Downloads all the wallpapers on Simple Desktops.

simple-desktops-scrape.php

#!/usr/bin/env php

<?php

for ($page = 1; true; $page += 1) {
  # Get the page of desktops
  $url = 'http://simpledesktops.com/browse/' . $page . '/';
  $html = @file_get_contents($url);

  # Bail if we hit a 404
  if ($html === false) {
    break;
  }

  # Go line-by-line
  foreach (explode("\n", $html) as $line) {
    # This is the desktop's title (name)
    if (!preg_match('/title="(.*?)"/', $line, $matches)) {
      continue;
    }
    $title = $matches[1];

    # This is the desktop's 295x184 URL
    if (!preg_match('/src="(.*?)"/', $line, $matches)) {
      continue;
    }
    # Get the desktop full-resolution URL
    $url = preg_replace('/\.png\..+$/', '.png', $matches[1]);

    # If the URL doesn't match YYYY/MM/DD, it's not a desktop
    if (!preg_match('/([0-9]{4})\/([0-9]{2})\/([0-9]{2})/', $url, $matches)) {
      continue;
    }
    list(, $year, $month, $day) = $matches;

    # Make a simple file name
    $name = strtolower($title);
    $name = preg_replace('/[^a-z]/', '-', $name);
    $name = preg_replace('/-+/', '-', $name);
    $name = preg_replace('/^-/', '', $name);
    $name = preg_replace('/-$/', '', $name);
    $name = $year . $month . $day . '-' . $name . '.png';

    # Don't re-download existing assets
    if (file_exists($name)) {
      continue;
    }

    # Download and write the desktop
    $image = file_get_contents($url);
    file_put_contents($name, $image);
    echo $name . "\n";
  }
}

?>