robotstxt-mlp2-mlp5/includes/class-robotstxt-mlp2-mlp5-reader-content-relations.php

<?php
/**
 * MLP2 content relations reader.
 *
 * Reads post and term relation data from the MLP2 backup content_relations
 * table. Relations are grouped by (ml_source_blogid, ml_source_elementid)
 * so that each group can be written as a single MLP5 relationship.
 *
 * MLP2 backup table schema (mlp2bak_content_relations or mlp2bak_relationships):
 *   ml_source_blogid    INT  -- source site ID
 *   ml_source_elementid INT  -- source element (post/term) ID
 *   ml_blogid           INT  -- target site ID
 *   ml_elementid        INT  -- target element ID
 *   ml_type             VARCHAR -- 'post' or 'term'
 *
 * @package Robotstxt_Mlp2_Mlp5
 */

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * Class Robotstxt_Mlp2_Mlp5_Reader_Content_Relations
 */
class Robotstxt_Mlp2_Mlp5_Reader_Content_Relations {

	/**
	 * Count distinct source groups for a given type.
	 *
	 * @param string $type 'post' or 'term'.
	 * @return int
	 */
	public static function count_groups( string $type ): int {
		global $wpdb;

		$mlp2  = new Robotstxt_Mlp2_Mlp5_Detector_Mlp2();
		$table = $mlp2->get_content_relations_table();

		if ( '' === $table ) {
			return 0;
		}

		list( $src_blog, $src_elem ) = self::source_columns( $mlp2, $table );
		if ( '' === $src_blog ) {
			return 0;
		}

		$type_safe = ( 'term' === $type ) ? 'term' : 'post';

		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		return (int) $wpdb->get_var(
			$wpdb->prepare(
				"SELECT COUNT(DISTINCT {$src_blog}, {$src_elem}) FROM `{$table}` WHERE ml_type = %s",
				$type_safe
			)
		);
		// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
	}

	/**
	 * Return a batch of distinct source groups, with all their translation members.
	 *
	 * Each returned element is:
	 *   [
	 *     'type'    => string,   // 'post' or 'term'
	 *     'members' => [
	 *       ['site_id' => int, 'element_id' => int],
	 *       ...
	 *     ]
	 *   ]
	 *
	 * The source element itself is always included as a member.
	 *
	 * @param string $type   'post' or 'term'.
	 * @param int    $offset Offset into the distinct source list.
	 * @param int    $limit  Number of groups to return.
	 * @return array[]
	 */
	public static function get_groups_batch( string $type, int $offset, int $limit ): array {
		global $wpdb;

		$mlp2  = new Robotstxt_Mlp2_Mlp5_Detector_Mlp2();
		$table = $mlp2->get_content_relations_table();

		if ( '' === $table ) {
			return array();
		}

		list( $src_blog, $src_elem ) = self::source_columns( $mlp2, $table );
		if ( '' === $src_blog ) {
			return array();
		}

		$type_safe = ( 'term' === $type ) ? 'term' : 'post';

		// Fetch a page of distinct sources.
		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$sources = $wpdb->get_results(
			$wpdb->prepare(
				"SELECT DISTINCT {$src_blog} AS source_blog, {$src_elem} AS source_elem FROM `{$table}` WHERE ml_type = %s ORDER BY {$src_blog}, {$src_elem} LIMIT %d OFFSET %d",
				$type_safe,
				$limit,
				$offset
			),
			ARRAY_A
		);
		// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

		if ( empty( $sources ) ) {
			return array();
		}

		$groups = array();

		foreach ( $sources as $source ) {
			$src_blog_id = (int) $source['source_blog'];
			$src_elem_id = (int) $source['source_elem'];

			// Fetch all translations for this source.
			// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
			$translations = $wpdb->get_results(
				$wpdb->prepare(
					"SELECT ml_blogid, ml_elementid FROM `{$table}` WHERE ml_type = %s AND {$src_blog} = %d AND {$src_elem} = %d",
					$type_safe,
					$src_blog_id,
					$src_elem_id
				),
				ARRAY_A
			);
			// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

			// Build members: source + all translations (deduplicated).
			$seen    = array();
			$members = array();

			$source_key          = $src_blog_id . ':' . $src_elem_id;
			$seen[ $source_key ] = true;
			$members[]           = array(
				'site_id'    => $src_blog_id,
				'element_id' => $src_elem_id,
			);

			foreach ( (array) $translations as $t ) {
				$key = ( (int) $t['ml_blogid'] ) . ':' . ( (int) $t['ml_elementid'] );
				if ( isset( $seen[ $key ] ) ) {
					continue;
				}
				$seen[ $key ] = true;
				$members[]    = array(
					'site_id'    => (int) $t['ml_blogid'],
					'element_id' => (int) $t['ml_elementid'],
				);
			}

			if ( count( $members ) >= 2 ) {
				$groups[] = array(
					'type'    => $type_safe,
					'members' => $members,
				);
			}
		}

		return $groups;
	}

	/**
	 * Detect the source-blog and source-element column names.
	 *
	 * Returns ['', ''] when columns cannot be determined.
	 *
	 * @param Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2  MLP2 detector.
	 * @param string                            $table Content relations table name.
	 * @return string[] [ $source_blog_col, $source_element_col ]
	 */
	private static function source_columns( Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2, string $table ): array {
		$cols = $mlp2->describe_table( $table );

		$blog_col = '';
		foreach ( array( 'ml_source_blogid', 'ml_blogid' ) as $c ) {
			if ( isset( $cols[ $c ] ) ) {
				$blog_col = $c;
				break;
			}
		}

		$elem_col = '';
		foreach ( array( 'ml_source_elementid', 'ml_elementid' ) as $c ) {
			if ( isset( $cols[ $c ] ) ) {
				$elem_col = $c;
				break;
			}
		}

		return array( $blog_col, $elem_col );
	}
}