/**
* @see ArticleSearchIndex::rebuildIndex()
*/
function callbackRebuildIndex($hookName, $params)
{
assert($hookName == 'ArticleSearchIndex::rebuildIndex');
$solrWebService = $this->getSolrWebService();
// Unpack the parameters.
list($log, $journal) = $params;
// If we got a journal instance then only re-index
// articles from that journal.
$journalIdOrNull = is_a($journal, 'Journal') ? $journal->getId() : null;
// Clear index (if the journal id is null then
// all journals will be deleted from the index).
if ($log) {
echo 'LucenePlugin: ' . __('search.cli.rebuildIndex.clearingIndex') . ' ... ';
}
$solrWebService->deleteArticlesFromIndex($journalIdOrNull);
if ($log) {
echo __('search.cli.rebuildIndex.done') . "\n";
}
// Re-build index, either of a single journal...
if (is_a($journal, 'Journal')) {
$journals = array($journal);
unset($journal);
// ...or for all journals.
} else {
$journalDao =& DAORegistry::getDAO('JournalDAO');
/* @var $journalDao JournalDAO */
$journalIterator =& $journalDao->getJournals();
$journals = $journalIterator->toArray();
}
// We re-index journal by journal to partition the task a bit
// and provide better progress information to the user.
foreach ($journals as $journal) {
if ($log) {
echo __('search.cli.rebuildIndex.indexing', array('journalName' => $journal->getLocalizedTitle())) . ' ';
}
// Mark all articles in the journal for re-indexing.
$numMarked = $this->_solrWebService->markJournalChanged($journal->getId());
// Pull or push?
if ($this->getSetting(0, 'pullIndexing')) {
// When pull-indexing is configured then we leave it up to the
// Solr server to decide when the updates will actually be done.
if ($log) {
echo '... ' . __('plugins.generic.lucene.rebuildIndex.pullResult', array('numMarked' => $numMarked)) . "\n";
}
} else {
// In case of push indexing we immediately update the index.
$numIndexed = 0;
do {
// We update the index in batches to reduce max memory usage
// and make a few intermediate commits.
$articlesInBatch = $solrWebService->pushChangedArticles(SOLR_INDEXING_MAX_BATCHSIZE, $journal->getId());
if (is_null($articlesInBatch)) {
$error = $solrWebService->getServiceMessage();
if ($log) {
echo ' ' . __('search.cli.rebuildIndex.error') . (empty($error) ? '' : ": {$error}") . "\n";
} else {
$this->_informTechAdmin($error, $journal);
}
return false;
}
if ($log) {
echo '.';
}
$numIndexed += $articlesInBatch;
} while ($articlesInBatch == SOLR_INDEXING_MAX_BATCHSIZE);
if ($log) {
echo ' ' . __('search.cli.rebuildIndex.result', array('numIndexed' => $numIndexed)) . "\n";
}
}
}
return true;
}
/**
* Start the solr server and wait for it to
* become online.
* @param $embeddedServer EmbeddedServer
*/
private function startSolrServer($embeddedServer)
{
$embeddedServer->start();
import('plugins.generic.lucene.classes.SolrWebService');
$solrWebService = new SolrWebService('http://localhost:8983/solr/ojs/search', 'admin', 'please change', 'test-inst');
$try = 0;
while ($solrWebService->getServerStatus() == SOLR_STATUS_OFFLINE && $try <= 10) {
sleep(1);
$try++;
}
self::assertEquals(SOLR_STATUS_ONLINE, $solrWebService->getServerStatus());
}
/**
* Rebuild the index for all journals or a single journal
* @param $log boolean Whether to write the log to standard output.
* @param $journal Journal If given, only re-index this journal.
* @param $buildIndex boolean Whether to rebuild the journal index.
* @param $buildDictionaries boolean Whether to rebuild dictionaries.
* @param $messages string Return parameter for log message output.
* @return boolean True on success, otherwise false.
*/
function _rebuildIndex($log, $journal, $buildIndex, $buildDictionaries, $updateBoostFile, &$messages)
{
// Rebuilding the index can take a long time.
@set_time_limit(0);
$solrWebService = $this->getSolrWebService();
if ($buildIndex) {
// If we got a journal instance then only re-index
// articles from that journal.
$journalIdOrNull = is_a($journal, 'Journal') ? $journal->getId() : null;
// Clear index (if the journal id is null then
// all journals will be deleted from the index).
