Allowing several bundles to share the same field name within a given entity type after field_mapping_alter

Patch as promised.

The two patches do the same thing, but one is more compact as it doesn't fix new spacing to account for added loops.

It's easier to read the patch without spaces, but the one with spaces is more correct.

R

It's hard to show an "after" snippet, because the result of the change is only visible in the effectively indexed Solr fields. However, I have this code I can show to enlighten the problem further.

The first piece of code is adapted from a custom module I developed that implements the field_mappings_alter hook. It basically takes semantically equal fields and translates them to a commonly named Solr field. An example usage would be where different fields only use and display a sub-set of a "common taxonomy" based on criteria we can't cover when re-using fields across content types.

/**
 * Implements hook_apachesolr_field_mappings_alter()
 */
function apachesolr_normalize_fields_apachesolr_field_mappings_alter(&$mappings, $entity_type) {

  // Enable indexing for text fields
  $mappings['my_text'] = array(
    'indexing_callback' => 'apachesolr_fields_default_indexing_callback',
    'map callback' => '',
    'index_type' => 'string',
    'facets' => TRUE,
    'facet missing allowed' => TRUE,
    'dependency plugins' => array('bundle', 'role'),
    'hierarchy callback' => FALSE,
    'name callback' => '',
    'facet mincount allowed' => FALSE,
    'multiple' => FALSE,
  );    
    
  $mappings['per-field']['field_educode_articles'] = $mappings['my_text'];
  $mappings['per-field']['field_educode_articles']['name callback'] = 'apachesolr_normalize_fields_nameconvert';
  $mappings['per-field']['field_educode_courses'] = $mappings['my_text'];
  $mappings['per-field']['field_educode_courses']['name callback'] = 'apachesolr_normalize_fields_nameconvert';  
}

/**
* Callback that generates a sub-part of a Solr field name from a Drupal field name
*   - simple test implementation that drops everything after the underscore
*        eg.: field_educode_courses >> field_ecucode
*               field_educode_articles >> field_educode
*/
function apachesolr_normalize_fields_nameconvert ($field) {
    $name = explode('_', $field['field_name']);
    unset($name[count($name)-1]);
    return implode('_', $name); 
}

Without the patch, only one of the two mapped fields are indexed, because of this line of code in apachesolr.module:

    $fields[$entity_type][apachesolr_index_key($row)] = $row;

The field that is indexed depends on what field is processed last. Changing this line of code to:

    $fields[$entity_type][apachesolr_index_key($row)][] = $row;

will "stack" the different fields on top of each other; thus needing a "loop" to process the stacked $rows (3 places in the code).

Was this what you had in mind when asking for snippets?

R

You understand correctly, no changes in the API have been done, nor should the change impact the API, and it only fixes this one issue.

R

Thanks Nick! I'll hopefully be using this module heavily in the time to come, so this might not be the last you hear from me. And if you don't, it just means everything works :)

R