<?php
/**
 * Wizard Step 2 — Analysis.
 *
 * Scans MLP2 backup tables and MLP5 tables to count records, detect
 * orphaned content, and validate schema compatibility before migration.
 *
 * @package Robotstxt_Mlp2_Mlp5
 */

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

/**
 * Class Robotstxt_Mlp2_Mlp5_Step_Analysis
 */
class Robotstxt_Mlp2_Mlp5_Step_Analysis {

	/**
	 * Run the full analysis and return a structured result.
	 *
	 * @return array {
	 *     @type int      $site_relations_count   MLP2 site relation rows.
	 *     @type int      $post_groups_count       Distinct MLP2 post-relation groups (by source).
	 *     @type int      $term_groups_count       Distinct MLP2 term-relation groups (by source).
	 *     @type int[]    $orphaned_site_ids       Site IDs referenced in MLP2 that no longer exist.
	 *     @type int      $orphaned_posts_count    Count of orphaned post references (sampled).
	 *     @type int      $orphaned_terms_count    Count of orphaned term references (sampled).
	 *     @type int      $duplicate_relations     Count of duplicate content relation rows.
	 *     @type bool     $mlp5_schema_ok          True if MLP5 tables have expected columns.
	 *     @type string   $mlp5_schema_error       Human-readable schema error, or empty string.
	 *     @type string   $mlp2_content_table      Backup content relations table name.
	 *     @type string   $mlp2_site_table         Backup site relations table name.
	 *     @type string   $mlp5_content_table      MLP5 content relations table name.
	 *     @type string   $mlp5_site_table         MLP5 site relations table name.
	 *     @type string   $mlp5_relationships_table MLP5 relationships group table name.
	 * }
	 */
	public static function run(): array {
		global $wpdb;

		$mlp2 = new Robotstxt_Mlp2_Mlp5_Detector_Mlp2();
		$mlp5 = new Robotstxt_Mlp2_Mlp5_Detector_Mlp5();

		$mlp2_site_table    = $mlp2->get_site_relations_table();
		$mlp2_content_table = $mlp2->get_content_relations_table();
		$mlp5_site_table    = $mlp5->get_site_relations_table();
		$mlp5_content_table = $mlp5->get_content_relations_table();
		$mlp5_rel_table     = $mlp5->get_relationships_table();

		// --- Counts ---
		$site_count = '';
		if ( '' !== $mlp2_site_table ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
			$site_count = (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$mlp2_site_table}`" );
		}

		$post_count = 0;
		$term_count = 0;
		if ( '' !== $mlp2_content_table ) {
			$col = self::source_blog_col( $mlp2, $mlp2_content_table );
			$id  = self::source_element_col( $mlp2, $mlp2_content_table );

			if ( '' !== $col && '' !== $id ) {
				// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
				$post_count = (int) $wpdb->get_var( "SELECT COUNT(DISTINCT {$col}, {$id}) FROM `{$mlp2_content_table}` WHERE ml_type = 'post'" );
				$term_count = (int) $wpdb->get_var( "SELECT COUNT(DISTINCT {$col}, {$id}) FROM `{$mlp2_content_table}` WHERE ml_type = 'term'" );
				// phpcs:enable
			}
		}

		// --- Orphaned sites ---
		$orphaned_sites = self::detect_orphaned_sites( $mlp2_site_table );

		// --- Orphaned posts/terms (sampled) ---
		$orphaned_posts = self::detect_orphaned_content( $mlp2_content_table, 'post', 100 );
		$orphaned_terms = self::detect_orphaned_content( $mlp2_content_table, 'term', 100 );

		// --- Duplicates ---
		$duplicates = self::count_duplicate_relations( $mlp2_content_table );

		// --- MLP5 schema validation ---
		list( $schema_ok, $schema_error ) = self::validate_mlp5_schema( $mlp5_site_table, $mlp5_content_table, $mlp5_rel_table, $mlp2, $mlp2_site_table, $mlp2_content_table );

		return array(
			'site_relations_count'     => (int) $site_count,
			'post_groups_count'        => $post_count,
			'term_groups_count'        => $term_count,
			'orphaned_site_ids'        => $orphaned_sites,
			'orphaned_posts_count'     => $orphaned_posts,
			'orphaned_terms_count'     => $orphaned_terms,
			'duplicate_relations'      => $duplicates,
			'mlp5_schema_ok'           => $schema_ok,
			'mlp5_schema_error'        => $schema_error,
			'mlp2_content_table'       => $mlp2_content_table,
			'mlp2_site_table'          => $mlp2_site_table,
			'mlp5_content_table'       => $mlp5_content_table,
			'mlp5_site_table'          => $mlp5_site_table,
			'mlp5_relationships_table' => $mlp5_rel_table,
		);
	}

