public function testGetNextWhitespace()
{
$this->assertSame(false, StringUtil::getNextWhitespace('Saya'));
$this->assertEquals(4, StringUtil::getNextWhitespace('Saya belajar segmentasi kalimat.'));
$this->assertEquals(12, StringUtil::getNextWhitespace('Saya belajar segmentasi kalimat.', 5));
// exclusive current position
$this->assertSame(2, StringUtil::getNextWhitespace(' S aya', 0));
}
private function getToken($text, $position)
{
if ($position >= strlen($text) - 1) {
return '';
}
$nextWs = StringUtil::getNextWhitespace($text, $position);
$prevWs = StringUtil::getPrevWhitespace($text, $position);
$tokenStart = $prevWs === false ? 0 : $prevWs + 1;
$tokenEnd = ($nextWs === false ? strlen($text) : $nextWs) - 1;
$token = substr($text, $tokenStart, $tokenEnd - $tokenStart);
return $token;
}
private function getToken($text, $position)
{
if ($position < strlen($text) - 1 && !StringUtil::isWhitespace(substr($text, $position + 1, 1))) {
$nextWs = StringUtil::getNextWhitespace($text, $position);
$prevWs = StringUtil::getPrevWhitespace($text, $position);
$tokenStart = $prevWs === false ? 0 : $prevWs + 1;
$tokenEnd = $nextWs === false ? strlen($text) : $nextWs;
$token = substr($text, $tokenStart, $tokenEnd - $tokenStart);
// strip trailing .
if (!empty($token) && in_array($token[strlen($token) - 1], $this->eosChars)) {
$token = substr($token, 0, strlen($token) - 1);
}
return $token;
} else {
return '';
}
}