ';
$q .= "and bi.start_date <= :end_date ";
$params['end_date'] = "{$endYear}-{$endMonth}-{$endDay}";
$q .= "and bi.end_date >= :start_date ";
$params['start_date'] = "{$startYear}-{$startMonth}-{$startDay}";
}
// Criteria for the pages themselves
$q .= 'where p.slug like :slug_pattern ';
// We often filter posts (not events) by a range of publication dates
if (isset($options['byPublishedAt']) && $options['byPublishedAt']) {
$q .= "and p.published_at <= :p_end_date ";
$params['p_end_date'] = "{$endYear}-{$endMonth}-{$endDay}";
$q .= "and p.published_at >= :p_start_date ";
$params['p_start_date'] = "{$startYear}-{$startMonth}-{$startDay}";
}
// In no case do we show unpublished material
$q .= 'and p.published_at <= NOW() and (p.archived IS NULL or p.archived IS FALSE) ';
// ... But only those matching the Lucene search that already gave us specific IDs.
// NOTE: if pageIds is not null and is empty, NOTHING should be returned
// (someone searched for something that doesn't appear in the system)
if (isset($pageIds)) {
if (count($pageIds)) {
$q .= 'and p.id in :pageIds ';
$params['pageIds'] = $pageIds;
} else {
$q .= 'and 0 <> 0 ';
}
}
if ($alphaSort) {
$pagesOrderBy = 's.value asc';
} elseif ($events) {
$pagesOrderBy = 'bi.start_date asc, bi.start_time asc';
} else {
// Oops: blog presentation is typically descending, not ascending
$pagesOrderBy = 'p.published_at desc';
}
// Separate queries, but quite fast because we're not bogged down in Doctrineland
$c_q = $q;
$t_q = $q;
$p_q = $q;
// We are filtering by this specific category
if ($hasCategorySlug) {
// Limit tags and pages by this specific category, but don't limit
// categories by it, otherwise we can't present a choice of categories
// meeting the other criteria
$t_q .= "and c.slug = :category_slug ";
$p_q .= "and c.slug = :category_slug ";
$params['category_slug'] = $options['categorySlug'];
}
if ($hasTag) {
// Limit pages and categories by this specific tag, but don't limit
// tags by it, otherwise we can't present a choice of tags
// meeting the other criteria
$p_q .= 'and t.name = :tag_name ';
$c_q .= 'and t.name = :tag_name ';
$params['tag_name'] = $options['tag'];
}
// In the cases where we are looking for categories or tags, be sure to
// discard the null rows from the LEFT JOINs. This is simpler than
// determining when to switch them to INNER JOINs
$result = array('categoriesInfo' => $mysql->query('select distinct c.slug, c.name ' . $c_q . 'and c.slug is not null order by c.name', $params), 'tagsByName' => $mysql->query('select t.name, count(distinct p.id) as t_count ' . $t_q . 'and t.name is not null group by t.name order by t.name', $params), 'tagsByPopularity' => $mysql->query('select t.name, count(distinct p.id) as t_count ' . $t_q . 'and t.name is not null group by t.name order by t_count desc limit 10', $params), 'pageIds' => $mysql->queryScalar('select distinct p.id ' . $p_q . ' order by ' . $pagesOrderBy, $params));
return $result;
}
public function updateLuceneIndex()
{
aZendSearch::updateLuceneIndex($this, array('type' => $this->getType(), 'title' => $this->getTitle(), 'description' => $this->getDescription(), 'credit' => $this->getCredit(), 'tags' => implode(", ", $this->getTags())));
}
/**
* DOCUMENT ME
* @param Doctrine_Query $q
* @param mixed $luceneQuery
* @return mixed
*/
public function addSearchQuery(Doctrine_Query $q = null, $luceneQuery)
{
return aZendSearch::addSearchQuery($this, $q, $luceneQuery);
}
public function executeSearch(sfWebRequest $request)
{
// create the array of pages matching the query
$q = $request->getParameter('q');
if ($request->hasParameter('x')) {
// We like to use input type="image" for presentation reasons, but it generates
// ugly x and y parameters with click coordinates. Get rid of those and come back.
