public function testEncoding()
{
if (PHP_OS == 'AIX') {
$this->markTestSkipped('Test not available on AIX');
}
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8 */
$analyzer = new Common\Utf8();
// UTF-8 text with a cyrillic symbols
$tokenList = $analyzer->tokenize(iconv('UTF-8', 'Windows-1251', 'Слово1 Слово2 ДругоеСлово'), 'Windows-1251');
$this->assertEquals(count($tokenList), 3);
$this->assertEquals($tokenList[0]->getTermText(), 'Слово');
$this->assertEquals($tokenList[0]->getStartOffset(), 0);
$this->assertEquals($tokenList[0]->getEndOffset(), 5);
$this->assertEquals($tokenList[0]->getPositionIncrement(), 1);
$this->assertEquals($tokenList[1]->getTermText(), 'Слово');
$this->assertEquals($tokenList[1]->getStartOffset(), 7);
$this->assertEquals($tokenList[1]->getEndOffset(), 12);
$this->assertEquals($tokenList[1]->getPositionIncrement(), 1);
$this->assertEquals($tokenList[2]->getTermText(), 'ДругоеСлово');
$this->assertEquals($tokenList[2]->getStartOffset(), 14);
$this->assertEquals($tokenList[2]->getEndOffset(), 25);
$this->assertEquals($tokenList[2]->getPositionIncrement(), 1);
}
public function __construct()
{
parent::__construct();
$this->addFilter(new TokenFilter\LowerCaseUtf8());
}
