/**
* @param ReferenceFactory $factory
* @param Extractor[] $extractors
* @param string $text
* @return Reference[]
* @throws MWException
* @throws \Flow\Exception\WikitextException
*/
protected function extractReferences(ReferenceFactory $factory, array $extractors, $text)
{
$dom = Utils::createDOM($text);
$output = array();
$xpath = new DOMXPath($dom);
foreach ($extractors as $extractor) {
$elements = $xpath->query($extractor->getXPath());
if (!$elements) {
$class = get_class($extractor);
throw new MWException("Malformed xpath from {$class}: " . $extractor->getXPath());
}
foreach ($elements as $element) {
try {
$ref = $extractor->perform($factory, $element);
} catch (InvalidReferenceException $e) {
wfDebugLog('Flow', 'Invalid reference detected, skipping element');
$ref = null;
}
// no reference was generated
if ($ref === null) {
continue;
}
// reference points to a special page
if ($ref->getSrcTitle()->isSpecialPage()) {
continue;
}
$output[] = $ref;
}
}
return $output;
}
/**
* Only extract templates to copy to Flow description.
* Requires Parsoid, to reliably extract templates.
*
* @param string $content
* @return string
*/
protected function extractTemplates($content)
{
$content = Utils::convert('wikitext', 'html', $content, $this->title);
$dom = Utils::createDOM($content);
$xpath = new \DOMXPath($dom);
$templates = $xpath->query('//*[@typeof="mw:Transclusion"]');
$content = '';
foreach ($templates as $template) {
$content .= $dom->saveHTML($template) . "\n";
}
return Utils::convert('html', 'wikitext', $content, $this->title);
}
/**
* Creates a DOM with extra considerations for BC with
* previous parsoid content
*
* @param string $content HTML from parsoid
* @return DOMDocument
*/
public static function createDOM($content)
{
/*
* The body tag is required otherwise <meta> tags at the top are
* magic'd into <head> rather than kept with the content.
*/
if (substr($content, 0, 5) !== '<body') {
// BC: content currently comes from parsoid and is stored
// wrapped in <body> tags, but prior to I0d9659f we were
// storing only the contents and not the body tag itself.
$content = "<body>{$content}</body>";
}
return Utils::createDOM($content);
}
/**
* @dataProvider createDomProvider
*/
public function testCreateDomErrorModes($message, $content)
{
$this->assertInstanceOf('DOMDocument', Utils::createDOM($content), $message);
}
