<?php
// $Id: boost.module,v 1.5.2.12 2011-02-23 18:22:05 mikeytown2 Exp $

/**
 * @file
 * Caches generated output as a static file to be served directly from the
 * webserver.
 */

/**
 * Default cacheablily setting.
 */
define('BOOST_CACHEABILITY_PAGES', '');

/**
 * Default cookie name.
 */
define('BOOST_COOKIE', 'DRUPAL_UID');

/**
 * Default value for sending out debugging info via a message.
 */
define('BOOST_MESSAGE_DEBUG', FALSE);

/**
 * Default value for the root cache dir.
 */
define('BOOST_ROOT_CACHE_DIR', 'cache');

/**
 * Default value for the normal cache dir.
 */
define('BOOST_NORMAL_DIR', 'normal');

/**
 * Default value for the character replacement of ? in the URL.
 */
define('BOOST_CHAR', '_');

/**
 * Default value for ignoring cron cache flush requests.
 */
define('BOOST_IGNORE_FLUSH', TRUE);

/**
 * Default value for ignoring cron cache flush requests.
 */
define('BOOST_EXPIRE_CRON', TRUE);

/**
 * Default etag settings.
 */
define('BOOST_APACHE_ETAG', 3);

/**
 * Default header setttings.
 */
define('BOOST_APACHE_XHEADER', 1);

/**
 * Default setting for forcing all content to be the charset defined below.
 */
define('BOOST_ADD_DEFAULT_CHARSET', TRUE);

/**
 * Default for content charset.
 */
define('BOOST_CHARSET_TYPE', 'utf-8');


/**
 * Implements hook_menu().
 */
function boost_menu() {
  $items = array();
  $path = drupal_get_path('module', 'boost');

  $items['admin/config/system/boost'] = array(
    'title' => 'Boost',
    'description' => 'Configuration for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_NORMAL_ITEM,
    'file' => 'boost.admin.inc',
    'file path' => $path,
  );
  $items['admin/config/system/boost/default'] = array(
    'title' => 'Boost Settings',
    'type' => MENU_DEFAULT_LOCAL_TASK,
  );
  $items['admin/config/system/boost/debug'] = array(
    'title' => 'Debug',
    'description' => 'Debug configuration for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_debug_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'boost.admin.debug.inc',
    'file path' => $path,
  );
  $items['admin/config/system/boost/filesystem'] = array(
    'title' => 'File System',
    'description' => 'File system configuration for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_filesystem_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'boost.admin.filesystem.inc',
    'file path' => $path,
  );
  $items['admin/config/system/boost/htaccess'] = array(
    'title' => '.htaccess',
    'description' => '.htaccess configuration for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_htaccess_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'boost.admin.htaccess.inc',
    'file path' => $path,
  );
  $items['admin/config/system/boost/htaccess/default'] = array(
    'title' => '.htaccess Settings',
    'type' => MENU_DEFAULT_LOCAL_TASK,
  );
  $items['admin/config/system/boost/htaccess/generator'] = array(
    'title' => '.htaccess Generation',
    'description' => '.htaccess generation for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_htaccess_generation'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'boost.admin.htaccess.inc',
    'file path' => $path,
  );
  $items['admin/config/system/boost/expiration'] = array(
    'title' => 'Cache Expiration',
    'description' => 'Cache expiration configuration for Boost.',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('boost_admin_expiration_settings'),
    'access arguments' => array('administer site configuration'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'boost.admin.expiration.inc',
    'file path' => $path,
  );

  return $items;
}

/**
 * Implements hook_init().
 *
 * Performs page setup tasks.
 */
function boost_init() {
  global $_boost;

  $_boost = boost_transform_url();
  // Make sure the page is/should be cached according to our current configuration.
  if (   strpos($_SERVER['SCRIPT_FILENAME'], 'index.php') === FALSE
      || ($_SERVER['REQUEST_METHOD'] != 'GET' && $_SERVER['REQUEST_METHOD'] != 'HEAD')
      || $_SERVER['SERVER_SOFTWARE'] === 'PHP CLI'
      || isset($_GET['nocache'])
      || $_boost['menu_item']['status'] != 200
      || variable_get('site_offline', 0)
  ) {
    $_boost['cache_this'] = FALSE;
  }

  // Give modules a chance to alter the cookie handler callback used.
  // hook_boost_cookie_handler_callback_alter
  $cookie_handler_callback = 'boost_cookie_handler';
  drupal_alter('boost_cookie_handler_callback', $cookie_handler_callback);
  if (function_exists($cookie_handler_callback)) {
    $cookie_handler_callback();
  }
}

/**
 * Implements boost_exit().
 */
function boost_exit($destination = NULL) {
  global $_boost;
  // Bail out of caching.
  if (!isset($_boost['cache_this'])) {
    if (!isset($_boost['is_cacheable'])) {
      return;
    }
    elseif (!$_boost['is_cacheable']) {
      return;
    }
  }
  elseif ($_boost['cache_this'] == FALSE) {
    return;
  }
  elseif (!$_boost['is_cacheable']) {
    return;
  }
  elseif ($_boost['menu_item']['status'] != 200) {
    return;
  }
  elseif (!drupal_page_is_cacheable()) {
    $_boost['is_cacheable'] = FALSE;
    return;
  }