return $this->redirect(sfContext::getInstance()->getController()->genUrl('a/search', true) . '?' . http_build_query(array("q" => $q)));
}
$key = strtolower(trim($q));
$key = preg_replace('/\\s+/', ' ', $key);
$replacements = sfConfig::get('app_a_search_refinements', array());
if (isset($replacements[$key])) {
$q = $replacements[$key];
}
$values = aZendSearch::searchLuceneWithValues(Doctrine::getTable('aPage'), $q, aTools::getUserCulture());
$nvalues = array();
foreach ($values as $value) {
if (!sfContext::getInstance()->getUser()->isAuthenticated()) {
if (isset($value->view_is_secure) && $value->view_is_secure) {
continue;
}
}
$nvalue = $value;
$nvalue->url = aTools::urlForPage($nvalue->slug, true);
$nvalue->class = 'aPage';
$nvalues[] = $nvalue;
}
$values = $nvalues;
if ($this->searchAddResults($values, $q)) {
usort($values, "aActions::compareScores");
}
$this->pager = new aArrayPager(null, sfConfig::get('app_a_search_results_per_page', 10));
$this->pager->setResultArray($values);
$this->pager->setPage($request->getParameter('page', 1));
$this->pager->init();
$this->pagerUrl = "a/search?" . http_build_query(array("q" => $q));
// setTitle takes care of escaping things
$this->getResponse()->setTitle(aTools::getOptionI18n('title_prefix') . 'Search for ' . $q);
$this->results = $this->pager->getResults();
}
public function executeContactList(sfWebRequest $request)
{
$this->checkProcess($request);
$nameQuery = $this->registration->getFullName();
$nameQuery = aZendSearch::luceneRemoveSpecialchars( $nameQuery );
$this->luceneQuery = ($request->getParameter('query')) ? $request->getParameter('query') : $nameQuery;
// $this->luceneQuery = ($this->luceneQuery == '') ? null : $this->luceneQuery;
$this->query = $this->luceneQuery;
//$this->luceneQuery = $this->registration->getFirstname();
$this->pager = $this->contact_getPager();
$this->contacts = $this->contact_buildQuery()->execute() ;
$this->setTemplate('1_contactList');
}
public function createLuceneQuery($query = '')
{
// remove multiple whitespaces and trim !
// $andQuery = // implode ( " AND ", explode( " ", trim( preg_replace('/\s+/', ' ',$query)) ));
$andQuery = aZendSearch::luceneCombineAnd( $query );
$hits = self::getLuceneIndex()->find($andQuery); // $query
$pks = array();
foreach ($hits as $hit)
{
$pks[] = $hit->pk;
}
if (empty($pks))
{
$pks[] = null;
// return array();
}
$q = $this->createQuery('a')
->andWhereIn('a.id', $pks);
// $q = $this->addActiveJobsQuery($q);
return $q;
}
public function updateLuceneIndex()
{
if ($this->getAdmin()) {
// Never index admin pages, that goes against the spirit of
// keeping them completely out of navigation, they are not
// a place for content in the normal sense, they are engines
// for administrative purposes
return;
}
$title = $this->getTitle();
$summary = $this->getSearchSummary();
$text = $this->getSearchText();
$slug = $this->getSlug();
$info = $this->getInfo();
// Already a separate field, so don't store it twice.
