/**
* Apply English normalizing logic to string.
*
* @param array $words
* @return array
*/
public static function stemWord($string)
{
if (extension_loaded('stem')) {
return stem_english($string);
}
return $string;
}
static function _stemWord($word)
{
$oldlocale = \setlocale(LC_ALL, 0);
\setlocale(LC_ALL, 'C');
$word = \stem_russian($word);
$word = \stem_english($word);
\setlocale(LC_ALL, $oldlocale);
return $word;
}
public static function Stem($words)
{
$result = array();
$ex = array();
foreach ($words as $key => $word) {
if (isset(self::$cache[$word])) {
$tmp = self::$cache[$word];
} else {
if (substr($key, 0, 2) == 'ru') {
$tmp = stem_russian_unicode($word);
} else {
$tmp = stem_english($word);
}
self::$cache[$word] = $tmp;
}
if (!isset($ex[$tmp])) {
$result[] = $tmp;
$ex[$tmp] = 1;
}
}
return $result;
}
public function sortOutHlCoords()
{
//Lucene operators
$operators = array("and", "or", "not");
$config = $this->getServiceLocator()->get('config');
$paramInfo = $this->sortOutParams($config);
//collect building blocks
$resLoc = $paramInfo['resLoc'];
$site = $paramInfo['site'];
$collection = $paramInfo['collection'];
$container = $paramInfo['container'];
$reel = $paramInfo['reel'];
$page = $paramInfo['page'];
//the all important query
$hl = $this->params()->fromRoute('hl', '');
//coordinates to pass back
$coords = [];
//pass back empty coordinate set if any of these parameters
//are missing
if ($this->isNullOrEmpty($reel) || $this->isNullOrEmpty($page) || $this->isNullOrEmpty($hl)) {
return array("imgloc" => '', "indloc" => '', "coords" => $coords);
}
//if
//location of files - ODW file layout
$resLoc .= '/' . $site . '/' . $collection . '/' . $container . '/' . $reel . '/odw/' . $page . '/';
$imgLoc = $resLoc . '../../' . $page . '.jpg';
$iaLoc = $resLoc . 'ia/' . $page . '.jpg';
//not all images will have IA derivative
if (file_exists($iaLoc) !== false) {
$imgLoc = $iaLoc;
}
$indLoc = $resLoc . 'index/imgworks';
//need index directory and segments file to be valid lucene layout
if (!file_exists($indLoc . '/segments.gen')) {
return array("imgloc" => $imgLoc, "indloc" => $indLoc, "coords" => $coords);
}
//get coordinates from Lucene index
$searchText = '';
//use Lucene tokens for searching
$queryTokens = Analyzer\Analyzer::getDefault()->tokenize($hl);
foreach ($queryTokens as $token) {
$searchTerm = $token->getTermText();
if (!in_array($searchTerm, $operators)) {
//no snowball analyzer or other stemming option
//in Lucene 2.x, so create stem seperately
$searchText .= stem_english($searchTerm);
//Lucene dropped this limitation after 2.x
//but this version won't wildcard without
//at least 3 characters in term
if (strlen($searchTerm) >= 3) {
$searchText .= "* ";
}
//if strlen
}
//if
}
//foreach
//now do search
$index = Lucene\Lucene::open($indLoc);
$searchResults = $index->find($searchText);
//assemble results
foreach ($searchResults as $searchResult) {
array_push($coords, [$searchResult->x1, $searchResult->y1, $searchResult->x2, $searchResult->y2]);
}
//foreach
//pass back image and index location in addition to results
return array("imgloc" => $imgLoc, "indloc" => $indLoc, "coords" => $coords);
}
