function testHasArticleIdsEmpty()
{
$is = new ImageServing(null, 200, 100);
$articles = array(1234);
$this->assertFalse($is->hasArticleIds($articles));
}
/**
* This is called by ImageServing if no representative image is found for a page.
* If this is a song page, will first check the album page, then the artist page for a good image.
* If this is an album page, will check the artist page for a good image.
* If this is an artist page, there is no good fallback, so 'out' will be returned unmodified.
*/
function lw_ImageServingFallback(ImageServing $imageServing, $n, &$out)
{
wfProfileIn(__METHOD__);
$articleTitlesToTry = array();
// For main namespace pages, fall back from song to album to artist.
if (isset($imageServing->articlesByNS[NS_MAIN])) {
$articles = $imageServing->articlesByNS[NS_MAIN];
foreach ($articles as $articleData) {
$title = $articleData['title'];
// If the title is an album, fall back to the artist... if the title is a song, fall back to the album.
$matches = array();
if (0 < preg_match("/^(.*?):.*[ _]\\([0-9]{4}\\)\$/", $title, $matches)) {
$articleTitlesToTry[] = $matches[1];
} else {
// TODO: Fall back to the album if there is one in the wikiText (which would automatically fall back to the artist if the album doesn't have a pic).
// NOTE: At the moment, I actually think artist pics might be better than album pics.
// TODO: Fall back to the album if there is one in the wikiText (which would automatically fall back to the artist if the album doesn't have a pic).
// Fall back to the artist if no album could be found in the wikiText.
// TODO: if we have the wikitext already & the artist can be found in there, use that instead of a regex on the title (it is more likely to be accurate since the artist name may have colons in it).
// TODO: if we have the wikitext already & the artist can be found in there, use that instead of a regex on the title (it is more likely to be accurate since the artist name may have colons in it).
if (0 < preg_match("/^(.*?):/", $title, $matches)) {
$articleTitlesToTry[] = $matches[1];
}
}
}
}
// If we found some reasonable fallback-pages above, pass those into ImageServing.
if (count($articleTitlesToTry) > 0) {
$articleIds = array();
foreach ($articleTitlesToTry as $titleStr) {
$title = Title::newFromText($titleStr);
// Titles with & break image serving, causing an endless loop here -- owen hack fix
if (strpos($titleStr, "&") == false && is_object($title)) {
$articleId = $title->getArticleID();
$article = Article::newFromID($articleId);
if (is_object($article)) {
/* @var $article WikiPage */
// Automatically follow redirects.
if ($article->isRedirect()) {
$title = $article->followRedirect();
$articleId = $title->getArticleID();
}
// The title str could be converted into an article-id, so add it to the array to be processed.
$articleIds[] = $articleId;
}
}
}
// Some of the titles were real pages, use ImageServing to get the results and store them (by reference) in '$out'.
if (count($articleIds) > 0) {
// the ImageServing::fallbackOnNoResults hook is triggered in ImageServing::getImages.
// we don't want to cycle through here repeatedly on the same article ids
if (!$imageServing->hasArticleIds($articleIds)) {
$imageServing = new ImageServing($articleIds);
$out = $imageServing->getImages($n);
}
}
}
wfProfileOut(__METHOD__);
return true;
// because this is a hook
}
