/**
* Creates a search indexer with the given document type
*
* @param string $document_type the shortname of the document type to
* index by.
* @param MDB2_Driver_Common $db the database connection used by this
* indexer.
* @param boolean $new if true, this is a new search index and all indexed
* words for the given document type are removed. If
* false, we are appending to an existing index.
* Defaults to false.
* @param boolean $append if true, keywords keywords for documents that
* are indexed are appended to the keywords that
* may already exist for the document in the index.
* Defaults to false.
*
* @see NateGoSearch::createDocumentType()
*
* @throws NateGoSearchDocumentTypeException if the document type shortname
* does not exist.
*/
public function __construct($document_type, MDB2_Driver_Common $db, $new = false, $append = false)
{
// cache mb_string overloading status
if (self::$use_mb_string === null) {
self::$use_mb_string = extension_loaded('mbstring') && (ini_get('mbstring.func_overload') & 2) === 2;
}
$type = NateGoSearch::getDocumentType($db, $document_type);
if ($type === null) {
throw new NateGoSearchDocumentTypeException("Document type {$document_type} does not exist and cannot be " . "indexed. Document types must be created before being used.", 0, $document_type);
}
$this->document_type = $type;
$this->db = $db;
$this->new = $new;
$this->append = $append;
}
protected function indexPhotos()
{
$spell_checker = new NateGoSearchPSpellSpellChecker('en_US', '', '', $this->getCustomWordList());
$photo_indexer = new NateGoSearchIndexer('photo', $this->db);
$photo_indexer->setSpellChecker($spell_checker);
$photo_indexer->addTerm(new NateGoSearchTerm('title', 5));
$photo_indexer->addTerm(new NateGoSearchTerm('tags', 2));
$photo_indexer->addTerm(new NateGoSearchTerm('description'));
$photo_indexer->setMaximumWordLength(32);
$photo_indexer->addUnindexedWords(NateGoSearchIndexer::getDefaultUnindexedWords());
$type = NateGoSearch::getDocumentType($this->db, 'photo');
$sql = sprintf('select PinholePhoto.title, PinholePhoto.id,
PinholePhoto.description, PinholePhoto.image_set
from PinholePhoto
inner join NateGoSearchQueue
on PinholePhoto.id = NateGoSearchQueue.document_id
and NateGoSearchQueue.document_type = %s
order by PinholePhoto.id', $this->db->quote($type, 'integer'));
$this->debug(Pinhole::_('Indexing photos ... ') . ' ');
$photos = SwatDB::query($this->db, $sql, SwatDBClassMap::get('PinholePhotoWrapper'));
$total = count($photos);
$count = 0;
$current_photo_id = null;
foreach ($photos as $photo) {
$ds = new SwatDetailsStore($photo);
$ds->title = $photo->getTitle();
$tags = '';
foreach ($photo->tags as $tag) {
$tags .= ' ' . $tag->title . ' ' . $tag->name;
}
$ds->tags = $tags;
if ($count % 10 == 0) {
$photo_indexer->commit();
$this->debug(str_repeat(chr(8), 3));
$this->debug(sprintf('%2d%%', $count / $total * 100));
}
$document = new NateGoSearchDocument($ds, 'id');
$photo_indexer->index($document);
$current_photo_id = $photo->id;
$count++;
$sql = sprintf('delete from NateGoSearchQueue where
document_type = %s and document_id = %s', $this->db->quote($type, 'integer'), $this->db->quote($photo->id, 'integer'));
SwatDB::exec($this->db, $sql);
}
if (count($photos) > 0 && isset($this->memcache)) {
$this->memcache->flushNs('photos');
}
$this->debug(str_repeat(chr(8), 3) . Pinhole::_('done') . "\n");
$photo_indexer->commit();
unset($photo_indexer);
}
protected function indexComments()
{
$type_shortname = 'comment';
$spell_checker = new NateGoSearchPSpellSpellChecker('en_US', '', '', $this->getCustomWordList());
$comment_indexer = new NateGoSearchIndexer($type_shortname, $this->db);
$comment_indexer->setSpellChecker($spell_checker);
