protected function doFormatFinalOutputFor($type, $linker)
{
$text = $this->dataValue->getDataItem()->getString();
$length = mb_strlen($text);
// Make a possibly shortened printout string for displaying the value.
// The result is only escaped to be HTML-safe if this is requested
// explicitly. The result will contain mark-up that must not be escaped
// again.
$abbreviate = $type === self::WIKI_LONG || $type === self::HTML_LONG;
$requestedLength = intval($this->dataValue->getOutputFormat());
// Appease the MW parser to correctly apply formatting on the
// first indent
if ($text !== '' && ($text[0] === '*' || $text[0] === '#' || $text[0] === ':')) {
$text = "\n" . $text . "\n";
}
if ($requestedLength > 0 && $requestedLength < $length) {
// Reduces the length and finish it with a whole word
return Normalizer::reduceLengthTo($text, $requestedLength) . ' …';
}
if ($type === self::HTML_SHORT || $type === self::HTML_LONG) {
$text = smwfXMLContentEncode($text);
}
return $abbreviate && $length > 255 ? $this->getAbbreviatedText($text, $length, $linker) : $text;
}
private function predictLanguage($text)
{
if ($this->languageDetection === array()) {
return null;
}
$languageDetector = $this->sanitizerFactory->newNullLanguageDetector();
if (isset($this->languageDetection['TextCatLanguageDetector'])) {
$languageDetector = $this->sanitizerFactory->newTextCatLanguageDetector();
$languageDetector->setLanguageCandidates($this->languageDetection['TextCatLanguageDetector']);
}
return $languageDetector->detect(Normalizer::reduceLengthTo($text, 200));
}
/**
* @since 0.1
*
* @param integer $length
*/
public function reduceLengthTo($length)
{
$this->string = Normalizer::reduceLengthTo($this->string, $length);
}
public function testToLowercase()
{
$this->assertEquals('abcdef', Normalizer::toLowercase('ABCDEF'));
}