run cron only on new or updated content rather than view

You're running the cron job in hook_nodeapi?

1. What conditions would this be run?
2. How would it look in the user administration interface?
3. Would it be run when content is submitted, saved, inserted, etc?
4. What about taxonomy?

It seems to me that running the cron job when a node is updated would bring a lot of overhead to the system. Not only would the cron run every hour or so, it would also run whenever someone updated any content. This would really slow down the system....

Hi,

Can I have a copy your mod?
I really need that so that newly added or modified contents is automatically indexed for my website.

Thanks.

@#2, #3:
As you know in Drupal the idea behind the cron hook is that it allows modules to define "maintenance tasks" that are performed periodically at a frequency set up by the site admin. Examples of such tasks include trimming stale data from some tables, refreshing aggregator feeds, and updating search indexes (which may include indexes for things that are not nodes). What you want to do is update just the node search index immediately a node is submitted (and perhaps also when a comment is submitted). Firing all the cron hooks when a node is submitted isn't the best way of going about this and IMO wouldn't be an appropriate mod to poormanscron, which only exists to provide a way of actually running cron without having to configure a cron job on the server.

A better way of doing this would be to use a tiny custom module to invoke http://api.drupal.org/api/function/node_update_index from the nodeapi hook (or even invoke http://api.drupal.org/api/function/_node_index_node on the node in question). Another approach would be to use Actions/Triggers to fire a custom action on node/comment submit.

Great solution gpk. I'm in the process of learning how to use custom modules with drupal. If you have some time, what is the exact php syntax to invoke the node_update_index function? I plan on hardwiring the php call into a custom php rule that fires after new content is published. I'll keep on looking for an example of invoking other modules that I can imitate. Thanks.

The exact PHP syntax would be just node_update_index() (see the link above: no arguments are shown). This would cause the entire search index for nodes to be updated. Probably there would only be the one node to do, but possibly it might take a few seconds to run if other updates are needed for some reason.

A better way might be to use _node_index_node($object) where $object is the node you have just created/edited and is passed in to your custom action by the trigger/actions system (you will have set the action's 'type' = 'node').

HTH,

Thanks for your help gpk. I successfully wired the node_update_index() function to the action module. I don't have time right now, but hopefully sometime in the future I'll do the more precise solution you suggested above.

For reference, here is the code I am currently using to force nodes to be indexed immediately on node creation/update:

immediate_index.module

/**
 * Implementation of hook_action_info()
 */
function immediate_index_action_info() {
  return array(
    'immediate_index_index_node_action' => array(
      'description' => t('Index post'),
      'type' => 'node',
      'configurable' => FALSE,
      'hooks' => array(
        // We can't use the 'presave' $op since the action of indexing a node
        // retrieves the node from the database, i.e. the action needs to take
        // place after saving the node.
        'nodeapi' => array('insert', 'update'),
        // Could in princple add any/all of the hook_comment() actions, i.e.
        // insert, update, delete, publish, unpublish.
      )
    )
  );
}

/**
 * Implementation of a Drupal action.
 * Indexes a node.
 *
 * This actually gets invoked via the trigger_nodeapi (and potentially
 * trigger_comment) hook implementation, i.e. after search_nodeapi has
 * marked the node as needing to be reindexed. 
 */
function immediate_index_index_node_action($node, $context = array()) {
  // If search.module is not enabled then do nothing. Could alternatively
  // put this functionality in a separate module with search as a dependency.
  if (module_exists('search')) {
    // Need to reset the static node cache!
    node_load(NULL, NULL, TRUE);
    global $user;
    // Save current user
    $current_user = $user;
    // Carry out indexing as anonymous user, for compatibility with CCK content
    // permissions module.
    // Avoid logging the user out in the event something goes wrong (e.g. a
    // drupal_goto() in a PHP node).
    session_save_session(FALSE);
    $user = drupal_anonymous_user();
    _node_index_node($node);
    watchdog('action', 'Indexed @type %title.', array('@type' => node_get_types('name', $node), '%title' => $node->title));
    // This is an essential final step of the indexing process.
    search_update_totals();
    // Restore current user.
    $user = $current_user;
    session_save_session(TRUE);
  }
}

This would also need a .info file of course, which I leave as an exercise for the reader ;-)

Also note that there is a risk (albeit tiny) of typos in the code above because I've renamed the module and therefore have slightly edited the code from what I have running live.