Bootie Top 10 Download Script

getBootie.php

<?php

// usage: `php getBootie.php <url> <album-title>`
// for a url like: http://bootiemashup.com/blog/2012/05/triple-tribute-bootie-mashup-pays-tribute-to-mca-of-beastie-boys-robin-gibb-of-bee-gees-donna-summer.html

// @see http://www.getid3.org/
require_once(dirname(__FILE__) . '/getid3/getid3.php');

$url = $argv[1];
$album_title = $argv[2];

$bd = new BootieAlbum($url, $album_title);
$bd->bootleg();

/**
 * A representation of an Bootie album - usually a monthly top-ten.
 */
class BootieAlbum {
	/**
	 * @var string $url URL to bootleg the album from
	 * @var string $album_title Album title.
	 * - TODO: pull from $url <h2>
	 * @var int $track_count Number of tracks on the album
	 * @var array $tracks Array of BootieTracks
	 */
	private $url = '';
	private $album_title = '';
	private $track_count = 0;
	private $tracks = array();
	
	/**
	 * Create a new album
	 * @param type $url
	 * @param type $album_title 
	 */
	public function __construct($url, $album_title) {
		$this->url = $url;
		$this->album_title = $album_title;
	}

	/**
	 * Find each track in the $url, and create a BootieTrack for each one
	 * Once tracks are downloaded, tag all successful downloads
	 */
	public function bootleg() {
		$dom = new DOMDocument;
		$dom->loadHTMLFile($this->url);

		$track_number = 0;
		// TODO: Why doesn't html_entity_decode() handle these? 
		//       I must be doing it wrong...
		$translations = array(
			'&ndash;' => '-',
			'&rsquo;' => "'",
		);

		foreach ($dom->getElementsByTagName('a') as $link) {
			if (preg_match('/\.mp3$/', $link->getAttribute('href'))) {
				$track_number++;
				$track_url = $link->getAttribute('href');
				$file_info = html_entity_decode(strip_tags($this->getInnerHTML($link)));
				$file_info = str_replace(array_keys($translations), array_values($translations), $file_info);
				// artist - title (contributors)
				preg_match('/^(.*) - (.*) \((.*)\)$/', $file_info, $matches);
				$track = new BootieTrack($track_url, $track_number, $matches[1], $matches[2], $matches[3]);
				$track->download();
				if ($track->getStatus() == 'success') {
					$this->tracks[] = $track;
				}
			}
		}
		
		$this->track_count = $track_number;
		
		foreach ($this->tracks as $track) {
			$track->tag($this->album_title, $this->track_count);
		}
	}

	// @see http://us2.php.net/manual/en/book.dom.php#105815
	private function getInnerHTML($element) {
		$innerHTML = ""; 
		$children = $element->childNodes; 
		foreach ($children as $child) { 
			$tmp_dom = new DOMDocument(); 
			$tmp_dom->appendChild($tmp_dom->importNode($child, true)); 
			$innerHTML.=trim($tmp_dom->saveHTML()); 
		} 
		return $innerHTML; 
	}	
}

class BootieTrack {
	private $url = '';
	private $track_number = 0;
	private $artist = '';
	private $title = '';
	private $contributors = '';
	private $filename = '';
	private $status = '';
	
	public function __construct($url, $track_number, $artist, $title, $contributors) {
		$this->url = $url;
		$this->track_number = $track_number;
		$this->artist = $artist;
		$this->title = $title;
		$this->contributors = $contributors;
		$this->status = 'new';
	}
	
	/**
	 * Download the mp3 for a track and update status
	 */
	public function download() {
		$this->filename = implode(' - ', array(
			$this->track_number,
			$this->artist,
			"$this->title ($this->contributors)",
		));
		$this->filename .= '.mp3';
		print "saving $this->filename...";
		if (file_put_contents('./' . $this->filename, file_get_contents($this->url))) {
			print "success!\n";
			$this->status = 'success';
		}
		else {
			print "failure\n";
			$this->status = 'failure';
		}
	}
	
	public function tag($album_title, $track_count) {
		$getID3 = new getID3;
		$getID3->setOption(array('encoding' => 'UTF_8'));
		getid3_lib::IncludeDependency(GETID3_INCLUDEPATH . 'write.php', __FILE__, true);

		$tagwriter = new getid3_writetags;
		$tagwriter->filename = './' . $this->filename;
		$tagwriter->tagformats = array('id3v1', 'id3v2.3');
		$tagwriter->overwrite_tags = true;
		$tagwriter->tag_encoding = 'UTF-8';
		$tagwriter->remove_other_tags = true;

		$tag = array(
			'title' => array("$this->title ($this->contributors)"),
			'artist' => array($this->artist),
			'album' => array($album_title),
			'year' => array(date('Y')), // NOTE: hard-coded
			'genre' => array('mashup'), // NOTE: hard-coded
			'track' => array("$this->track_number/$track_count"),
			'comment' => array($this->contributors),
		);

		$tagwriter->tag_data = $tag;
		print "updating id3 tag for $this->filename...";
		if ($tagwriter->WriteTags()) {
			print "updated!\n";
		}
		else {
			print "failed.\n";
		}
	}
	
	public function getStatus() {
		return $this->status;
	}
}