  // Get the data to cache.
  $data = ob_get_contents();

  // Get header info.
  $_boost['header_info'] = boost_get_header_info();
  $_boost['matched_header_info'] = boost_match_header_attributes($_boost['header_info']);
  if ($_boost['matched_header_info']['enabled'] === FALSE) {
    return;
  }

  // Add note to bottom of content if possible.
  if ($_boost['matched_header_info']['commment_start'] && $_boost['matched_header_info']['commment_end']) {
    $expire = $_boost['matched_header_info']['lifetime_max'];
    $cached_at = date('Y-m-d H:i:s', REQUEST_TIME);
    $expires_at = date('Y-m-d H:i:s', REQUEST_TIME + $expire);
    $note = "\n" . $_boost['matched_header_info']['commment_start'] . 'Page cached by Boost @ ' . $cached_at . ', expires @ ' . $expires_at . ', lifetime ' . format_interval($expire) . $_boost['matched_header_info']['commment_end'];
    $data .= $note;
  }

  // Write data to a file.
  if ($_boost['filename']) {
    // Attach extension to filename.
    $_boost['filename'] .= '.' . $_boost['matched_header_info']['extension'];
    // Write to file.
    boost_write_file($_boost['filename'], $data);
  }
}

/**
 * Implements hook_cron(). Performs periodic actions.
 */
function boost_cron() {
  // Remove expired files from the cache.
  global $_boost;
  $_boost['cron'] = TRUE;
  if (   isset($_boost['base_dir'])
      && variable_get('boost_expire_cron', BOOST_EXPIRE_CRON)
        ) {
    $count = _boost_rmdir($_boost['base_dir'], FALSE);
    watchdog('boost', 'Expired %count stale files from static page cache.', array('%count' => $count), WATCHDOG_NOTICE);
  }
}

/**
 * Implements hook_flush_caches(). Deletes all static files.
 */
function boost_flush_caches() {
  // Remove all files from the cache
  global $_boost;
  if (   isset($_boost['base_dir'])
      && (empty($_boost['cron']) || variable_get('boost_ignore_flush', BOOST_IGNORE_FLUSH) == FALSE)
        ) {
    $count = _boost_rmdir($_boost['base_dir'], TRUE);
    watchdog('boost', 'Flushed all files (%count) from static page cache.', array('%count' => $count), WATCHDOG_NOTICE);
  }
  return;
}

/**
 * Implements hook_page_delivery_callback_alter().
 */
function boost_page_delivery_callback_alter(&$callback, $set = FALSE) {
  if ($callback == 'drupal_deliver_html_page') {
    $callback = 'boost_deliver_html_page';
  }
}

/**
 * Given a URL give back eveything we know
 *
 * @param $url
 *   Full URL
 * @param $b_path
 *   Base Path
 */
function boost_transform_url($url = NULL, $b_path = NULL) {
  global $base_root, $base_path;
  $items = &drupal_static(__FUNCTION__);

  // Set defaults if none passed in.
  if ($url === NULL) {
    $url = $base_root . request_uri();
  }
  if ($b_path == NULL) {
    $b_path = $base_path;
  }
  $hash = $url . ' ' . $b_path;

  if (!isset($items[$hash])) {
    $parts = boost_parse_url($url, $b_path);
    if (!$parts) {
      $items[$hash] = array('cache_this' => FALSE);
      return $items[$hash];
    }
    $parts['base_dir'] = boost_get_normal_cache_dir() . '/' . $parts['host'] . $b_path;
    $parts['filename'] = $parts['base_dir'] . $parts['path'] . variable_get('boost_char', BOOST_CHAR) . $parts['query'];
    $parts['directory'] = dirname($parts['filename']);

    // Get the internal path (node/8).
    if (drupal_is_front_page()) {
      $parts['normal_path'] = variable_get('site_frontpage', 'node');
    }
    else {
      $parts['normal_path'] = drupal_get_normal_path($parts['path']);
    }
    // Get the alias (content/about-us).
    $parts['path_alias'] = drupal_get_path_alias($parts['normal_path']);
    // Get all args.
    $args = arg(NULL, $parts['normal_path']);
    // Prevent array warnings.
    $args[0] = empty($args[0]) ? '' : $args[0];
    $args[1] = empty($args[1]) ? '' : $args[1];
    $args[2] = empty($args[2]) ? '' : $args[2];
    $parts['args'] = $args;

    // Get content type.
    $parts = _boost_get_menu_router($parts);
    // See if url is cacheable.
    $parts = boost_is_cacheable($parts);
    $items[$hash] = $parts;
  }

  return $items[$hash];
}

/**
 * Returns the relative normal cache dir. cache/normal.
 */
function boost_get_normal_cache_dir() {
  return variable_get('boost_root_cache_dir', BOOST_ROOT_CACHE_DIR) . '/' . variable_get('boost_normal_dir', BOOST_NORMAL_DIR);
}

/**
 * parse_url that takes into account the base_path
 *
 * @param $url
 *   Full URL
 * @param $b_path
 *   Base Path
 */
function boost_parse_url($url = NULL, $b_path = NULL) {
  global $base_root, $base_path;
  // Set defaults.
  if ($url === NULL) {
    $url = $base_root . request_uri();
  }
  if ($b_path == NULL) {
    $b_path = $base_path;
  }

