<?php
/**
 * Which database tables belong to this install's migration.
 *
 * @package Shiftora\Migration
 */

declare(strict_types=1);

namespace Shiftora\Migration;

// Prevent direct file access.
if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * Pure name logic, so the exporter, the engine and the tests agree on one rule.
 *
 * Left out: tables outside the prefix, Shiftora's own tables (including staging
 * copies), Action Scheduler tables, and every table of a second WordPress install
 * that sits under the prefix (Plugin Check's runtime install creates wp_pc_*).
 * Moving those would copy install-bound state, and a staging table copied into
 * its own staging table doubles the space the destination needs.
 */
final class TableEligibility {

	public const REASON_FOREIGN_PREFIX   = 'outside this site\'s prefix';
	public const REASON_SHIFTORA         = 'Shiftora table';
	public const REASON_ACTION_SCHEDULER = 'Action Scheduler table';
	public const REASON_NESTED_INSTALL   = 'belongs to another WordPress install';

	/**
	 * Splits full table names into this install's tables and the rest.
	 *
	 * @param string[] $tables Full table names.
	 *
	 * @return array{ kept: string[], left_out: array<string, string> } Left out maps name to reason.
	 */
	public static function partition( array $tables, string $prefix ): array {
		$tables   = array_values( array_map( 'strval', $tables ) );
		$nested   = self::nested_prefixes( $tables, $prefix );
		$kept     = [];
		$left_out = [];

		foreach ( $tables as $table ) {
			$reason = self::reason( $table, $prefix, $nested );
			if ( null === $reason ) {
				$kept[] = $table;
			} else {
				$left_out[ $table ] = $reason;
			}
		}

		return [ 'kept' => $kept, 'left_out' => $left_out ];
	}

	/**
	 * Screens a table list the remote site advertised, or a selection for a pull.
	 *
	 * Names may be full or bare. With no known remote prefix only the name rules
	 * that need no prefix apply: bare shiftora_ and Action Scheduler names.
	 *
	 * @param string[] $names Remote table names.
	 *
	 * @return array{ kept: string[], left_out: array<string, string> }
	 */
	public static function screen_remote( array $names, string $remote_prefix ): array {
		$names = array_values( array_map( 'strval', $names ) );

		if ( '' === $remote_prefix ) {
			$kept     = [];
			$left_out = [];
			foreach ( $names as $name ) {
				if ( str_starts_with( $name, 'shiftora_' ) ) {
					$left_out[ $name ] = self::REASON_SHIFTORA;
				} elseif ( self::is_action_scheduler_table( $name ) ) {
					$left_out[ $name ] = self::REASON_ACTION_SCHEDULER;
				} else {
					$kept[] = $name;
				}
			}

			return [ 'kept' => $kept, 'left_out' => $left_out ];
		}

		$full_of = static fn( string $name ): string => str_starts_with( $name, $remote_prefix ) ? $name : $remote_prefix . $name;
		$fulls   = array_map( $full_of, $names );
		$nested  = self::nested_prefixes( $fulls, $remote_prefix );
		$kept     = [];
		$left_out = [];

		foreach ( $names as $i => $name ) {
			$reason = self::reason( $fulls[ $i ], $remote_prefix, $nested );
			if ( null === $reason ) {
				$kept[] = $name;
			} else {
				$left_out[ $name ] = $reason;
			}
		}

		return [ 'kept' => $kept, 'left_out' => $left_out ];
	}

	/**
	 * One line for the migration log: how many tables were left out, and which.
	 *
	 * @param array<string, string> $left_out Name to reason.
	 */
	public static function summarise( array $left_out ): string {
		$count = count( $left_out );

		return sprintf(
			'Left out %d %s that do not belong to this site\'s data: %s.',
			$count,
			1 === $count ? 'table' : 'tables',
			implode( ', ', array_keys( $left_out ) )
		);
	}

	/**
	 * Why a table is not eligible, or null when it is.
	 *
	 * @param string[] $nested Longer prefixes that hold a second install.
	 */
	private static function reason( string $table, string $prefix, array $nested ): ?string {
		if ( ! str_starts_with( $table, $prefix ) ) {
			return self::REASON_FOREIGN_PREFIX;
		}

		if ( str_starts_with( substr( $table, strlen( $prefix ) ), 'shiftora_' ) ) {
			return self::REASON_SHIFTORA;
		}

		if ( self::is_action_scheduler_table( $table ) ) {
			return self::REASON_ACTION_SCHEDULER;
		}

		foreach ( $nested as $inner ) {
			if ( str_starts_with( $table, $inner ) ) {
				return self::REASON_NESTED_INSTALL;
			}
		}

		return null;
	}

	/**
	 * Prefixes under this one that carry a full second install: both options and
	 * posts exist there. A purely numeric segment is a multisite blog (wp_2_), not
	 * another install, and a plugin's tables (wp_wc_orders) never have both.
	 *
	 * @param string[] $tables Full names.
	 *
	 * @return string[] Full prefixes, such as wp_pc_.
	 */
	private static function nested_prefixes( array $tables, string $prefix ): array {
		$present = array_flip( $tables );
		$nested  = [];

		foreach ( $tables as $table ) {
			if ( ! str_starts_with( $table, $prefix ) || ! str_ends_with( $table, 'options' ) ) {
				continue;
			}

			$candidate = substr( $table, 0, -strlen( 'options' ) );
			$segments  = substr( $candidate, strlen( $prefix ) );
			if ( 1 !== preg_match( '/^(?:[^_]+_)+$/', $segments ) || 1 === preg_match( '/^(?:\d+_)+$/', $segments ) ) {
				continue;
			}

			if ( isset( $present[ $candidate . 'posts' ] ) ) {
				$nested[] = $candidate;
			}
		}

		return $nested;
	}

	/** Action Scheduler names its tables with this segment under any prefix. */
	private static function is_action_scheduler_table( string $table ): bool {
		$table = strtolower( $table );

		return str_contains( $table, 'action_scheduler_' ) || str_contains( $table, 'actionscheduler_' );
	}
}