What is the correct alternative approach to the hook_apachesolr_node_exclude

You should write a status callback and implement

function hook_apachesolr_entity_info_alter(&$entity_info) {

If you have a good example, please post it here so others can use this as a guidance

This code is 5 lines and you are looking for a more elegant solution? :-)

I'm all up for improvements in documentation ;-) Will commit something in this line to the api docs

I found I had to include the apachesolr.index.inc file before calling apachesolr_index_node_status_callback() otherwise it caused fatal errors when publishing/unpublishing content.

  ...
  // added here
  module_load_include('inc', 'apachesolr', 'apachesolr.index');
  // Use the default apachesolr implementation otherwise.
  return apachesolr_index_node_status_callback($node, $type);
}

Committed, thanks!

thank you for this example, im having trouble getting the hook_apachesolr_entity_info_alter to fire. im on BETA16 and added #2 to my custom module

ive tried
drush solr-delete-index, drush solr-mark-all, drush solr-index
and tried to update my content and publish.

in my callback im just checking for nid == x to remove single nodes buy nid but dont see any changes to my results.

I'm also having trouble with this with BETA16.

I can get the hook_apachesolr_entity_info_alter() to fire (twice) by clearing the cache, but not on indexing content - so my custom node_status_callback isn't getting picked up either.

I'll report back when I find out what's going on...

If you are marking all content to reindex, the process doesn't use the status callback.
The reindex callback is used which basically selects all bundles to be indexed straight from the node table.

See the default function in apachsolr.index.inc:

function apachesolr_index_node_solr_reindex($env_id) {
  $indexer_table = apachesolr_get_indexer_table('node');
  $transaction = db_transaction();
  try {
    db_delete($indexer_table)
      ->condition('entity_type', 'node')
      ->execute();
    $select = db_select('node', 'n');
    $select->addExpression("'node'", 'entity_type');
    $select->addField('n', 'nid', 'entity_id');
    $select->addField('n', 'type', 'bundle');
    $select->addField('n', 'status', 'status');
    $select->addExpression(REQUEST_TIME, 'changed');
    $select->condition('n.type', apachesolr_get_index_bundles($env_id, 'node'), 'IN');

    $insert = db_insert($indexer_table)
      ->fields(array('entity_id', 'bundle', 'status', 'entity_type', 'changed'))
      ->from($select)
      ->execute();
  }
  catch (Exception $e) {
    $transaction->rollback();
    //drupal_set_message($e->getMessage(), 'error');
    watchdog_exception('Apache Solr', $e);
    return FALSE;
  }

  return TRUE;
}

This means if you have something specific in your status callback such as excluding pages that are tagged with a certain term (e.g. no_index), the reindex function won't take that into account and will simply mark all content to index, regardless of whether or not that content is tagged with "no_index".

I had to alter the select query to index only the content I wanted indexed.

That sounds like a bug indeed, we might need to loop over all content to make sure we use this exclude function.

duplicate of #1509790: Status is not checked when a reindex is initiated

Since this issue is linked to in apachesolr.api.php, it'd be helpful to note the correct way to add a status callback, since 'status callback' is now an array (to allow multiple status callbacks, which is great). Updated example from #2:

                  
/**
* Implements hook_apachesolr_entity_info_alter().
*/
function mymodule_apachesolr_entity_info_alter(&$entity_info) {
  $entity_info['node']['status callback'][] = 'mymodule_apachesolr_index_node_status_callback';
}

/**
* Status callback for ApacheSolr, for nodes.
*/
function mymodule_apachesolr_index_node_status_callback($entity_id, $type) {
  $node = node_load($entity_id, NULL, TRUE);
  if ($node->type == 'news' && $node->some_custom_param == TRUE) {
    // Exclude this node which meets some custom criteria.
    return 0;
  }

  return 1;
} 

Thanks!

This issue is not a duplicate, because its about "What is the correct alternative approach to the hook_apachesolr_node_exclude".

And I am wondering, because hook_apachesolr_node_exclude() is still there. So I guess we don't need an alternative anymore. See also #904428: Document how to remove excluded nodes from index.

Setting this to "needs review" in case I am overlooking something.

There is no alternative :) The hook is back in the module