  // Parse url.
  $parts = parse_url($url);
  if (empty($parts['host']) || empty($parts['path'])) {
    return FALSE;
  }
  if (!isset($parts['query'])) {
    $parts['query'] = '';
  }
  $parts['path'] = preg_replace('/^' . preg_quote($b_path, '/') .'/i', '', $parts['path']);
  $parts['base_path'] = $b_path;
  $parts['query_array'] = array();
  parse_str($parts['query'], $parts['query_array']);

  // Get page number and info from the query string.
  if (!empty($parts['query_array'])) {
    $query = array();
    foreach ($parts['query_array'] as $key => $val) {
      if ($key != 'q' && $key != 'destination' && $key != 'page' && !empty($val)) {
        $query[$key] = $val;
      }
      if ($key == 'page' && is_numeric($val)) {
        $parts['page_number'] = $val;
      }
    }
    ksort($query);
    $parts['query_extra'] = str_replace('&amp;', '&', urldecode(http_build_query($query)));
  }

  // Get fully decoded URL.
  $decoded1 = urldecode($parts['base_path'] . $parts['path'] . variable_get('boost_char', BOOST_CHAR) . $parts['query']);
  $decoded2 = urldecode($decoded1);
  while ($decoded1 != $decoded2) {
    $decoded1 = urldecode($decoded2);
    $decoded2 = urldecode($decoded1);
  }
  $decoded = $decoded2;
  unset($decoded2);
  unset($decoded1);

  $parts['url_full'] = $parts['host'] . $parts['base_path'] . $parts['path'] . variable_get('boost_char', BOOST_CHAR) . $parts['query'];
  $parts['url'] = $url;
  $parts['url_decoded'] = $decoded;
  return $parts;
}

/**
 * Determines whether a given url can be cached or not by boost.
 *
 * TODO: Add in support for the menu_item
 *
 * @param $parts
 *   $parts
 *
 * @return $parts
 */
function boost_is_cacheable($parts) {
  // Set local variables.
  $path = $parts['path'];
  $query = $parts['query'];
  $full = $parts['url_full'];
  $normal_path = $parts['normal_path'];
  $alias = $parts['path_alias'];
  $decoded = $parts['url_decoded'];

  // Never cache
  //  the user autocomplete/login/registration/password/reset/logout pages
  //  any admin pages
  //  comment reply pages
  //  node add page
  //  openid login page
  //  URL variables that contain / or \
  //  if incoming URL contains '..' or null bytes
  //  if decoded URL contains :// outside of the host portion of the url
  //  Limit the maximum directory nesting depth of the path
  //  Do not cache if destination is set.
  if (   $normal_path == 'user'
      || preg_match('!^user/(autocomplete|login|register|password|reset|logout)!', $normal_path)
      || preg_match('!^admin!', $normal_path)
      || preg_match('!^comment/reply!', $normal_path)
      || preg_match('!^node/add!', $normal_path)
      || preg_match('!^openid/authenticate!', $normal_path)
      || strpos($query, '/') !== FALSE
      || strpos($query, "\\") !== FALSE
      || strpos($full, '..') !== FALSE
      || strpos($full, "\0") !== FALSE
      || count(explode('/', $path)) > 10
      || strpos($decoded, "://") !== FALSE
      || !empty($query_array['destination'])
    ) {
    $parts['is_cacheable'] = FALSE;
    return $parts;
  }

  // Check for reserved characters if on windows.
  // http://en.wikipedia.org/wiki/Filename#Reserved_characters_and_words
  // " * : < > |
  $chars = '"*:<>|';
  if (stristr(PHP_OS, 'WIN') && preg_match("/[" . $chars . "]/", $full)) {
    $parts['is_cacheable'] = FALSE;
    return $parts;
  }

  // Match the user's cacheability settings against the path.
  // See http://api.drupal.org/api/function/block_block_list_alter/7
  $visibility = variable_get('boost_cacheability_option', BLOCK_VISIBILITY_NOTLISTED);
  $pages_setting = variable_get('boost_cacheability_pages', BOOST_CACHEABILITY_PAGES);
  if ($pages_setting) {
    // Convert path string to lowercase. This allows comparison of the same path
    // with different case. Ex: /Page, /page, /PAGE.
    $pages = drupal_strtolower($pages_setting);
    if ($visibility < BLOCK_VISIBILITY_PHP) {
      // Convert the alias to lowercase.
      $path = drupal_strtolower($alias);
      // Compare the lowercase internal and lowercase path alias (if any).
      $page_match = drupal_match_path($path, $pages);
      if ($path != $normal_path) {
        $page_match = $page_match || drupal_match_path($normal_path, $pages);
      }
      // When 'boost_cacheability_option' has a value of 0 (BLOCK_VISIBILITY_NOTLISTED),
      // Boost will cache all pages except those listed in 'boost_cacheability_pages'.
      // When set to 1 (BLOCK_VISIBILITY_LISTED), Boost will only cache those
      // pages listed in 'boost_cacheability_pages'.
      $page_match = !($visibility xor $page_match);
    }
    elseif (module_exists('php')) {
      $page_match = php_eval($pages_setting);
    }
    else {
      $page_match = FALSE;
    }
  }
  else {
    $page_match = TRUE;
  }
  $parts['is_cacheable'] = $page_match;
  if (!$parts['is_cacheable']) {
    return $parts;
  }