	/**
	 * Render the Step 2 UI.
	 */
	public static function render(): void {
		$data   = Robotstxt_Mlp2_Mlp5_Wizard_State::get_step_data( 2 );
		$ran    = ! empty( $data );
		$ran_ok = $ran && ! empty( $data['mlp5_schema_ok'] );
		?>
		<h2><?php esc_html_e( 'Step 2 — Analysis', 'robotstxt-mlp2-mlp5' ); ?></h2>

		<?php if ( ! $ran ) : ?>
			<p><?php esc_html_e( 'Click "Run Analysis" to scan the MLP2 backup database and validate MLP5 schema compatibility before migrating.', 'robotstxt-mlp2-mlp5' ); ?></p>
			<form method="post" action="<?php echo esc_url( admin_url( 'admin-post.php' ) ); ?>">
				<?php wp_nonce_field( 'mlp_migrator_analysis', 'mlp_migrator_nonce' ); ?>
				<input type="hidden" name="action" value="mlp_migrator_step" />
				<input type="hidden" name="mlp_migrator_step_action" value="2_run" />
				<p><button type="submit" class="button button-primary"><?php esc_html_e( 'Run Analysis', 'robotstxt-mlp2-mlp5' ); ?></button></p>
			</form>
		<?php else : ?>

			<?php if ( ! $data['mlp5_schema_ok'] ) : ?>
				<div class="notice notice-error inline" style="margin:0 0 1em;">
					<p><strong><?php esc_html_e( 'MLP5 schema incompatibility detected:', 'robotstxt-mlp2-mlp5' ); ?></strong> <?php echo esc_html( $data['mlp5_schema_error'] ); ?></p>
				</div>
			<?php endif; ?>

			<table class="widefat striped" style="border:0;margin-bottom:1.5em;">
				<thead><tr><th><?php esc_html_e( 'Category', 'robotstxt-mlp2-mlp5' ); ?></th><th><?php esc_html_e( 'Records found', 'robotstxt-mlp2-mlp5' ); ?></th></tr></thead>
				<tbody>
					<tr><td><?php esc_html_e( 'Site relations', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['site_relations_count'] ) ); ?></td></tr>
					<tr><td><?php esc_html_e( 'Post relation groups', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['post_groups_count'] ) ); ?></td></tr>
					<tr><td><?php esc_html_e( 'Term relation groups', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['term_groups_count'] ) ); ?></td></tr>
				</tbody>
			</table>

			<?php if ( ! empty( $data['orphaned_site_ids'] ) || $data['orphaned_posts_count'] > 0 || $data['orphaned_terms_count'] > 0 || $data['duplicate_relations'] > 0 ) : ?>
				<h3><?php esc_html_e( 'Conflicts (will be skipped and logged)', 'robotstxt-mlp2-mlp5' ); ?></h3>
				<table class="widefat striped" style="border:0;margin-bottom:1.5em;">
					<tbody>
						<?php if ( ! empty( $data['orphaned_site_ids'] ) ) : ?>
							<tr><td><?php esc_html_e( 'Orphaned sites', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( implode( ', ', array_map( 'intval', $data['orphaned_site_ids'] ) ) ); ?></td></tr>
						<?php endif; ?>
						<?php if ( $data['orphaned_posts_count'] > 0 ) : ?>
							<tr><td><?php esc_html_e( 'Orphaned post references', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['orphaned_posts_count'] ) ); ?></td></tr>
						<?php endif; ?>
						<?php if ( $data['orphaned_terms_count'] > 0 ) : ?>
							<tr><td><?php esc_html_e( 'Orphaned term references', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['orphaned_terms_count'] ) ); ?></td></tr>
						<?php endif; ?>
						<?php if ( $data['duplicate_relations'] > 0 ) : ?>
							<tr><td><?php esc_html_e( 'Duplicate relation rows', 'robotstxt-mlp2-mlp5' ); ?></td><td><?php echo esc_html( number_format_i18n( (int) $data['duplicate_relations'] ) ); ?></td></tr>
						<?php endif; ?>
					</tbody>
				</table>
			<?php endif; ?>

