parse.php

<?php

function parse_cve_list($html) {
  $urls = array();
  if (preg_match_all('~<h2><a href="\/node\/(\d+)">.*?SA~', $html, $matches)) {
    foreach ($matches[1] as $node) {
      $urls[] = 'https://www.drupal.org/node/' . $node;
    }
  }
  return $urls;
}

function parse_cve($html) {
  $u = array();

  // First, parse the meta for the advisory id, project name, and versions
  if (preg_match('~<meta name="description" content="(.*?)"~', $html, $m)) {
    $meta = $m[1];
  } else {
    return false;
  }

  // Advisory ID
  if (preg_match('~Advisory ID:\s?(.*?)\s~', $meta, $matches))
    $u['advisory_id'] = $matches[1];
  else if (preg_match('~(.*?)Project~', $meta, $matches)) // missing tag
    $u['advisory_id'] = $matches[1];

  // Project Name
  if (preg_match('~Projects?: (.*?)Version~', $meta, $matches))
    $u['project_name'] = str_replace(array('(third-party modules)', '(third-party module)'), '', $matches[1]);
  
  // Versions affected
  if (preg_match('~.*Version(s?): (.*?)(Date|Sec)~', $meta, $matches)) {
    //$u['versions'] = $matches[2];
    $versions = explode(',', $matches[2]);

    foreach ($versions as $k => $v)
      $versions[$k] = trim($v);

    if (in_array('7.x', $versions))
      $u['drupal_7'] = 'Yes';
    if (in_array('6.x', $versions))
      $u['drupal_6'] = 'Yes';
    if (in_array('5.x', $versions))
      $u['drupal_5'] = 'Yes';
    if (in_array('4.7.x', $versions))
      $u['drupal_4'] = 'Yes';
  }
  
  if (preg_match('~Date: (.*?) ~', $meta, $matches)) {
    if (strtotime($matches[1]) !== false) {
      $u['advisory_date'] = date("Y-m-d", strtotime($matches[1]));
    } else { // Date is missing or corrupt on a few cves
      if (preg_match('~<time.*>(.*?) at.*?</time>~', $html, $matches)) {
        $u['advisory_date'] = date("Y-m-d", strtotime($matches[1]));
      }
    }
  }
    
  // Vulnerability Hunt -- seems more accurate to check the whole doc vs just the meta
  if (stristr($html, 'xss'))
    $u['s_xss'] = 'Yes';

  if (stristr($html, 'cross site scripting'))
    $u['s_xss'] = 'Yes';

  if (stristr($html, 'csrf'))
    $u['s_csrf'] = 'Yes';

  if (stristr($html, 'cross site request forgery'))
    $u['s_csrf'] = 'Yes';

  if (stristr($html, 'access bypass'))
    $u['s_access_bypass'] = 'Yes';

  if (stristr($html, 'code execution'))
    $u['s_arbitrary_code_execution'] = 'Yes';

  if (stristr($html, 'open redirect'))
    $u['s_open_redirect'] = 'Yes';

  if (stristr($html, 'information disclosure'))
    $u['s_information_disclosure'] = 'Yes';

  if (stristr($html, 'sql injection'))
    $u['s_sql_injection'] = 'Yes';

  if (stristr($html, 'session fixation'))
    $u['s_session_fixation'] = 'Yes';

  if (stristr($html, 'denial of service'))
    $u['s_dos'] = 'Yes';

  if (stristr($html, 'filesystem overwrite')) 
    $u['s_file_access'] = 'Yes'; // generic for various file issues

  if (stristr($html, 'privilege escalation'))
    $u['s_privilege_escalation'] = 'Yes'; // generic for various file issues

  if (stristr($html, 'spam'))
    $u['s_spam'] = 'Yes'; // generic for various file issues

  if (preg_match('/<h2>Reported by<\/h2>.*?<li>(.*?)<\/li>/ms', $html, $matches))
    $u['reported_by'] = trim($matches[1]);

  if (preg_match('/<h2>Fixed by<\/h2>.*?<li>(.*?)<\/li>/ms', $html, $matches))
    $u['fixed_by'] = trim($matches[1]);

  if (preg_match('/<h2>Coordinated by<\/h2>.*?<li>(.*?)<\/li>/ms', $html, $matches))
    $u['coordinated_by'] = trim($matches[1]);

  // Unsupported? TODO: Improve check for this.
  if (stristr($html, 'unsupported'))
    $u['unsupported'] = 'Yes';

  // Project short name 
  if (preg_match('~a href="https?://www.drupal.org/project/(.*?)"~', $html, $m))
    $u['project_short_name'] = $m[1];

  // Account for project differences in the short name
  if (empty($u['project_short_name']) || $u['project_short_name'] === 'drupal') {
    if (preg_match('~a href="https?://drupal.org/project/(.*?)"~', $html, $m)) {
      $u['project_short_name'] = $m[1];
    } else {
      die("Cannot find the short name for {$u['project_name']}\r\n");
    }
  }

  // Project url and git repo
  $u['project_url'] = "https://www.drupal.org/project/{$u['project_short_name']}";
  $u['project_git_url'] = "http://git.drupal.org/project/{$u['project_short_name']}.git";

  // Trim whitespace
  foreach ($u as $k => $v)
    $u[$k] = trim($v);

  return $u;
}