$comment_indexer->addTerm(new NateGoSearchTerm('fullname', 30));
$comment_indexer->addTerm(new NateGoSearchTerm('email', 20));
$comment_indexer->addTerm(new NateGoSearchTerm('bodytext', 1));
$comment_indexer->setMaximumWordLength(32);
$comment_indexer->addUnindexedWords(NateGoSearchIndexer::getDefaultUnindexedWords());
$type = NateGoSearch::getDocumentType($this->db, $type_shortname);
$sql = sprintf('select BlorgComment.*
from BlorgComment
inner join NateGoSearchQueue
on BlorgComment.id = NateGoSearchQueue.document_id
and NateGoSearchQueue.document_type = %s
order by BlorgComment.id', $this->db->quote($type, 'integer'));
$this->debug(Blorg::_('Indexing comments... ') . ' ');
$comments = SwatDB::query($this->db, $sql, SwatDBClassMap::get('BlorgCommentWrapper'));
$total = count($comments);
$count = 0;
foreach ($comments as $comment) {
$ds = new SwatDetailsStore($comment);
if ($count % 10 == 0) {
$comment_indexer->commit();
$this->debug(str_repeat(chr(8), 3));
$this->debug(sprintf('%2d%%', $count / $total * 100));
}
$document = new NateGoSearchDocument($ds, 'id');
$comment_indexer->index($document);
$count++;
}
$this->debug(str_repeat(chr(8), 3) . Blorg::_('done') . "\n");
$comment_indexer->commit();
unset($comment_indexer);
$sql = sprintf('delete from NateGoSearchQueue where document_type = %s', $this->db->quote($type, 'integer'));
SwatDB::exec($this->db, $sql);
}
/**
* Queries the NateGoSearch index with a set of keywords
*
* Querying does not directly return a set of results. This is due to the
* way NateGoSearch is designed. The document ids from this search are
* stored in a results table and accessed through a unique identifier.
*
* @param string $keywords the search string to query.
*
* @return NateGoSearchResult an object containing result information.
*
* @see NateGoSearchResult::getUniqueId()
*/
public function query($keywords)
{
static $unique_counter = 0;
$id = sha1(uniqid($unique_counter, true));
$keywords = $this->normalizeKeywordsForSpelling($keywords);
if ($this->spell_checker === null) {
$misspellings = array();
} else {
$misspellings = $this->spell_checker->getMisspellingsInPhrase($keywords);
}
$misspellings = $this->getPopularReplacements($keywords, $misspellings);
$keywords = $this->normalizeKeywordsForSearching($keywords);
$keywords_hash = sha1($keywords);
$results = new NateGoSearchResult($this->db, $id, $keywords, $this->document_types);
$results->addMisspellings($misspellings);
$searched_keywords = array();
$keyword = strtok($keywords, ' ');
while ($keyword) {
if (in_array($keyword, $this->blocked_words)) {
$results->addBlockedWords($keyword);
} else {
$searched_keywords[] = NateGoSearchIndexer::stemKeyword($keyword);
$results->addSearchedWords($keyword);
}
$keyword = strtok(' ');
}
$keywords = implode(' ', $searched_keywords);
if (count($this->document_types) > 0) {
$this->db->loadModule('Function');
$params = array($this->db->quote($keywords, 'text'), $this->db->quote($keywords_hash, 'text'), $this->quoteArray($this->document_types), $this->db->quote($id, 'text'));
$types = array('text');
$rs = $this->db->function->executeStoredProc('nateGoSearch', $params, $types);
if (MDB2::isError($rs)) {
throw new NateGoSearchDBException($rs);
}
$unique_id = $rs->fetchOne();
if (MDB2::isError($unique_id)) {
throw new NateGoSearchDBException($unique_id);
}
$unique_counter++;
$results->setUniqueId($unique_id);
$sql = sprintf('select count(document_id) from %s
where unique_id = %s', $results->getResultTable(), $this->db->quote($unique_id, 'text'));
$document_count = $this->db->queryOne($sql);
if (MDB2::isError($document_count)) {
throw new NateGoSearchDBException($document_count);
}
$results->setDocumentCount($document_count);
}
return $results;
}