Rewrite github generated HTML to Backdrop filtered HTML

degithub

#!/usr/bin/php
<?php
if ($_SERVER['argc'] == 2) {
  $file = $_SERVER['argv'][1];
}
else {
  $file = 'php://stdin';
}

$fp = fopen($file, 'r');
if (!$fp) {
  die("Cannot read from ${file}.");
}

while (!feof($fp)) {
  $line = fgets($fp, 4096);

  $sane = rtrim($line);
  $sane = strip_tags($sane, '<a><em><strong><cite><blockquote><code><ul><ol><li><h2><h3><h4>');
  $sane = preg_replace('/([^<a]) id="([^"])+"/', '\\1', $sane);
  $sane = preg_replace('/ class="([^"])+"/', '', $sane);
  $sane = preg_replace('/ data-lang="([^"])+"/', '', $sane);
  $sane = strtr($sane, array('&#39;' => "'", '&quot;' => '"', '&gt;' => '>', '&lt;' => '<', '&nbsp;' => ' '));
  $sane = html_entity_decode($sane);

  printf("%s\n", $sane);
}

fclose($fp);

++