  // Invoke hook_boost_is_cacheable($path).
  $modules = boost_module_implements('boost_is_cacheable', 'boost');
  foreach ($modules as $module) {
    if (($result = module_invoke($module, 'boost_is_cacheable', $parts)) !== NULL) {
      return $result;
    }
  }
}

/**
 * Implements hook_boost_is_cacheable().
 *
 * TODO support for node types, etc.
 *
 * @param $parts
 *   info about this request
 *
 * @return $parts
 */
function boost_boost_is_cacheable($parts) {
  global $user;
  if ($user->uid != 0) {
    $parts['is_cacheable'] = FALSE;
  }
  else {
    $parts['is_cacheable'] = TRUE;
  }
  return $parts;
}

/**
 * Sets a special cookie preventing authenticated users getting served pages
 * from the static page cache.
 *
 * @param $uid
 *   User ID Number
 * @param $expires
 *   Expiration time
 */
function boost_set_cookie($uid, $expires = NULL) {
  if (!$expires) {
    $expires = ini_get('session.cookie_lifetime');
    $expires = (!empty($expires) && is_numeric($expires)) ? REQUEST_TIME + (int)$expires : 0;
    setcookie(BOOST_COOKIE, strval($uid), $expires, ini_get('session.cookie_path'), ini_get('session.cookie_domain'), ini_get('session.cookie_secure') == '1');
  }
  else {
    setcookie(BOOST_COOKIE, '0', $expires, ini_get('session.cookie_path'), ini_get('session.cookie_domain'), ini_get('session.cookie_secure') == '1');
  }
}

/**
 * Logic for the setting and removal of the boost cookie.
 */
function boost_cookie_handler() {
  global $user;

  // Check if Drupal is started from index.php - could cause problems with other
  // contrib modules like ad module.
  $uid = isset($user->uid) ? $user->uid : 0;
  if (strpos($_SERVER['SCRIPT_FILENAME'], 'index.php') !== FALSE) {
    // Remove Boost cookie at logout if it still exists.
    if (isset($_COOKIE[BOOST_COOKIE]) && $uid == 0) {
      boost_set_cookie($uid, REQUEST_TIME - 86400);
    }
    // Remove Boost cookie if set to -1.
    elseif (isset($_COOKIE[BOOST_COOKIE]) && $_COOKIE[BOOST_COOKIE] == '-1') {
      boost_set_cookie($uid, REQUEST_TIME - 86400);
    }
    // Set Boost cookie if it doesn't exists and user is logged in.
    elseif (!isset($_COOKIE[BOOST_COOKIE]) && $uid != 0) {
      boost_set_cookie($uid);
    }
  }
}

/**
 * Gets menu router contex.
 *
 * Allows for any content type to have it's own cache expiration among
 * other things.
 *
 * @param $parts
 *
 * @return $parts
 *
 */
function _boost_get_menu_router($parts) {
  // Declare array keys.
  $router_item = array();
  $router_item['page_type'] = '';
  $router_item['page_id'] = '';
  // Load the menu item.
  $item = menu_get_item($parts['normal_path']);
  if (is_array($item)) {
    $router_item += $item;
    if ($router_item['access']) {
      $router_item['status'] = 200;
    }
    else {
      $router_item['status'] = 403;
    }
  }
  else {
    $router_item['status'] = 404;
  }

  // Get any extra arguments.
  if (!empty($router_item['path'])) {
    $menu_args = arg(NULL, $router_item['path']);
    $diff = array();
    foreach ($parts['args'] as $key => $value) {
      if (!empty($value)) {
        if (isset($menu_args[$key])) {
          if ($value !== $menu_args[$key] && $menu_args[$key] !== '%') {
            $diff[] = $value;
          }
        }
        else {
          $diff[] = $value;
        }
      }
    }
  }
  if (!empty($diff)) {
    $router_item['extra_arguments'] = implode('/', $diff);
  }
  else {
    $router_item['extra_arguments'] = '';
  }

  // Make sure function for menu callback is loaded.
  // See menu_execute_active_handler()
  if (!empty($router_item['include_file'])) {
    require_once DRUPAL_ROOT . '/' . $router_item['include_file'];
  }

  $parts['menu_item'] = $router_item;
  // Invoke hook_boost_menu_router($router_item).
  $modules = boost_module_implements('boost_menu_router', 'boost');
  foreach ($modules as $module) {
    if (($result = module_invoke($module, 'boost_menu_router', $parts)) !== NULL) {
      break;
    }
  }

  // Remove extra data from the load function
  unset($result['menu_item']['map']);
  unset($result['menu_item']['page_arguments']);

  return $result;
}

/**
 * Implements hook_boost_menu_router().
 *
 * TODO Better support for arguments.
 *
 * @param $parts
 *   info about this request
 *
 * @return $parts
 */
function boost_boost_menu_router($parts) {
  // Handle nodes.
  if ($parts['args'][0] == 'node' && is_numeric($parts['args'][1])) {
    $node = node_load($parts['args'][1]);
    $parts['menu_item']['page_callback'] = 'node';
    $parts['menu_item']['page_id'] = $parts['args'][1];
    if ($node) {
      $parts['menu_item']['page_type'] = $node->type;
    }
    return $parts;
  }

