', + '#markup' => t('In order for multisite setting to work, and to keep the main robots.txt file in place, ' . $suggestion['message'], array('%file' => $suggestion['file'])), + '#suffix' => '
' . $suggestion['code'], + ); + $form['root']['note'] = array( + '#type' => 'markup', + '#prefix' => '', + ); + } + else { + $form['root']['software'] = array( + '#type' => 'markup', + '#prefix' => '', + ); + } + + // Build a table + $marker = _robotstxt_check_marker(); + + $rows = array( + 0 => array( + 'data' => array( + array('data' => $marker['site']), + array('data' => $marker['original'] ? 'NO' : 'YES'), + array('data' => $marker['path']), + array('data' => $marker['exists'] ? 'OK' : 'Missing'), + ), + 'no_striping' => TRUE, + 'class' => $marker['exists'] ? 'ok' : 'error', + ) + ); + + $form['markers'] = array( + '#theme' => 'table', + '#header' => array('Site / Configurtion Folder', 'Alias', 'Marker file path', 'Status'), + '#rows' => $rows, + ); + + $form['actions'] = array('#type' => 'actions'); + if (!$marker['exists']) { + $form['actions']['generate'] = array( + '#type' => 'submit', + '#value' => t('Generate marker'), + '#submit' => array('robotstxt_multisite_settings_submit_generate'), + ); + } + else { + $form['actions']['delete'] = array( + '#type' => 'submit', + '#value' => t('Delete marker'), + '#submit' => array('robotstxt_multisite_settings_submit_delete'), + ); + } + + return $form; +} + +/** + * Submit callback + * + * @param $form + * @param $form_state + */ +function robotstxt_multisite_settings_submit_generate($form, &$form_state) { + $marker = _robotstxt_check_marker(); + if (!$marker['exists']) { + if (!touch($marker['path'])) { + drupal_set_message(t('Could not create file !file', array('!file' => $marker['path'])), 'error'); + } + } +} + +function robotstxt_multisite_settings_submit_delete($form, &$form_state) { + $marker = _robotstxt_check_marker(); + if ($marker['exists']) { + if (!unlink($marker['path'])) { + drupal_set_message(t('Could not create file !file', array('!file' => $marker['path'])), 'error'); + } + } +} + +/** + * Generates a list of marker files and checks their existence. + * + * Marker files are necessary for Apache/Nginx to craft a special request so that the + * module could return custom robots.txt file. Each file is associated with the site name, + * either original or an alias defined in sites.php. + * + * @return array Array of marker files data. Each entry contains the following: + * - original: bool, whether this file is placed in the conf folder or is an alias + * - sites: bool, whether this file is created for a alias defined in sites.php + * - file: string, file path relative to the DRUPAL_ROOT + * - exists: bool. whether file exists or not + */ +function _robotstxt_check_marker() { + global $base_url; + + $conf_dir = conf_path(); + $site_dir = substr(conf_path(), 6); + $host = parse_url($base_url, PHP_URL_HOST); + $fs = realpath(DRUPAL_ROOT . DIRECTORY_SEPARATOR . variable_get('file_public_path', $conf_dir . '/files')); + + $marker = array( + 'site' => $site_dir, + 'original' => TRUE, + ); + + // If $host matches $conf_dir - we got direct hit to the site folder! + if ($host === $site_dir) { + $marker['path'] = $fs . DIRECTORY_SEPARATOR . '.robotstxt'; + } + else if ('default' === $site_dir) { + $marker['path'] = $fs . DIRECTORY_SEPARATOR . ".{$host}.robotstxt"; + $marker['original'] = FALSE; + } + else { + // Look for suggested default folder + $default = variable_get('robotstxt_override_default', 'sites/default/files'); + $fs = realpath(DRUPAL_ROOT . DIRECTORY_SEPARATOR . $default); + $marker['path'] = $fs . DIRECTORY_SEPARATOR . ".{$host}.robotstxt"; + $marker['original'] = FALSE; + } + + $marker['exists'] = file_exists( $marker['path']); + + return $marker; +} + +/** + * Helper function + * Provides suggestions for server configuration changes + * + * @param string $server Server name in lower case + * @return bool|string[] Suggested server data, or FALSE + */ +function _robotstxt_server_suggestions($server) { + $suggestions = array( + 'apache' => array( + 'file' => '.htaccess', + 'message' => 'the following addition to your %file file is necessary:', + 'code' => "