			<p>
				<form method="post" action="<?php echo esc_url( admin_url( 'admin-post.php' ) ); ?>" style="display:inline;">
					<?php wp_nonce_field( 'mlp_migrator_analysis', 'mlp_migrator_nonce' ); ?>
					<input type="hidden" name="action" value="mlp_migrator_step" />
					<input type="hidden" name="mlp_migrator_step_action" value="2_run" />
					<button type="submit" class="button button-secondary"><?php esc_html_e( 'Re-run Analysis', 'robotstxt-mlp2-mlp5' ); ?></button>
				</form>
				&nbsp;
				<form method="post" action="<?php echo esc_url( admin_url( 'admin-post.php' ) ); ?>" style="display:inline;">
					<?php wp_nonce_field( 'mlp_migrator_step_2', 'mlp_migrator_nonce' ); ?>
					<input type="hidden" name="action" value="mlp_migrator_step" />
					<input type="hidden" name="mlp_migrator_step_action" value="2" />
					<button type="submit" class="button button-primary"<?php echo $ran_ok ? '' : ' disabled'; ?>><?php esc_html_e( 'Next: Migrate Site Relations', 'robotstxt-mlp2-mlp5' ); ?></button>
				</form>
			</p>
		<?php endif; ?>
		<?php
	}

	// -------------------------------------------------------------------------
	// Private helpers
	// -------------------------------------------------------------------------

	/**
	 * Detect orphaned site IDs in the MLP2 site relations table.
	 *
	 * @param string $table MLP2 backup site relations table name.
	 * @return int[] Site IDs that no longer exist in the network.
	 */
	private static function detect_orphaned_sites( string $table ): array {
		if ( '' === $table ) {
			return array();
		}

		global $wpdb;

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$rows = $wpdb->get_results( "SELECT site_1, site_2 FROM `{$table}`", ARRAY_A );

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

		$ids = array();
		foreach ( $rows as $row ) {
			$ids[] = (int) $row['site_1'];
			$ids[] = (int) $row['site_2'];
		}
		$ids = array_unique( $ids );

		$existing = get_sites(
			array(
				'site__in' => $ids,
				'fields'   => 'ids',
				'number'   => count( $ids ),
			)
		);
		$existing = array_map( 'intval', $existing );

		return array_values( array_diff( $ids, $existing ) );
	}

	/**
	 * Sample up to $sample content relation rows and count orphaned elements.
	 *
	 * @param string $table  MLP2 backup content relations table name.
	 * @param string $type   'post' or 'term'.
	 * @param int    $sample Maximum rows to inspect.
	 * @return int Count of orphaned references found in the sample.
	 */
	private static function detect_orphaned_content( string $table, string $type, int $sample ): int {
		if ( '' === $table ) {
			return 0;
		}

		global $wpdb;

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

		// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$rows = $wpdb->get_results(
			$wpdb->prepare(
				"SELECT ml_blogid, ml_elementid FROM `{$table}` WHERE ml_type = %s LIMIT %d",
				$type_safe,
				$sample
			),
			ARRAY_A
		);
		// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared

		if ( empty( $rows ) ) {
			return 0;
		}

		$orphans = 0;
		foreach ( $rows as $row ) {
			$site_id    = (int) $row['ml_blogid'];
			$element_id = (int) $row['ml_elementid'];

			if ( ! get_site( $site_id ) ) {
				++$orphans;
				continue;
			}

			switch_to_blog( $site_id );
			if ( 'term' === $type_safe ) {
				$exists = (bool) get_term( $element_id );
			} else {
				$exists = (bool) get_post( $element_id );
			}
			restore_current_blog();

			if ( ! $exists ) {
				++$orphans;
			}
		}

		return $orphans;
	}

	/**
	 * Count duplicate rows in the content relations table.
	 *
	 * @param string $table MLP2 backup content relations table name.
	 * @return int Count of duplicate rows.
	 */
	private static function count_duplicate_relations( string $table ): int {
		if ( '' === $table ) {
			return 0;
		}

		global $wpdb;