  // Handle taxonomy.
  if ($parts['args'][0] == 'taxonomy' && is_numeric($parts['args'][2])) {
    $term = taxonomy_term_load($parts['args'][2]);
    $parts['menu_item']['page_callback'] = 'taxonomy';
    $parts['menu_item']['page_id'] = $parts['args'][2];
    if ($term) {
      $vocab = taxonomy_vocabulary_load($term->vid);
      $parts['menu_item']['page_type'] = $vocab->name;
    }
    return $parts;
  }

  // Handle users.
  if ($parts['args'][0] == 'user' && is_numeric($parts['args'][1])) {
    $user = user_load($parts['args'][1]);
    $parts['menu_item']['page_callback'] = 'user';
    $parts['menu_item']['page_id'] = $parts['args'][1];
    if ($user !== FALSE) {
      $parts['menu_item']['page_type'] = implode(', ', $user->roles);
    }
    return $parts;
  }

  // Handle views.
  if (isset($parts['menu_item']['page_callback']) && $parts['menu_item']['page_callback'] == 'views_page') {
    $parts['menu_item']['page_callback'] = 'view';
    $parts['menu_item']['page_type'] = array_shift($parts['menu_item']['page_arguments']);
    $parts['menu_item']['page_id'] = array_shift($parts['menu_item']['page_arguments']);
    // See http://drupal.org/node/651798 for the reason why this if is needed
    if (is_array($parts['menu_item']['page_id'])) {
      $parts['menu_item']['page_id'] = array_shift($parts['menu_item']['page_id']);
    }
    return $parts;
  }

  // Handle panels.
  if (isset($parts['menu_item']['page_callback']) && $parts['menu_item']['page_callback'] == 'page_manager_page_execute') {
    $subtask_id = array_shift($parts['menu_item']['page_arguments']);
    $page = page_manager_page_load($subtask_id);
    $task = page_manager_get_task($page->task);
    if ($function = ctools_plugin_get_function($task, 'page callback')) {
      $parts['menu_item']['page_callback'] = $function;
    }
    $parts['menu_item']['page_type'] = $page->task;
    $parts['menu_item']['page_id'] = $page->name;
    return $parts;
  }

  // Try to handle everything else.
  elseif (isset($parts['menu_item']['page_arguments']) && is_array($parts['menu_item']['page_arguments'])) {
    foreach ($parts['menu_item']['page_arguments'] as $string) {
      if (is_string($string) && empty($parts['menu_item']['page_type'])) {
        $parts['menu_item']['page_type'] = $string;
      }
      elseif (is_string($string)) {
        $parts['menu_item']['page_id'] .= $string;
      }
    }
  }

  // If router doesn't hold the the arguments, get them from the URL.
  if (empty($parts['menu_item']['page_type'])) {
    $parts['menu_item']['page_type'] = $parts['menu_item']['extra_arguments'];
  }
  elseif (empty($parts['menu_item']['page_id'])) {
    $parts['menu_item']['page_id'] = $parts['menu_item']['extra_arguments'];
  }

  // Try populating with the query string.
  if (empty($parts['menu_item']['page_type']) && !empty($parts['query_extra'])) {
    $parts['menu_item']['page_type'] = $parts['query_extra'];
  }
  elseif (empty($parts['menu_item']['page_id']) && !empty($parts['query_extra'])) {
    $parts['menu_item']['page_id'] = $parts['query_extra'];
  }

  return $parts;
}

/**
 * Alters module_implements to set a hook to fire at the end.
 *
 * @param $hook
 *   Name of hook.
 * @param $name
 *   Module name to shift to the end of the array.
 *
 * @return array
 */
function boost_module_implements($hook, $name) {
  $modules = module_implements($hook);
  // Make $names built in hook the last one.
  $pos = array_search($name, $modules);
  if ($pos !== FALSE) {
    $temp = $modules[$pos];
    unset($modules[$pos]);
    $modules[] = $temp;
  }
  return $modules;
}

/**
 * Gets and parses the header info.
 *
 * @see drupal_send_headers()
 *
 * @return array
 *   Contains info about the page that is about to be sent.
 */
function boost_get_header_info() {
  $headers = drupal_get_http_header();
  $status = '200 OK';
  $status_number = '200';
  $content_type = 'text/html; charset=utf-8';
  $content_type_basic = 'text/html';
  $charset = 'utf-8';

  foreach ($headers as $name_lower => $value) {
    if ($name_lower == 'status') {
      $status = $value;
    }
    elseif ($name_lower == 'content-type') {
      $content_type = $value;
    }
  }

  preg_match('!^(\d+)!', $status, $matches);
  if (isset($matches[1])) {
    $status_number = (int) $matches[1];
  }
  $content_type_info = explode('; charset=', $content_type);
  $content_type_basic = array_shift($content_type_info);
  if (!empty($content_type_info)) {
    $charset = array_shift($content_type_info);
  }

  return array(
    'status' => $status,
    'status-number' => $status_number,
    'content-type' => $content_type,
    'content-type-basic' => $content_type_basic,
    'charset' => $charset,
    'headers_sent' => headers_sent(),
  );
}