// Otherwise though the info structure is well worth it because
// it lets us check explicit privileges
unset($info['title']);
aZendSearch::updateLuceneIndex($this, array('text' => $text), $this->getCulture(), array('title' => $title, 'summary' => $summary, 'slug' => $slug, 'info' => serialize($info)));
}
/**
* Just a hook used by the above
* @param mixed $query
* @param mixed $culture
* @return mixed
*/
public function searchLuceneWithScores($query, $culture)
{
return aZendSearch::searchLuceneWithScores($this, $query, $culture);
}
/**
* DOCUMENT ME
* @param mixed $arguments
* @param mixed $options
*/
protected function execute($arguments = array(), $options = array())
{
// We've come a long way in reducing memory usage here, but it's still an expensive job
ini_set('memory_limit', '256M');
// initialize the database connection
$databaseManager = new sfDatabaseManager($this->configuration);
$connection = $databaseManager->getDatabase($options['connection'] ? $options['connection'] : null)->getConnection();
// Initialize the context, which loading use of helpers, notably url_for
// First set config vars so that reasonable siteless-but-rooted URLs can be generated
// TODO: think about ways to make this work for people who like frontend_dev.php etc., although
// we're doing rather well with an index.php that suits each environment
sfConfig::set('sf_no_script_name', true);
$_SERVER['PHP_SELF'] = '';
$_SERVER['SCRIPT_NAME'] = '';
$context = sfContext::createInstance($this->configuration);
if (isset($options['table'])) {
$indexes = array($options['table']);
} else {
$indexes = sfConfig::get('app_aToolkit_indexes', array());
}
$count = 0;
foreach ($indexes as $index) {
$table = Doctrine::getTable($index);
if ($index === 'aPage') {
aZendSearch::purgeLuceneIndex($table);
// We're about to request updates of all page/culture combinations. Don't
// add that to an existing workload which could result in a huge pileup of
// repeat requests if someone starts interrupting this task and trying again, etc.
$this->query('DELETE FROM a_lucene_update');
$pages = Doctrine::getTable('aPage')->createQuery('p')->innerJoin('p.Areas a')->execute(array(), Doctrine::HYDRATE_ARRAY);
foreach ($pages as $page) {
$cultures = array();
foreach ($page['Areas'] as $area) {
$cultures[$area['culture']] = true;
}
$cultures = array_keys($cultures);
foreach ($cultures as $culture) {
$this->query('INSERT INTO a_lucene_update (page_id, culture) VALUES (:page_id, :culture)', array('page_id' => $page['id'], 'culture' => $culture));
}
}
while (true) {
$result = $this->query('SELECT COUNT(id) AS total FROM a_lucene_update');
$count = $result[0]['total'];
if ($count == 0) {
break;
}
if ($options['verbose']) {
$this->logSection('toolkit', "{$count} pages remain to be indexed, starting another update pass...");
}
$this->update('aPage', $options);
}
} else {
if ($table->hasField('lucene_dirty')) {
aZendSearch::purgeLuceneIndex($table);
$tableSqlName = $table->getTableName();
// Use Doctrine update and count queries to get the performance while
// retaining compatibility with aggregate inheritance "tables" like
// dukeTubesArticle and dukeTubesEvent. With raw SQL we get confused
// because we run out of objects that Doctrine recognizes as being of the
// relevant type but we marked everything in the table as "dirty"
Doctrine_Query::create()->update($index)->set('lucene_dirty', true)->execute();
while (true) {
$count = $table->createQuery('q')->where('q.lucene_dirty IS TRUE')->count();
if ($count == 0) {
break;
}
if ($options['verbose']) {
$this->logSection('toolkit', "{$count} {$index} objects remain to be indexed, starting another update pass...");
}
$this->update($index, $options);
}
} else {
// We don't have a deferred update feature for other tables,
// so we'll have to get them done in the memory available
$table->rebuildLuceneIndex();
}
}
if ($options['verbose']) {
$this->logSection('toolkit', sprintf('Index for "%s" rebuilt', $index));
}
}
}
/**
* DOCUMENT ME
* @return mixed
*/
public static function registerZend()
{
if (self::$zendLoaded) {
return;
}
// Zend 1.8.0 and thereafter
include_once 'Zend/Loader/Autoloader.php';
$loader = Zend_Loader_Autoloader::getInstance();
// NOT the default autoloader, Symfony's is the default.
