search-it-indexing

Search It – Indexing

Search It stores all searchable content in database tables (rex_tmp_search_it_index, rex_tmp_search_it_keywords). Content must be indexed before it appears in search results.

What gets indexed

By default:

• Articles – all online articles in all languages, rendered to plaintext
• Media files – if configured in backend settings (Additional sources)
• DB columns – any table/column pair configured in backend settings
• PDF files – text extracted via pdftotext (must be installed on server)
• URL addon URLs – if URL addon is installed and enabled in settings

Full reindex

Via backend

Backend > Search It > Generate index > "Generate index" button.

Via console

php redaxo/bin/console search_it:reindex          # full reindex
php redaxo/bin/console search_it:clearCache        # clear search cache only

Via PHP

use FriendsOfRedaxo\SearchIt\SearchIt;

$search = new SearchIt();
$search->generateIndex();   // full reindex

Index individual items

$search = new SearchIt();

// Single article (optionally for specific language)
$search->indexArticle(42);                    // all languages
$search->indexArticle(42, rex_clang::getCurrentId());  // current language only

// Database column
$search->indexColumn(rex::getTable('my_table'), 'description');

// File
$search->indexFile('document.pdf');

Clear index and cache

$search = new SearchIt();
$search->deleteIndex();     // drop entire index (requires regeneration)
$search->deleteCache();     // clear cached search results
$search->deleteKeywords();  // clear similarity keyword index

Automatic re-indexing

Search It hooks into REDAXO extension points via EventHandler and automatically re-indexes articles when they are saved, published or deleted. This happens for:

• ART_ADDED, ART_UPDATED, ART_DELETED
• ART_STATUS (online/offline toggle)
• SLICE_ADDED, SLICE_UPDATED, SLICE_DELETED
• MEDIA_ADDED, MEDIA_UPDATED

No manual action needed for article content changes. But if you change backend settings (e.g. add a new DB column source), you must trigger a full reindex.

Cronjobs

In the backend under Cronjob addon, two cronjob types are available:

• Search it: Reindex – scheduled reindex (full, articles only, columns only, or URLs only)
• Search it: Clear Cache – scheduled cache clearing

Useful for sites with frequently changing external data sources (e.g. DB columns filled by imports).

Plaintext conversion

Articles are fetched via HTTP (or socket), rendered, then converted to plaintext. The PlaintextConverter strips HTML, applies CSS selector exclusions, runs regex replacements and optionally parses Textile.

Configure in backend: Settings > Plaintext settings:

• CSS selectors to exclude – e.g. nav, .no-search, footer (content in these elements is not indexed)
• Regex replacements – custom find/replace patterns applied before indexing
• Strip HTML tags – remove all remaining tags after processing

Customise plaintext via extension point

rex_extension::register('SEARCH_IT_PLAINTEXT', function(rex_extension_point $ep) {
    $text = $ep->getSubject();
    // Remove specific content before indexing
    $text = preg_replace('/<div class="no-index">.*?<\/div>/s', '', $text);
    return $text;
});

Return an array to control further processing:

return ['text' => $cleanedText, 'process' => true];
// process = true: standard plaintext conversion still runs after your hook
// process = false: use your text as-is, skip built-in conversion

PDF indexing

Requires pdftotext (from poppler-utils) on the server:

apt-get install poppler-utils    # Debian/Ubuntu

Search It uses PdfConverter to extract text from PDF files in the media pool. Enable file indexing and add pdf to the allowed extensions in backend settings.

Database tables

Table	Purpose
rex_tmp_search_it_index	Main index (plaintext, metadata per article/column/file)
rex_tmp_search_it_keywords	Keywords for similarity search
rex_tmp_search_it_cache	Cached search results
rex_tmp_search_it_cacheindex_ids	Links cache entries to index entries
rex_tmp_search_it_stats_searchterms	Search term statistics

Tables use the tmp_ prefix because they are regenerable – the index can always be rebuilt from source content.

Common pitfalls

• Adding a new DB column in the backend settings but forgetting to reindex – the column data does not appear until the next full index run.
• Running generateIndex() on every page load – this is an expensive operation. Only call it from console, cronjob, or backend.
• Expecting immediate results after indexArticle() – if the search cache contains stale results, call deleteCache() as well.
• Forgetting poppler-utils on the server – PDF files are silently skipped if pdftotext is not available.
• Indexing articles on a domain that requires HTTPS but having SSL verification enabled in settings when using a self-signed cert – indexing fails silently. Disable SSL verification for local development.