search for tags when a partial string is provided

modsearch_extensionstart.php

<?php
/**
 * @auth: d.gupta
 * @date: Apr/11/12
 * @desc: This file (start.php) along a manifest file constitute a plugin namely "search_extension" that extends the "search"

The core plugin "search" displays tag results only if the inputted string matches the tags exactly and completely
[concerned code: search_tag_hook(..) in /mod/search/search_hooks.php]

However, the following code overrides it by correctly displaying the tag-search-results even when the searched string is a substring of the tag

ANY IMPROVEMENTS/FEEDBACK IS HIGHLY APPRECIATED.

Why this isn't posted as a plugin yet:
From the comments, it seems that there is work and improvements already on its way by the core developers. I am hopeful
to have this worked out in the future versions of the Elgg thereby not requiring this as a plugin at all. Or else, if
 not, I look forward to have some feedback before going down that path

 */

elgg_register_event_handler('init','system','search_extension_init');

function search_extension_init(){
    elgg_unregister_plugin_hook_handler('search','tags','search_tags_hook');
    elgg_register_plugin_hook_handler('search', 'tags', 'search_extension_tags_hook');
}

/**
 * Return default results for searches on tags.
 *
 * @param unknown_type $hook
 * @param unknown_type $type
 * @param unknown_type $value
 * @param unknown_type $params
 * @return unknown_type
 */
function search_extension_tags_hook($hook, $type, $value, $params) {
    $db_prefix = elgg_get_config('dbprefix');
 static $iteration = 0;
    $iteration++;
    $valid_tag_names = elgg_get_registered_tag_metadata_names();

    // @todo will need to split this up to support searching multiple tags at once.
    $query = sanitise_string($params['query']);

    // if passed a tag metadata name, only search on that tag name.
    // tag_name isn't included in the params because it's specific to
    // tag searches.
    if ($tag_names = get_input('tag_names')) {
        if (is_array($tag_names)) {
            $search_tag_names = $tag_names;
        } else {
            $search_tag_names = array($tag_names);
        }
        // check these are valid to avoid arbitrary metadata searches.
        foreach ($search_tag_names as $i => $tag_name) {
            if (!in_array($tag_name, $valid_tag_names)) {
                unset($search_tag_names[$i]);
            }
        }
    } else {
        $search_tag_names = $valid_tag_names;
    }

    if (!$search_tag_names) {
        return array('entities' => array(), 'count' => $count);
    }


    // don't use elgg_get_entities_from_metadata() here because of
    // performance issues.  since we don't care what matches at this point
    // use an IN clause to grab everything that matches at once and sort
    // out the matches later.
    $params['joins'][] = "JOIN {$db_prefix}metadata md on e.guid = md.entity_guid";
    $params['joins'][] = "JOIN {$db_prefix}metastrings msn on md.name_id = msn.id";
    $params['joins'][] = "JOIN {$db_prefix}metastrings msv on md.value_id = msv.id";

    $access = get_access_sql_suffix('md');
    $sanitised_tags = array();

    foreach ($search_tag_names as $tag) {
        $sanitised_tags[] = '"' . sanitise_string($tag) . '"';
    }

    $tags_in = implode(',', $sanitised_tags);

    /** @auth: d.gupta @date: Apr/11/12
     * changed query to LIKE %query% so as to match tags with partial string
     */
    $params['wheres'][] = "(msn.string IN ($tags_in) AND msv.string LIKE '%$query%' AND $access)";

    $params['count'] = TRUE;
    $count = elgg_get_entities($params);

    // no need to continue if nothing here.
    if (!$count) {
        return array('entities' => array(), 'count' => $count);
    }

    $params['count'] = FALSE;
    $entities = elgg_get_entities($params);

    // add the volatile data for why these entities have been returned.
    foreach ($entities as $entity) {
        $matched_tags_strs = array();

        // get tags for each tag name requested to find which ones matched.
        foreach ($search_tag_names as $tag_name) {
            $tags = $entity->getTags($tag_name);

            // @todo make one long tag string and run this through the highlight
            // function.  This might be confusing as it could chop off
            // the tag labels.

            /* @auth: d.gupta @date: Apr/11/12 ----------- */
            /* @desc: this is done for getting display of tags-search-result right. following code modified from the original for replacing the condtion: "in_array(strtolower($query), array_map('strtolower', $tags))" by our calculated boolean: "$query_is_in_tags" (boolean)"
             * how this makes a difference: our calculated condition ($query_is_in_tags) returns true if the searched string is contained/is a substring of the actual tag of the entity
             *the former/orginial code displayed & highlighted the searched tag only if the searched string exactly matched the entity's tag whereas
             */
            $query_lower = strtolower($query);
            $tags_lower = array_map('strtolower',$tags);
            $query_is_in_tags = false;
            foreach($tags_lower as $tag_element){
                if(strpos($tag_element,$query_lower)>=0){
                    $query_is_in_tags = true;
                }
            }
            if ($query_is_in_tags) {
            // ------------------------------------------------------//
                if (is_array($tags)) {
                    $tag_name_str = elgg_echo("tag_names:$tag_name");
                    $matched_tags_strs[] = "$tag_name_str: " . implode(', ', $tags);
                }
            }
        }

        // different entities have different titles
        switch($entity->type) {
            case 'site':
            case 'user':
            case 'group':
                $title_tmp = $entity->name;
                break;

            case 'object':
                $title_tmp = $entity->title;
                break;
        }

        // Nick told me my idea was dirty, so I'm hard coding the numbers.
        $title_tmp = strip_tags($title_tmp);
        if (elgg_strlen($title_tmp) > 297) {
            $title_str = elgg_substr($title_tmp, 0, 297) . '...';
        } else {
            $title_str = $title_tmp;
        }

        $desc_tmp = strip_tags($entity->description);
        if (elgg_strlen($desc_tmp) > 297) {
            $desc_str = elgg_substr($desc_tmp, 0, 297) . '...';
        } else {
            $desc_str = $desc_tmp;
        }

        $tags_str = implode('. ', $matched_tags_strs);

        $tags_str = search_get_highlighted_relevant_substrings($tags_str, $params['query']);

        $entity->setVolatileData('search_matched_title', $title_str);
        $entity->setVolatileData('search_matched_description', $desc_str);
        $entity->setVolatileData('search_matched_extra', $tags_str);
    }

    return array(
        'entities' => $entities,
        'count' => $count,
    );
}