// Thanks to Guglielmo Celata
// $loader->setFallbackAutoloader(true);
$loader->suppressNotFoundWarnings(false);
// Before Zend 1.8.0
// require_once 'Zend/Loader.php';
// Zend_Loader::registerAutoload();
self::$zendLoaded = true;
// UTF8 tokenizer can be turned off if you don't have now off by default because it is really, really ignorant of English,
// it can't even cope with plural vs singular, much less stemming
// Thanks to Fotis. Also thanks to the Zend Lucene source
// for the second bit. iconv doesn't mean that PCRE was compiled
// with support for Unicode character classes, which the Lucene
// cross-language tokenizer requires to work. Lovely
if (function_exists('iconv') && @preg_match('/\\pL/u', 'a') == 1) {
Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8());
}
}
/**
* DOCUMENT ME
* @param sfWebRequest $request
* @return mixed
*/
public function executeSearch(sfWebRequest $request)
{
$now = date('YmdHis');
// create the array of pages matching the query
$q = $request->getParameter('q');
if ($request->hasParameter('x')) {
// We sometimes like to use input type="image" for presentation reasons, but it generates
// ugly x and y parameters with click coordinates. Get rid of those and come back.
return $this->redirect(sfContext::getInstance()->getController()->genUrl('a/search', true) . '?' . http_build_query(array("q" => $q)));
}
$key = strtolower(trim($q));
$key = preg_replace('/\\s+/', ' ', $key);
$replacements = sfConfig::get('app_a_search_refinements', array());
if (isset($replacements[$key])) {
$q = $replacements[$key];
}
try {
$values = aZendSearch::searchLuceneWithValues(Doctrine::getTable('aPage'), $q, aTools::getUserCulture());
} catch (Exception $e) {
// Lucene search error. TODO: display it nicely if they are always safe things to display. For now: just don't crash
$values = array();
}
// The truth is that Zend cannot do all of our filtering for us, especially
// permissions-based. So we can do some other filtering as well, although it
// would be bad not to have Zend take care of the really big cuts (if 99% are
// not being prefiltered by Zend, and we have a Zend max results of 1000, then
// we are reduced to working with a maximum of 10 real results).
$nvalues = array();
$index = Doctrine::getTable('aPage')->getLuceneIndex();
foreach ($values as $value) {
$document = $index->getDocument($value->id);
// $published_at = $value->published_at;
// New way: don't touch anything but $hit->id directly and you won't force a persistent
// use of memory for the lazy loaded columns http://zendframework.com/issues/browse/ZF-8267
$published_at = $document->getFieldValue('published_at');
if ($published_at > $now) {
continue;
}
// 1.5: the names under which we store columns in Zend Lucene have changed to
// avoid conflict with also indexing them
$info = unserialize($document->getFieldValue('info_stored'));
if (!aPageTable::checkPrivilege('view', $info)) {
continue;
}
$slug = $document->getFieldValue('slug_stored');
if (substr($slug, 0, 1) !== '@' && strpos($slug, '/') === false) {
// A virtual page slug which is not a route is not interested in being part of search results
continue;
}
$nvalues[] = $value;
}
$values = $nvalues;
if ($this->searchAddResults($values, $q)) {
foreach ($values as $value) {
if (get_class($value) === 'stdClass') {
// bc with existing implementations of searchAddResults
if (!isset($value->slug_stored)) {
if (isset($value->slug)) {
$value->slug_stored = $value->slug;
} else {
$value->slug_stored = null;
}
}
if (!isset($value->title_stored)) {
$value->title_stored = $value->title;
}
if (!isset($value->summary_stored)) {
$value->summary_stored = $value->summary;
}
if (!isset($value->engine_stored)) {
if (isset($value->engine)) {
$value->engine_stored = $value->engine;
} else {
$value->engine_stored = null;
}
}
}
}
// $value = new stdClass();
// $value->url = $url;
