Configuring MySQL Full-Text Indexes for 1C-Bitrix

-- Statistics of InnoDB FULLTEXT indexes
SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_CACHE;
SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_TABLE;
-- Forced rebuild
SET GLOBAL innodb_optimize_fulltext_only = ON;
OPTIMIZE TABLE b_search_content;
SET GLOBAL innodb_optimize_fulltext_only = OFF;

These queries help track the index population and rebuild it when necessary.

Developing a custom component for FULLTEXT

Overriding search in Bitrix

The standard bitrix:search.page does not use FULLTEXT — it works through Bitrix ORM with LIKE. To connect FULLTEXT, we override the query in a custom wrapper component or via the OnBeforeIBlockElementGetList event.

Minimal custom search via FULLTEXT:

namespace Local\Search;

class FulltextSearcher { private \Bitrix\Main\DB\Connection $db; public function __construct() { $this->db = \Bitrix\Main\Application::getConnection(); }

    public function search(string $query, int $page = 1, int $limit = 20): array
    {
        $query = $this->sanitizeQuery($query);
        $offset = ($page - 1) * $limit;

        // BOOLEAN MODE with prefix search
        $boolQuery = '+' . implode('* +', explode(' ', $query)) . '*';

        $sql = "
            SELECT sc.ID, sc.TITLE, sc.URL, sc.MODULE_ID, sc.ITEM_ID,
                   MATCH(sc.TITLE, sc.BODY) AGAINST (? IN BOOLEAN MODE) AS relevance
            FROM b_search_content sc
            WHERE MATCH(sc.TITLE, sc.BODY) AGAINST (? IN BOOLEAN MODE)
              AND sc.SITE_ID = ?
              AND sc.PUBLIC = 'Y'
            ORDER BY relevance DESC
            LIMIT ? OFFSET ?
        ";

$result = $this->db->query($sql, [$boolQuery, $boolQuery, SITE_ID, $limit, $offset]); $rows = []; while ($row = $result->fetch()) { $rows[] = $row; }

return $rows; } public function count(string $query): int { $boolQuery = '+' . implode('* +', explode(' ', $query)) . '*';

        $result = $this->db->query(
            "SELECT COUNT(*) AS cnt FROM b_search_content WHERE MATCH(TITLE, BODY) AGAINST (? IN BOOLEAN MODE) AND SITE_ID = ? AND PUBLIC = 'Y'",
            [$boolQuery, SITE_ID]
        );

        return (int)$result->fetch()['cnt'];
    }

    private function sanitizeQuery(string $query): string
    {
        $query = preg_replace('/[+\-><()\~*"@]+/', ' ', $query);
        $query = preg_replace('/\s+/', ' ', trim($query));
        return mb_substr($query, 0, 255);
    }
}

Custom search component

The component template uses FulltextSearcher instead of the standard module:

// /local/components/local/search.fulltext/component.php
if (!defined('B_PROLOG_INCLUDED') || B_PROLOG_INCLUDED !== true) die();

$query = trim($_GET['q'] ?? '');
if (mb_strlen($query) < 2) {
    $this->arResult['ITEMS'] = [];
    $this->arResult['TOTAL'] = 0;
    $this->IncludeComponentTemplate();
    return;
}

$searcher = new \Local\Search\FulltextSearcher();
$page = max(1, (int)($_GET['PAGEN_1'] ?? 1));
$this->arResult['ITEMS'] = $searcher->search($query, $page);
$this->arResult['TOTAL'] = $searcher->count($query);
$this->arResult['QUERY'] = htmlspecialchars($query);
$this->arResult['PAGE'] = $page;
$this->SetResultCacheKeys([]); // we do not cache search
$this->IncludeComponentTemplate();

Performance and parser comparison

Comparative metrics

How to implement FULLTEXT in 5 steps

1. Audit: measure the current search time, analyze the data volume.
2. MySQL configuration: set ngram_token_size=2, disable stopwords.
3. Creating indexes: run ALTER TABLE with FULLTEXT and the ngram parser.
4. Component development: create a custom FulltextSearcher and template.
5. Testing: load testing, "before/after" comparison.

What is included in the work

• Audit of the current search index, measurement of query time.
• Configuring MySQL: ngram_token_size, disabling stopwords.
• Creating FULLTEXT indexes on b_search_content and/or infoblock tables.
• Developing a custom search component with FULLTEXT queries.
• Rebuilding the Bitrix search index (BXSearch::reindex()).
• Load testing with a "before/after" report.
• Providing documentation and migration scripts.
• Free support for one month.

Our experience in Bitrix development is over 10 years, we have successfully accelerated search on 50+ projects. Get a consultation — write to us, we will evaluate your project within 1 day.

According to the MySQL 8.0 documentation, the ngram parser ensures correct indexing of Cyrillic without external dependencies (MySQL Ngram Full-Text Parser).