Scraping Bandcamp for artist info.

getBandcampArtist.php

<?php

function curlURL($url) {

    $return = array();
  
    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $url);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($ch, CURLOPT_VERBOSE, 1);
    curl_setopt($ch, CURLOPT_HEADER, 1);
    $response = curl_exec($ch);
    $header_size = curl_getinfo($ch, CURLINFO_HEADER_SIZE);
    $return['header'] = substr($response, 0, $header_size);
    $return['body'] = substr($response, $header_size);
    curl_close($ch);

    return $return;

}

function getBandcampArtist($artist="") {

  $return = array();

  trim($artist);

  $artist = str_replace(
      array(' ','á','é','í','ó','&','-','/','\''),
      array('','a','e','i','o','and','','',''),
      strtolower($artist)
  );

  $words = ['music','band'];

  $url = 'https://'.$artist.'.bandcamp.com';

  $response = curlURL($url);

  if (!stristr($response['header'],'200 OK')) {
    if (stristr($response['header'], 'Location: '.$url.'/releases')) {
      $url.= '/releases';
      $response = curlURL($url);
    }
  }

  if (stristr($response['header'],'200 OK')) {

    $return['url'] = $url;

    $parser = new DOMDocument();
    if (@$parser -> loadHTML($response['body'])) {
      $xpath = new DOMXPath($parser);
      foreach ($xpath -> query('//*[local-name(.) = "meta"][contains(@property, "og:video:secure_url")][@content]') as $node) {
        $return['player'] = $node -> getAttribute('content');
      }
      foreach ($xpath -> query('//*[local-name(.) = "meta"][contains(@property, "og:image")][@content]') as $node) {
        $return['image'] = $node -> getAttribute('content');
      }
    }
  
  } else {
    foreach ($words as $word) {
      if (!stristr($artist, $word)) {
        $return = getBandcampArtist($artist.$word);
        break;
      } else {
        $artist = str_replace($word, '', $artist);
      }
    }
  }

  return $return;

}