// $value->title = $title;
// $value->score = $scores[$id];
// $value->summary = $summary;
// $value->class = 'Article';
// $values[] = $value;
usort($values, "aActions::compareScores");
}
$this->pager = new aArrayPager(null, sfConfig::get('app_a_search_results_per_page', 10));
$this->pager->setResultArray($values);
$this->pager->setPage($request->getParameter('page', 1));
$this->pager->init();
$this->pagerUrl = "a/search?" . http_build_query(array("q" => $q));
// setTitle takes care of escaping things
$this->getResponse()->setTitle(aTools::getOptionI18n('title_prefix') . 'Search for ' . $q . aTools::getOptionI18n('title_suffix'));
$results = $this->pager->getResults();
// Now that we have paginated and obtained the short list of results we really
// care about it's OK to use the lazy load features of Lucene for the last mile
$nresults = array();
foreach ($results as $value) {
$nvalue = $value;
$nvalue->slug = $nvalue->slug_stored;
$nvalue->title = $nvalue->title_stored;
$nvalue->summary = $nvalue->summary_stored;
if (strlen($nvalue->engine_stored)) {
$helperClass = $nvalue->engine_stored . 'SearchHelper';
if (class_exists($helperClass)) {
$searchHelper = new $helperClass();
$nvalue->partial = $searchHelper->getPartial();
}
}
if (!isset($nvalue->url)) {
if (substr($nvalue->slug, 0, 1) === '@') {
// Virtual page slug is a named Symfony route, it wants search results to go there
$nvalue->url = $this->getController()->genUrl($nvalue->slug, true);
} else {
$slash = strpos($nvalue->slug, '/');
if ($slash === false) {
// A virtual page (such as global) that isn't the least bit interested in
// being part of search results
continue;
}
if ($slash > 0) {
// A virtual page slug which is a valid Symfony route, such as foo/bar?id=55
$nvalue->url = $this->getController()->genUrl($nvalue->slug, true);
} else {
// A normal CMS page
$nvalue->url = aTools::urlForPage($nvalue->slug);
}
}
}
$nvalue->class = 'aPage';
$nresults[] = $nvalue;
}
$this->results = $nresults;
}
/**
* DOCUMENT ME
* @return mixed
*/
public function updateLuceneIndex()
{
if ($this->getAdmin()) {
// Never index admin pages, that goes against the spirit of
// keeping them completely out of navigation, they are not
// a place for content in the normal sense, they are engines
// for administrative purposes
return;
}
$title = $this->getTitle();
$engine = $this->getEngine();
$summary = $this->getSearchSummary();
$text = $this->getSearchText();
$tags = implode(',', $this->getTags());
$categories = array();
$categoryObjects = $this->getCategories();
foreach ($categoryObjects as $category) {
$categories[] = $category->getName();
}
$categories = implode(',', $categories);
$metaDescription = $this->getMetaDescription();
$slug = $this->getSlug();
$info = $this->getInfo();
// Already separate fields, so don't store them twice.
unset($info['title'], $info['engine']);
$tags = $this->getTags();
$args = array('object' => $this, 'indexed' => array('text' => $text, 'slug' => $slug, 'title' => $title, 'tags' => implode(', ', $tags), 'categories' => $categories, 'metadescription' => $metaDescription, 'engine' => $engine), 'culture' => $this->getCulture(), 'stored' => array('title_stored' => $title, 'summary_stored' => $summary, 'slug_stored' => $slug, 'engine_stored' => strlen($engine) ? $engine : '', 'info_stored' => serialize($info)), 'boosts' => array('tags' => 2.0, 'metadescription' => 1.2, 'title' => 3.0), 'keywords' => array('published_at' => preg_replace('/[^\\d]/', '', $this->published_at)));
if (strlen($engine)) {
$helperClass = $engine . 'SearchHelper';
if (class_exists($helperClass)) {
$searchHelper = new $helperClass();
$args = $searchHelper->filterUpdateLuceneIndex($args);
}
}
// If a filter returns false, it might be associated with a missing event.
// don't index
if (is_array($args)) {
aZendSearch::updateLuceneIndex($args);
}
}
