wpseek.com
A WordPress-centric search engine for devs and theme authors

function _wp_batch_split_terms() {
	global $wpdb;

	$lock_name = 'term_split.lock';

	// Try to lock.
	$lock_result = $wpdb->query( $wpdb->prepare( "INSERT IGNORE INTO `$wpdb->options` ( `option_name`, `option_value`, `autoload` ) VALUES (%s, %s, 'off') /* LOCK */", $lock_name, time() ) );

	if ( ! $lock_result ) {
		$lock_result = get_option( $lock_name );

		// Bail if we were unable to create a lock, or if the existing lock is still valid.
		if ( ! $lock_result || ( $lock_result > ( time() - HOUR_IN_SECONDS ) ) ) {
			wp_schedule_single_event( time() + ( 5 * MINUTE_IN_SECONDS ), 'wp_split_shared_term_batch' );
			return;
		}
	}

	// Update the lock, as by this point we've definitely got a lock, just need to fire the actions.
	update_option( $lock_name, time() );

	// Get a list of shared terms (those with more than one associated row in term_taxonomy).
	$shared_terms = $wpdb->get_results(
		"SELECT tt.term_id, t.*, count(*) as term_tt_count FROM {$wpdb->term_taxonomy} tt
		 LEFT JOIN {$wpdb->terms} t ON t.term_id = tt.term_id
		 GROUP BY t.term_id
		 HAVING term_tt_count > 1
		 LIMIT 10"
	);

	// No more terms, we're done here.
	if ( ! $shared_terms ) {
		update_option( 'finished_splitting_shared_terms', true );
		delete_option( $lock_name );
		return;
	}

	// Shared terms found? We'll need to run this script again.
	wp_schedule_single_event( time() + ( 2 * MINUTE_IN_SECONDS ), 'wp_split_shared_term_batch' );

	// Rekey shared term array for faster lookups.
	$_shared_terms = array();
	foreach ( $shared_terms as $shared_term ) {
		$term_id                   = (int) $shared_term->term_id;
		$_shared_terms[ $term_id ] = $shared_term;
	}
	$shared_terms = $_shared_terms;

	// Get term taxonomy data for all shared terms.
	$shared_term_ids = implode( ',', array_keys( $shared_terms ) );
	$shared_tts      = $wpdb->get_results( "SELECT * FROM {$wpdb->term_taxonomy} WHERE `term_id` IN ({$shared_term_ids})" );

	// Split term data recording is slow, so we do it just once, outside the loop.
	$split_term_data    = get_option( '_split_terms', array() );
	$skipped_first_term = array();
	$taxonomies         = array();
	foreach ( $shared_tts as $shared_tt ) {
		$term_id = (int) $shared_tt->term_id;

		// Don't split the first tt belonging to a given term_id.
		if ( ! isset( $skipped_first_term[ $term_id ] ) ) {
			$skipped_first_term[ $term_id ] = 1;
			continue;
		}

		if ( ! isset( $split_term_data[ $term_id ] ) ) {
			$split_term_data[ $term_id ] = array();
		}

		// Keep track of taxonomies whose hierarchies need flushing.
		if ( ! isset( $taxonomies[ $shared_tt->taxonomy ] ) ) {
			$taxonomies[ $shared_tt->taxonomy ] = 1;
		}

		// Split the term.
		$split_term_data[ $term_id ][ $shared_tt->taxonomy ] = _split_shared_term( $shared_terms[ $term_id ], $shared_tt, false );
	}

	// Rebuild the cached hierarchy for each affected taxonomy.
	foreach ( array_keys( $taxonomies ) as $tax ) {
		delete_option( "{$tax}_children" );
		_get_term_hierarchy( $tax );
	}

	update_option( '_split_terms', $split_term_data );

	delete_option( $lock_name );
}