/**
* @desc Process Album article page
*
* @param Article $article
* @return array
*/
public function processArticle(Article $article)
{
$albumData = ['article_id' => $article->getId()];
$albumData = array_merge($albumData, $this->getHeader($article));
$albumData['album_lowercase'] = LyricsUtils::lowercase($albumData['Album']);
$albumData['genres'] = $this->getGenres($article);
if (isset($albumData['Genre']) && !in_array($albumData['Genre'], $albumData['genres'])) {
$albumData['genres'][] = $albumData['Genre'];
}
return array_merge($albumData, $this->getFooter($article));
}
/**
* @desc Process Song article page
*
* @param Article $article
* @return array
*/
public function processArticle(Article $article)
{
$songArticleId = $article->getId();
$songData = ['article_id' => $songArticleId];
$songData = array_merge($songData, $this->getFooter($article));
$songData['lyrics'] = $this->getLyrics($article);
// MOB-1367 - make sure the song name is the same as song's article title
$songTitle = $article->getTitle();
$songName = !is_null($songTitle) ? $this->getSongFromArtistTitle($songTitle->getText()) : null;
if (!is_null($songName)) {
$songData['song'] = $songName;
$songData['song_lowercase'] = LyricsUtils::lowercase($songName);
} else {
wfDebugLog(__METHOD__, sprintf('Scraped song without title (%d) or with invalid name', $songArticleId));
}
return $songData;
}
/**
* @desc Returns lower cased value of field
* @param $fieldName
* @return string
*/
public function getLowerCaseField($fieldName)
{
return LyricsUtils::lowercase($this->getField($fieldName));
}
/**
* @desc Extract album data from section heading
*
* @param string $heading - section heading
* @return array
*/
public function getAlbumData($heading)
{
//==[[Entombed:Serpent Saints The Ten Amendments (2007)|Serpent Saints - The Ten Amendments (2007)]]==
$result = [];
$headingArr = explode('|', trim($heading, '][= '));
$result['title'] = false;
if (count($headingArr) > 1) {
$result['title'] = trim($headingArr[0]);
$result['year'] = '';
$heading = $headingArr[1];
}
if (preg_match('#(.+)\\(([\\d]+)\\)#', $heading, $matches)) {
$result['Album'] = trim($matches[1]);
$result['year'] = trim($matches[2]);
} else {
$result['Album'] = trim($heading);
}
$result['album_name_lc'] = LyricsUtils::lowercase($result['Album']);
return $result;
}