public static function getForLexem($lexem)
{
$lexemSources = LexemSource::get_all_by_lexemId($lexem->id);
return util_objectProperty($lexemSources, 'sourceId');
}
public static function searchFullText($words, $hasDiacritics, $sourceId)
{
$field = $hasDiacritics ? 'formNoAccent' : 'formUtf8General';
$intersection = null;
$stopWords = array();
$lmMap = array();
foreach ($words as $word) {
// Get all LexemModels generating this form
$lms = Model::factory('LexemModel')->table_alias('L')->select('L.id')->distinct()->join('InflectedForm', 'I.lexemModelId = L.id', 'I')->where("I.{$field}", $word)->find_many();
$lmIds = util_objectProperty($lms, 'id');
$lmMap[] = $lmIds;
// Get the FullTextIndex records for each LexemModels. Note that the FTI excludes stop words.
$defIds = FullTextIndex::loadDefinitionIdsForLexemModels($lmIds, $sourceId);
// Determine whether the word is a stop word.
if (empty($defIds)) {
$isStopWord = Model::factory('InflectedForm')->table_alias('I')->join('LexemModel', 'I.lexemModelId = LM.id', 'LM')->join('Lexem', 'LM.lexemId = L.id', 'L')->where("I.{$field}", $word)->where('L.stopWord', 1)->count();
} else {
$isStopWord = false;
}
if ($isStopWord) {
$stopWords[] = $word;
} else {
$intersection = $intersection === null ? $defIds : util_intersectArrays($intersection, $defIds);
}
}
if (empty($intersection)) {
// This can happen when the query is all stopwords or the source selection produces no results
return array(array(), $stopWords);
}
if (count($words) == 1) {
// For single-word queries, skip the ordering part.
// We could sort the definitions by lexicon, but it is very expensive.
return array($intersection, $stopWords);
}
// Now compute a score for every definition
DebugInfo::resetClock();
$positionMap = FullTextIndex::loadPositionsByLexemIdsDefinitionIds($lmMap, $intersection);
$shortestIntervals = array();
foreach ($intersection as $defId) {
$shortestIntervals[] = util_findSnippet($positionMap[$defId]);
}
if ($intersection) {
array_multisort($shortestIntervals, $intersection);
}
DebugInfo::stopClock("Computed score for every definition");
return array($intersection, $stopWords);
}
/**
* Load all lexems having the same form as one of the given lexems, but exclude the given lexems.
**/
function loadSetHomonyms($lexems)
{
if (count($lexems) == 0) {
return array();
}
$names = util_objectProperty($lexems, 'formNoAccent');
$ids = util_objectProperty($lexems, 'id');
return Model::factory('Lexem')->where_in('formNoAccent', $names)->where_not_in('id', $ids)->find_many();
}
