public function testLinkHtmlSpecialCharsAreDecodedByExtractLink()
{
$this->aggregatorConfiguration->expects($this->atLeastOnce())->method('getLinkExtractPattern')->willReturn('/^(.*?)$/');
$testString = 'http://example.com/file.bin?x=abc&y=123';
$expectedResult = 'http://example.com/file.bin?x=abc&y=123';
$this->assertSame($expectedResult, $this->subjectUnderTest->extractLink($testString));
}
/**
* Parses given XML and returns array with titles & processed links.
*
* @param string $feed Valid RSS XML.
*
* @return array|false Key contains title, value contains processed link. It can also return bool false on error.
*
* @todo This method sucks...
*/
public function parseFeedXml($feed)
{
$feed = simplexml_load_string($feed);
if ($feed === false) {
$this->logger->error('RSS XML parsing failed in ' . $this->getName(), libxml_get_errors());
return false;
}
if (!isset($feed->channel->item)) {
$this->logger->error($this->getName() . ' failed to extract items from XML');
return false;
}
$nameTag = $this->configuration->getNameTagName();
$linkTag = $this->configuration->getLinkTagName();
$nameExtract = $this->configuration->getNameExtractPattern();
$linkExtract = $this->configuration->getLinkExtractPattern();
$linkTransform = $this->configuration->getLinkTransformPattern();
$result = [];
$counter = 0;
foreach ($feed->channel->item as $item) {
if (!isset($item->{$nameTag})) {
$this->logger->warning('Unable to locate name tag <' . $nameTag . '> in ' . $this->getName() . ' for item#' . $counter . '. Skipping item.');
continue;
}
if (!isset($item->{$linkTag})) {
$this->logger->warning('Unable to locate name tag <' . $linkTag . '> in ' . $this->getName() . ' for item#' . $counter . '. Skipping item.');
continue;
}
$nameMatches = [];
if (preg_match($nameExtract, $item->{$nameTag}, $nameMatches) < 1 || !isset($nameMatches[1])) {
$this->logger->warning('Extracting name from "' . $item->{$nameTag} . '" failed in ' . $this->getName() . ' for item#' . $counter . '. Skipping item.');
continue;
}
if (isset($result[$nameMatches[1]])) {
$this->logger->warning('Duplicate name "' . $item->{$nameTag} . '" in ' . $this->getName() . ' for item#' . $counter . '. Skipping item.');
continue;
}
$linkMatches = [];
if (preg_match($linkExtract, $item->{$linkTag}, $linkMatches) < 1 || !isset($linkMatches[1])) {
$this->logger->warning('Extracting link from "' . $item->{$linkTag} . '" failed in ' . $this->getName() . ' for item#' . $counter . '. Skipping item.');
continue;
}
if ($linkTransform !== null) {
$linkMatches[1] = preg_replace($linkTransform[0], $linkTransform[1], $linkMatches[1]);
}
$result[$nameMatches[1]] = $linkMatches[1];
$counter++;
}
return $result;
}
public function testConfigurationIsConsideredValidAfterSettingUrl()
{
$this->subjectUnderTest->setUrl('http://example.org/feed.rss');
$this->assertTrue($this->subjectUnderTest->isValid());
}