/**
 * Given header info, match it to a file extension
 *
 * @param $header_info
 *   array contains header info
 *
 * @return array
 */
function boost_match_header_attributes($header_info) {
  $type = $header_info['content-type-basic'];
  $enabled = variable_get('boost_enabled_' . $type, -1);
  if ($enabled === -1) {
    boost_get_storage_types();
    $enabled = variable_get('boost_enabled_' . $type, -1);
    if ($enabled === -1) {
      variable_set('boost_enabled_' . $type, FALSE);
      $enabled = FALSE;
    }
  }
  if (!$enabled) {
    return array('enabled' => FALSE);
  }

  $keys = array(
    'enabled',
    'extension',
    'lifetime_max',
    'lifetime_min',
    'commment_start',
    'commment_end',
  );
  $info = array();
  foreach ($keys as $key) {
    $info[$key] = variable_get('boost_' . $key . '_' . $type, -1);
    if ($info[$key] === -1) {
      return array('enabled' => FALSE);
    }
  }
  return $info;
}

/**
 * Get the storage types for the boost cache.
 *
 * @return $types
 *   array('title' => array('content-type' => $values));
 */
function boost_get_storage_types() {
  $primary_types = array();
  $secondary_types = array();

  // Force boost to be last in the array
  $modules = boost_module_implements('boost_storage_types', 'boost');

  // Module with the lightest weight runs last. They will then be able to change
  // the defaults and modules that are heavier
  $modules = array_reverse($modules);
  foreach ($modules as $module) {
    // Invoke hook_boost_storage_types()
    if (($result = module_invoke($module, 'boost_storage_types')) !== NULL) {
      foreach ($result as $key => $values) {
        if (!empty($values['variants'])) {
          foreach ($values['variants'] as $type) {
            $secondary_types[$type] = $values;
            unset($secondary_types[$type]['variants']);
          }
        }
        unset($values['variants']);
        $primary_types[$key] = $values;
      }
    }
  }

  // Merge it all back together
  $merged = array_merge($primary_types, $secondary_types);
  $types = array();
  foreach ($merged as $type => $values) {
    if ($values['title']) {
      $title = $values['title'];
      unset($values['title']);
      $types[$title][$type] = $values;
    }
  }

  // Apply what's saved in the database
  $keys = array(
    'enabled',
    'extension',
    'gzip',
    'lifetime_max',
    'lifetime_min',
    'commment_start',
    'commment_end',

  );
  foreach ($types as $title => $content_types) {
    foreach ($content_types as $type => $values) {
      foreach ($keys as $key) {
        $types[$title][$type][$key] = variable_get('boost_' . $key . '_' . $type, $values[$key]);
        // Set variable in DB if it has never been set before
        if (variable_get('boost_' . $key . '_' . $type, -1) === -1) {
          variable_set('boost_' . $key . '_' . $type, $values[$key]);
        }
      }
    }
  }

  return $types;
}

/**
 * Implements hook_boost_storage_types().
 *
 * @return $types
 */
function boost_boost_storage_types() {
  $types = array();

  $types['text/html'] = array(
    'title'           => t('HTML'),
    'description'     => t('HTML output, usually a webpage'),
    'extension'       => 'html',
    'enabled'         => TRUE,
    'gzip'            => TRUE,
    'lifetime_max'    => 3600,
    'lifetime_min'    => 0,
    'commment_start'  => '<!-- ',
    'commment_end'    => ' -->',
    'variants'        => array(),
  );
  $types['text/xml'] = array(
    'title'           => t('XML'),
    'description'     => t('XML output, usually a feed'),
    'extension'       => 'xml',
    'enabled'         => FALSE,
    'gzip'            => TRUE,
    'lifetime_max'    => 3600,
    'lifetime_min'    => 0,
    'commment_start'  => '<!-- ',
    'commment_end'    => ' -->',
    'variants'        => array(
      'application/xml',
      'application/rss',
      'application/rss+xml',
    ),
  );
  $types['text/javascript'] = array(
    'title'           => t('AJAX/JSON'),
    'description'     => t('JSON output, usually a response to a AJAX request'),
    'extension'       => 'json',
    'enabled'         => FALSE,
    'gzip'            => TRUE,
    'lifetime_max'    => 3600,
    'lifetime_min'    => 0,
    'commment_start'  => '/* ',
    'commment_end'    => ' */',
    'variants'        => array(),
  );
  return $types;
}

/**
 * Recursive version of rmdir(); use with extreme caution.
 *
 * Function also checks file age and only removes expired files.
 *
 * @param $dir
 *   The top-level directory that will be recursively removed.
 * @param $flush
 *   Instead of removing expired cached files, remove all files.
 */
function _boost_rmdir($dir, $flush = TRUE) {
  static $lifetimes = array();
  static $counter = 0;
  if (is_dir($dir) == FALSE) {
    return FALSE;
  }

  if (!boost_in_cache_dir($dir)) {
    return FALSE;
  }

  // Map extensions to cache lifetimes
  if (empty($lifetimes)) {
    $types = boost_get_storage_types();
    foreach ($types as $title => $content_types) {
      foreach ($content_types as $type => $values) {
        $lifetimes[$values['extension']] = $values['lifetime_max'];
      }
    }