+RewriteCond %{REQUEST_URI} ^/robots.txt$
+RewriteCond %{DOCUMENT_ROOT}/sites/%{HTTP_HOST}/files/.robotstxt -f [OR]
+RewriteCond %{DOCUMENT_ROOT}/sites/default/files/.%{HTTP_HOST}.robotstxt -f
+RewriteRule ^ index.php?q=robots.txt [L]
+",
+ ),
+ 'nginx' => array(
+ 'file' => 'rules',
+ 'message' => 'the following update to your %file file is necessary:',
+ 'code' => '
+location = /robots.txt {
+ allow all;
+ log_not_found off;
+ access_log off;
+ if (-f $document_root/sites/$host/files/.robotstxt) {
+ rewrite ^ /index.php?q=robots.txt;
+ }
+ if (-f $document_root/sites/default/files/.$host.robotstxt) {
+ rewrite ^ /index.php?q=robots.txt;
+ }
+}
+',
+ )
+ );
+ return isset($suggestions[$server]) ? $suggestions[$server] : FALSE;
+}
+
+
+/**
+ * Checks whether the default folder exists
+ *
+ * * Should the check for the /sites/default/[files] folder fail - aliases will not be recognized
+ *
+ * The web server and Drupal do not share information about Drupal's file system, neither for
+ * single site nor multi-site configuration. In order to do our best and allow the web server
+ * to recognize aliased web sites, we are making the following assumptions:
+ * - /sites/default folder exists
+ * - /sites/default/[files] public file system folder exists and is writable, where we assume
+ * that [files] part is the standard and is the same for any other site on the platform. There
+ * is no true reason to have this configured differently.
+ *
+ * Further, we will use default public file system folder to store alias-specific markers thus
+ * allowing the web server check against this markers for sites which are accessed via an alias.
+ *
+ * !!! IT IS STRICTLY REQUIRED FOR THE MULTI-SITE ENVIRONMENT TO HAVE STANDARD !!!
+ * PUBLIC FILE SYSTEM FOLDER SETTING, OTHERWISE THIS MODULE IS
+ * LIMITED IN ITS ABILITY TO RESOLVE ROBOTS.TXT SETTINGS CORRECTLY.
+ *
+ * @return boolean TRUE is folder exists, FALSE otherwise
+ */
+function _robotstxt_check_default_folder() {
+ $fs = variable_get('robotstxt_platform_site_filesystem', '/files');
+ $dir = join(DIRECTORY_SEPARATOR, array(DRUPAL_ROOT,'sites','default', $fs));
+ return is_dir($dir) && is_writable($dir);
+}
diff --git a/robotstxt.module b/robotstxt.module
index 6e23fd6..61a660f 100644
--- a/robotstxt.module
+++ b/robotstxt.module
@@ -27,6 +27,10 @@ function robotstxt_permission() {
'title' => t('Administer robots.txt'),
'description' => t('Perform maintenance tasks for robots.txt.'),
),
+ 'administer multisite robots.txt' => array(
+ 'title' => t('Administer multisite robots.txt'),
+ 'description' => t('Perform maintenance tasks for robots.txt in a multisite environment.'),
+ ),
);
}
@@ -47,6 +51,20 @@ function robotstxt_menu() {
'access arguments' => array('administer robots.txt'),
'file' => 'robotstxt.admin.inc',
);
+ $items['admin/config/search/robotstxt/content'] = array(
+ 'title' => 'Content',
+ 'description' => 'Manage contents of your robots.txt file.',
+ 'type' => MENU_DEFAULT_LOCAL_TASK
+ );
+ $items['admin/config/search/robotstxt/multisite'] = array(
+ 'title' => 'Multisite Support',
+ 'description' => 'Manage robots.txt file multisite support.',
+ 'page callback' => 'drupal_get_form',
+ 'page arguments' => array('robotstxt_multisite_settings'),
+ 'access arguments' => array('administer multisite robots.txt'),
+ 'file' => 'robotstxt.admin.inc',
+ 'type' => MENU_LOCAL_TASK
+ );
return $items;
}