public function testGetPrevNonWhitespace()
{
$this->assertSame(false, StringUtil::getPrevNonWhitespace(' '));
$this->assertSame(26, StringUtil::getPrevNonWhitespace('natural language processing'));
$this->assertEquals(17, StringUtil::getPrevNonWhitespace('segmentasi kalimat'));
$this->assertEquals(11, StringUtil::getPrevNonWhitespace('Saya belajar segmentasi kalimat.', 12));
// exclusive current position
$this->assertEquals(4, StringUtil::getPrevNonWhitespace('bahasa', 5));
}
private function getToken($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 $token;
} else {
return '';
}
}
private function getLeftoverSpan($text, array $positions)
{
if ($positions[count($positions) - 1] != strlen($text) - 1) {
$start = StringUtil::getNextNonWhitespace($text, $positions[count($positions) - 1]);
$end = StringUtil::getPrevNonWhitespace($text);
if ($start !== false && $end - $start > 0) {
return new Span($start, $end + 1);
}
}
}