    // Be sure to recreate the htaccess file just in case.
    boost_form_submit_handler();
  }

  $objects = scandir($dir);
  $empty_dir = TRUE;
  foreach ($objects as $object) {
    if ($object == "." || $object == "..") {
      continue;
    }
    if ($object == ".htaccess") {
      $empty_dir = FALSE;
      continue;
    }

    $file = $dir . "/" . $object;
    if (is_dir($file)) {
      _boost_rmdir($file, $flush);
    }
    elseif ($flush) {
      unlink($file);
      $counter++;
    }
    else {
      // Need to handle gzipped files.
      // Nice if  it supported multi level cache expiration per content type.
      $ext = substr(strrchr($file, '.'), 1);
      $age = boost_file_get_age($file);
      if (isset($lifetimes[$ext]) && $age > $lifetimes[$ext]) {
        unlink($file);
        $counter++;
      }
      else {
        $empty_dir = FALSE;
      }
    }
  }
  if ($empty_dir) {
    rmdir($dir);
  }
  return $counter;
}

/**
 * Returns the age of a cached file, measured in seconds since it was last
 * updated.
 *
 * @param $filename
 *   Name of cached file
 * @return int
 */
function boost_file_get_age($filename) {
  return REQUEST_TIME - filemtime($filename);
}

/**
 * Returns a nice html formatted version of print_r
 *
 * @param $data
 *
 * @return string
 */
function boost_print_r($data) {
  return str_replace('    ', '&nbsp;&nbsp;&nbsp;&nbsp;', nl2br(htmlentities(print_r($data, TRUE))));
}

/**
 * Write to a file. Ensures write is atomic via rename operation.
 *
 * @param $filename
 *  relative filename.
 * @param $data
 *  data to write to the file.
 */
function boost_write_file($filename, &$data) {
  // Create directory if it doesn't exist.
  $directory = dirname($filename);
  if (!boost_mkdir($directory)) {
    return FALSE;
  }

  // Save data to a temp file.
  // file_unmanaged_save_data does not use rename.
  $tempname = drupal_tempnam($directory, 'boost');
  if (file_put_contents($tempname, $data) === FALSE) {
    watchdog('boost', 'Could not create the file %file on your system', array('%file' => $tempname), WATCHDOG_ERROR);
    @unlink($tempname);
    return FALSE;
  }

  // Move temp file to real filename; windows can not do a rename replace.
  if (@rename($tempname, $filename) === FALSE) {
    $oldname = $tempname. 'old';
    if (@rename($filename, $oldname) !== FALSE) {
      if (@rename($tempname, $filename) === FALSE) {
        watchdog('boost', 'Could not rename the file %file on your system', array('%file' => $filename), WATCHDOG_ERROR);
        @unlink($tempname);
        @rename($oldname, $filename);
        return FALSE;
      }
      else {
        @unlink($oldname);
      }
    }
  }

  // chmod file so webserver can send it out.
  drupal_chmod($filename);
  return TRUE;
}

/**
 * Create a directory.
 *
 * @param $directory
 *  relative directory.
 */
function boost_mkdir($directory) {
  global $_boost;

  // Only do something if it's not a dir.
  if (!is_dir($directory)) {
    if (!boost_in_cache_dir($directory)) {
      return FALSE;
    }

    // Try to create the directory.
    $mode = variable_get('file_chmod_directory', 0775);
    if (!mkdir($directory, $mode, TRUE)) {
      watchdog('boost', 'Could not create the directory %dir on your system', array('%dir' => $directory), WATCHDOG_ERROR);
      return FALSE;
    }
    drupal_chmod($directory);
  }
  return TRUE;
}

/**
 * Verify that the operation is going to operate in the cache dir.
 *
 * @param $file
 *  relative directory or file.
 */
function boost_in_cache_dir($file) {
  global $_boost;

  $good = TRUE;
  $real_file = realpath($file);
  $cache_dir = variable_get('boost_root_cache_dir', BOOST_ROOT_CACHE_DIR);
  $real_cache_dir = realpath($cache_dir);

  // Only operate in the cache dir.
  // Check the real path.
  if (   strpos($file, $cache_dir) !== 0
      || ($real_file && $real_cache_dir && strpos($real_file, $real_cache_dir) !== 0)
        ) {
    $good = FALSE;
  }

  // Send error to watchdog.
  if (!$good) {
    watchdog('boost', 'An operation outside of the cache directory was attempted on your system. %file or %real_file is outside the cache directory %cache or %real_cache. Debug info below <br> !debug', array(
      '%file'       => $file,
      '%real_file'  => $real_file,
      '%cache'      => $cache_dir,
      '%real_cache' => $real_cache_dir,
      '!debug'      => boost_print_r($_boost),
      ),
    WATCHDOG_CRITICAL);
  }

  return $good;
}

/**
 * Capture error conditions.
 *
 * @param $page_callback_result
 *   The result of a page callback. Can be one of:
 *   - NULL: to indicate no content.
 *   - An integer menu status constant: to indicate an error condition.
 *   - A string of HTML content.
 *   - A renderable array of content.
 *
 * @see drupal_deliver_html_page()
 */
function boost_deliver_html_page($page_callback_result) {
  global $_boost;

