private function scanCharPositions($text)
{
$charPositions = array();
$skippedCharPositions = array();
for ($i = 0; $i < strlen($text); $i++) {
if (!StringUtil::isWhitespace($text[$i])) {
if (empty($charPositions)) {
$charPositions[] = $i;
} elseif (in_array($i - 1, $charPositions) || in_array($i - 1, $skippedCharPositions)) {
$skippedCharPositions[] = $i;
continue;
} else {
$charPositions[] = $i;
}
}
}
return $charPositions;
}
public function testIsWhitespace()
{
$this->assertEquals(true, StringUtil::isWhitespace(' '));
$this->assertEquals(true, StringUtil::isWhitespace("\t"));
$this->assertEquals(true, StringUtil::isWhitespace("\n"));
}
public static function findToken($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], self::$standardEosChars)) {
$token = substr($token, 0, strlen($token) - 1);
$span = new Span($tokenStart, $tokenEnd - 1);
} else {
$span = new Span($tokenStart, $tokenEnd);
}
return $span;
}
}