		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$total = (int) $wpdb->get_var( "SELECT COUNT(*) FROM `{$table}`" );
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared
		$distinct = (int) $wpdb->get_var( "SELECT COUNT(*) FROM (SELECT DISTINCT ml_source_blogid, ml_source_elementid, ml_blogid, ml_elementid, ml_type FROM `{$table}`) AS d" );

		return max( 0, $total - $distinct );
	}

	/**
	 * Validate that MLP5 tables have the expected columns.
	 *
	 * @param string                            $mlp5_site     MLP5 site relations table.
	 * @param string                            $mlp5_content  MLP5 content relations table.
	 * @param string                            $mlp5_rel      MLP5 relationships table.
	 * @param Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2 MLP2 detector.
	 * @param string                            $mlp2_site     MLP2 backup site relations table.
	 * @param string                            $mlp2_content  MLP2 backup content relations table.
	 * @return array [ bool $ok, string $error_message ]
	 */
	private static function validate_mlp5_schema(
		string $mlp5_site,
		string $mlp5_content,
		string $mlp5_rel,
		Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2,
		string $mlp2_site,
		string $mlp2_content
	): array {
		if ( '' === $mlp5_site || '' === $mlp5_content ) {
			return array( false, __( 'MLP5 tables not found. Ensure MultilingualPress 5 is active and has been initialized.', 'robotstxt-mlp2-mlp5' ) );
		}

		// Site relations: expect site_1 and site_2 columns.
		$site_cols = $mlp2->describe_table( $mlp5_site );
		if ( ! isset( $site_cols['site_1'], $site_cols['site_2'] ) ) {
			return array( false, sprintf( /* translators: %s: table name */ __( 'Table %s does not have expected columns (site_1, site_2).', 'robotstxt-mlp2-mlp5' ), $mlp5_site ) );
		}

		// Content relations: expect relationship_id, site_id, content_id.
		$content_cols = $mlp2->describe_table( $mlp5_content );
		if ( ! isset( $content_cols['relationship_id'], $content_cols['site_id'], $content_cols['content_id'] ) ) {
			return array( false, sprintf( /* translators: %s: table name */ __( 'Table %s does not have expected columns (relationship_id, site_id, content_id).', 'robotstxt-mlp2-mlp5' ), $mlp5_content ) );
		}

		// Relationships group table: expect id and type columns.
		if ( '' !== $mlp5_rel ) {
			$rel_cols = $mlp2->describe_table( $mlp5_rel );
			if ( ! isset( $rel_cols['id'], $rel_cols['type'] ) ) {
				return array( false, sprintf( /* translators: %s: table name */ __( 'Table %s does not have expected columns (id, type).', 'robotstxt-mlp2-mlp5' ), $mlp5_rel ) );
			}
		}

		// MLP2 backup: check the content table has the known columns.
		if ( '' !== $mlp2_content ) {
			$mlp2_cols = $mlp2->describe_table( $mlp2_content );
			$required  = array( 'ml_blogid', 'ml_elementid', 'ml_type' );
			foreach ( $required as $col ) {
				if ( ! isset( $mlp2_cols[ $col ] ) ) {
					return array( false, sprintf( /* translators: 1: column name, 2: table name */ __( 'MLP2 backup table %2$s is missing expected column "%1$s". Schema may differ from expected.', 'robotstxt-mlp2-mlp5' ), $col, $mlp2_content ) );
				}
			}
		}

		return array( true, '' );
	}

	/**
	 * Detect the source-blog column name for the MLP2 content relations table.
	 *
	 * @param Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2  MLP2 detector.
	 * @param string                            $table Content relations table name.
	 * @return string Column name or empty string.
	 */
	private static function source_blog_col( Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2, string $table ): string {
		$cols = $mlp2->describe_table( $table );
		foreach ( array( 'ml_source_blogid', 'ml_blogid' ) as $candidate ) {
			if ( isset( $cols[ $candidate ] ) ) {
				return $candidate;
			}
		}
		return '';
	}

	/**
	 * Detect the source-element column name for the MLP2 content relations table.
	 *
	 * @param Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2  MLP2 detector.
	 * @param string                            $table Content relations table name.
	 * @return string Column name or empty string.
	 */
	private static function source_element_col( Robotstxt_Mlp2_Mlp5_Detector_Mlp2 $mlp2, string $table ): string {
		$cols = $mlp2->describe_table( $table );
		foreach ( array( 'ml_source_elementid', 'ml_elementid' ) as $candidate ) {
			if ( isset( $cols[ $candidate ] ) ) {
				return $candidate;
			}
		}
		return '';
	}
}