  // Menu status constants are integers; page content is a string or array.
  if (is_int($page_callback_result)) {
    // @todo: Break these up into separate functions?
    switch ($page_callback_result) {
      case MENU_NOT_FOUND:
        // 404 page.
        $_boost['menu_item']['status'] = 404;
        break;

      case MENU_ACCESS_DENIED:
        // 403 page.
        $_boost['menu_item']['status'] = 403;
        break;

      case MENU_SITE_OFFLINE:
        // 503 page.
        $_boost['menu_item']['status'] = 503;
        break;
    }
  }

  // Call original function
  drupal_deliver_html_page($page_callback_result);

  // Last possible place to send a watchdog message without a shutdown function.
  if (variable_get('boost_message_debug', BOOST_MESSAGE_DEBUG) && $_boost['args'][0] != 'admin') {
    watchdog('boost', boost_print_r($_boost), array(), WATCHDOG_DEBUG);
  }
}

/**
 * Always run these functions after a form submit from boost.
 */
function boost_form_submit_handler() {
  register_shutdown_function('boost_htaccess_cache_dir_put');
}

/**
 * Overwrite old htaccess rules with new ones.
 */
function boost_htaccess_cache_dir_put() {
  global $_boost, $base_path, $base_root;
  if (empty($_boost['base_dir'])) {
    $url = $base_root . request_uri();
    $parts = parse_url($url);
    $_boost['base_dir'] = boost_get_normal_cache_dir() . '/' . $parts['host'] . $base_path;
  }
  boost_write_file($_boost['base_dir'] . '.htaccess', boost_htaccess_cache_dir_generate());
}

/**
 * Generate htaccess rules for the cache directory.
 */
function boost_htaccess_cache_dir_generate() {
  $char_type = variable_get('boost_charset_type', BOOST_CHARSET_TYPE);
  $etag = variable_get('boost_apache_etag', BOOST_APACHE_ETAG);

  // Go through every storage type getting data needed to build htaccess file.
  $gzip = FALSE;
  $data = array();
  $files = array();
  $types = boost_get_storage_types();
  foreach ($types as $title => $content_types) {
    foreach ($content_types as $type => $values) {
      if ($values['enabled']) {
        $forcetype = '\.' . $values['extension'];
        if ($values['gzip']) {
          $forcetype .= '(?:\.gz)?$';
          $gzip = TRUE;
        }
        else {
          $forcetype .= '$';
        }
        $files[$values['extension']] = $values['extension'];
        $data[$values['extension']] = array(
          'type' => $type,
          'forcetype' => $forcetype,
        );
      }
    }
  }
  if (empty($data) || empty($files)) {
    return FALSE;
  }

  // Add in default charset
  $string = '';
  if (variable_get('boost_add_default_charset', BOOST_ADD_DEFAULT_CHARSET)) {
    $string .= "AddDefaultCharset " . $char_type . "\n";
  }

  // Set FileETag
  if ($etag == 1) {
    $string .= "FileETag None\n";
  }
  elseif ($etag == 2) {
    $string .= "FileETag All\n";
  }
  elseif ($etag == 3) {
    $string .= "FileETag MTime Size\n";
  }

  // Set html expiration time to the past and put in boost header if desired.
  $files = implode('|' , $files);
  if ($gzip) {
    $files .= '(?:\.gz)?';
  }
  $string .= "<FilesMatch \"\.($files)$\">\n";
  $string .= "  <IfModule mod_expires.c>\n";
  $string .= "    ExpiresDefault A5\n";
  $string .= "  </IfModule>\n";
  $string .= "  <IfModule mod_headers.c>\n";
  $string .= "    Header set Expires \"Sun, 19 Nov 1978 05:00:00 GMT\"\n";
  $string .= "    Header unset Last-Modified\n";
  $string .= "    Header append Vary Accept-Encoding\n";
  $string .= "    Header set Cache-Control \"no-store, no-cache, must-revalidate, post-check=0, pre-check=0\"\n";
  if (variable_get('boost_apache_xheader', BOOST_APACHE_XHEADER) > 0) {
    $string .= "    Header set X-Cached-By \"Boost\"\n";
  }
  $string .= "  </IfModule>\n";
  $string .= "</FilesMatch>\n";

  // Set charset and content encoding.
  $string .= "<IfModule mod_mime.c>\n";
  foreach ($data as $extension => $values) {
    $string .= "  AddCharset " . $char_type . " ." . $extension . "\n";
  }
  $string .= $gzip ? "  AddEncoding gzip .gz\n" : '';
  $string .= "</IfModule>\n";

  // Fix for versions of apache that do not respect the T='' RewriteRule
  foreach ($data as $extension => $values) {
    $forcetype = $values['forcetype'];
    $type = $values['type'];
    $string .= "<FilesMatch \"$forcetype\">\n";
    $string .= "  ForceType " . $type . "\n";
    $string .= "</FilesMatch>\n";
  }

  // Make sure files can not execute in the cache dir.
  $string .= "\n";
  $string .= "SetHandler Drupal_Security_Do_Not_Remove_See_SA_2006_006\n";
  $string .= "Options None\n";
  $string .= "Options +FollowSymLinks\n";
  $string .= "\n";

  return $string;
}