$this->_indexingMessage($log, 'LucenePlugin: ' . __('search.cli.rebuildIndex.clearingIndex') . ' ... ', $messages);
$solrWebService->deleteArticlesFromIndex($journalIdOrNull);
$this->_indexingMessage($log, __('search.cli.rebuildIndex.done') . PHP_EOL, $messages);
// Re-build index, either of a single journal...
if (is_a($journal, 'Journal')) {
$journals = array($journal);
unset($journal);
// ...or for all journals.
} else {
$journalDao = DAORegistry::getDAO('JournalDAO');
/* @var $journalDao JournalDAO */
$journalIterator = $journalDao->getAll();
$journals = $journalIterator->toArray();
}
// We re-index journal by journal to partition the task a bit
// and provide better progress information to the user.
foreach ($journals as $journal) {
$this->_indexingMessage($log, 'LucenePlugin: ' . __('search.cli.rebuildIndex.indexing', array('journalName' => $journal->getLocalizedName())) . ' ', $messages);
// Mark all articles in the journal for re-indexing.
$numMarked = $this->_solrWebService->markJournalChanged($journal->getId());
// Pull or push?
if ($this->getSetting(0, 'pullIndexing')) {
// When pull-indexing is configured then we leave it up to the
// Solr server to decide when the updates will actually be done.
$this->_indexingMessage($log, '... ' . __('plugins.generic.lucene.rebuildIndex.pullResult', array('numMarked' => $numMarked)) . PHP_EOL, $messages);
} else {
// In case of push indexing we immediately update the index.
$numIndexed = 0;
do {
// We update the index in batches to reduce max memory usage
// and make a few intermediate commits.
$articlesInBatch = $solrWebService->pushChangedArticles(SOLR_INDEXING_MAX_BATCHSIZE, $journal->getId());
if (is_null($articlesInBatch)) {
$error = $solrWebService->getServiceMessage();
$this->_indexingMessage($log, ' ' . __('search.cli.rebuildIndex.error') . (empty($error) ? '' : ": {$error}") . PHP_EOL, $messages);
if (!$log) {
// If logging is switched off then inform the
// tech admin with an email (e.g. in the case of
// an OJS upgrade).
$this->_informTechAdmin($error, $journal);
}
return true;
}
$this->_indexingMessage($log, '.', $messages);
$numIndexed += $articlesInBatch;
} while ($articlesInBatch == SOLR_INDEXING_MAX_BATCHSIZE);
$this->_indexingMessage($log, ' ' . __('search.cli.rebuildIndex.result', array('numIndexed' => $numIndexed)) . PHP_EOL, $messages);
}
}
}
if ($buildDictionaries) {
// Rebuild dictionaries.
$this->_indexingMessage($log, 'LucenePlugin: ' . __('plugins.generic.lucene.rebuildIndex.rebuildDictionaries') . ' ... ', $messages);
$solrWebService->rebuildDictionaries();
if ($updateBoostFile) {
$this->_indexingMessage($log, __('search.cli.rebuildIndex.done') . PHP_EOL, $messages);
}
}
// Remove the field cache file as additional fields may be available after re-indexing. If we don't
// do this it may seem that indexing didn't work as the cache will only be invalidated after 24 hours.
$cacheFile = 'cache/fc-plugins-lucene-fieldCache.php';
if (file_exists($cacheFile)) {
if (is_writable(dirname($cacheFile))) {
unlink($cacheFile);
} else {
$this->_indexingMessage($log, 'LucenePlugin: ' . __('plugins.generic.lucene.rebuildIndex.couldNotDeleteFieldCache') . PHP_EOL, $messages);
}
}
if ($updateBoostFile) {
// Update the boost file.
$this->_indexingMessage($log, 'LucenePlugin: ' . __('plugins.generic.lucene.rebuildIndex.updateBoostFile') . ' ... ', $messages);
$this->_updateBoostFiles();
}
$this->_indexingMessage($log, __('search.cli.rebuildIndex.done') . PHP_EOL, $messages);
return true;
}
/**
* Check that the given article is not indexed.
* @param $articleId integer
*/
private function articleNotInIndex($articleId)
{
$article = $this->solrWebService->getArticleFromIndex($articleId);
self::assertTrue(empty($article));
}
