<?php
/**
 * Autoloaded options grouped by owner.
 *
 * @package Bloatwright
 */

namespace Bloatwright\Services;

use Bloatwright\Profiler\Owner;

/**
 * AutoloadAnalyzer.
 */
class AutoloadAnalyzer {

	/**
	 * Core option names that belong to WordPress itself.
	 */
	const CORE = array( 'siteurl', 'home', 'blogname', 'blogdescription', 'users_can_register', 'admin_email', 'start_of_week', 'use_balanceTags', 'use_smilies', 'require_name_email', 'comments_notify', 'posts_per_rss', 'rss_use_excerpt', 'mailserver_url', 'mailserver_login', 'mailserver_pass', 'mailserver_port', 'default_category', 'default_comment_status', 'default_ping_status', 'default_pingback_flag', 'posts_per_page', 'date_format', 'time_format', 'links_updated_date_format', 'comment_moderation', 'moderation_notify', 'permalink_structure', 'rewrite_rules', 'hack_file', 'blog_charset', 'moderation_keys', 'active_plugins', 'category_base', 'ping_sites', 'comment_max_links', 'gmt_offset', 'default_email_category', 'recently_edited', 'template', 'stylesheet', 'comment_registration', 'html_type', 'use_trackback', 'default_role', 'db_version', 'uploads_use_yearmonth_folders', 'upload_path', 'blog_public', 'default_link_category', 'show_on_front', 'tag_base', 'show_avatars', 'avatar_rating', 'upload_url_path', 'thumbnail_size_w', 'thumbnail_size_h', 'thumbnail_crop', 'medium_size_w', 'medium_size_h', 'avatar_default', 'large_size_w', 'large_size_h', 'image_default_link_type', 'image_default_size', 'image_default_align', 'close_comments_for_old_posts', 'close_comments_days_old', 'thread_comments', 'thread_comments_depth', 'page_comments', 'comments_per_page', 'default_comments_page', 'comment_order', 'sticky_posts', 'widget_categories', 'widget_text', 'widget_rss', 'uninstall_plugins', 'timezone_string', 'page_for_posts', 'page_on_front', 'default_post_format', 'link_manager_enabled', 'finished_splitting_shared_terms', 'site_icon', 'medium_large_size_w', 'medium_large_size_h', 'wp_page_for_privacy_policy', 'show_comments_cookies_opt_in', 'admin_email_lifespan', 'disallowed_keys', 'comment_previously_approved', 'auto_plugin_theme_update_emails', 'auto_update_core_dev', 'auto_update_core_minor', 'auto_update_core_major', 'wp_force_deactivated_plugins', 'wp_attachment_pages_enabled', 'initial_db_version', 'fresh_site', 'WPLANG', 'cron', 'sidebars_widgets', 'nonce_key', 'nonce_salt', 'wp_user_roles', 'current_theme', 'theme_mods_', 'recovery_keys', 'https_detection_errors', 'can_compress_scripts', 'db_upgraded', 'blog_upload_space', 'user_count', 'widget_block', 'widget_', 'theme_switched', 'auto_core_update_notified', 'wp_calendar_block_has_published_posts', 'wp_force_deactivated_plugins', 'https_migration_required', 'core_updater.lock', 'doing_cron', 'update_core', 'update_plugins', 'update_themes', 'theme_roots', 'dirsize_cache', 'wp_core_block_css_files', 'global_styles', 'health-check-site-status-result', 'is_multi_author', 'wp_theme_files_patterns-' );

	/**
	 * Analyse.
	 *
	 * @param int $top_rows Rows kept per owner.
	 * @return array{owners: array<string, array{bytes:int,count:int,rows:array}>, descriptors: array, unknown: array, total_bytes:int, total_count:int}
	 */
	public function analyze( int $top_rows = 10 ): array {
		global $wpdb;
		$rows = $wpdb->get_results( $wpdb->prepare( 'SELECT option_name, LENGTH(option_value) AS bytes FROM %i WHERE autoload IN (%s, %s, %s, %s) ORDER BY bytes DESC', $wpdb->options, 'yes', 'on', 'auto-on', 'auto' ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery -- analysis query.
		if ( ! function_exists( 'get_plugins' ) ) {
			require_once ABSPATH . 'wp-admin/includes/plugin.php'; // Lists installed but inactive plugins as owners too.
		}
		$overrides   = (array) get_option( 'bloatwright_overrides', array() );
		$owner       = new Owner( $overrides );
		$plugins     = array();
		$descriptors = array();
		foreach ( $owner->all() as $key => $d ) {
			if ( in_array( $d['type'], array( 'plugin', 'inactive', 'theme' ), true ) ) {
				$plugins[ Aggregator::slug( $d ) ] = $key;
			}
		}
		$prefixes = self::prefix_overrides( $overrides, array_keys( $owner->all() ) );
		$owners   = array();
		$total    = 0;
		$count    = 0;
		$unknown  = array();
		foreach ( (array) $rows as $r ) {
			$name   = (string) $r['option_name'];
			$bytes  = (int) $r['bytes'];
			$total += $bytes;
			++$count;
			$key = $this->owner_for_option( $name, $plugins, $prefixes );
			if ( Owner::UNKNOWN === $key ) {
				if ( count( $unknown ) < 20 ) {
					$unknown[] = array(
						'name'  => $name,
						'bytes' => $bytes,
					);
				}
			}
			if ( ! isset( $owners[ $key ] ) ) {
				$owners[ $key ]      = array(
					'bytes' => 0,
					'count' => 0,
					'rows'  => array(),
				);
				$descriptors[ $key ] = $owner->describe( $key );
			}
			$owners[ $key ]['bytes'] += $bytes;
			++$owners[ $key ]['count'];
			if ( count( $owners[ $key ]['rows'] ) < $top_rows ) {
				$owners[ $key ]['rows'][] = array(
					'name'  => $name,
					'bytes' => $bytes,
				);
			}
		}
		uasort(
			$owners,
			static function ( $a, $b ) {
				return $b['bytes'] <=> $a['bytes'];
			}
		);
		return array(
			'owners'      => $owners,
			'descriptors' => $descriptors,
			'unknown'     => $unknown,
			'total_bytes' => $total,
			'total_count' => $count,
		);
	}

	/**
	 * Attribution overrides whose key is an option-name prefix rather than a directory (no slash or
	 * backslash, not a Windows drive path), longest first: prefix => owner key.
	 *
	 * @param array    $overrides Stored overrides (directory or option prefix => owner key).
	 * @param string[] $known Owner keys that exist.
	 * @return array<string,string>
	 */
	public static function prefix_overrides( array $overrides, array $known ): array {
		$out = array();
		foreach ( $overrides as $prefix => $key ) {
			$prefix = trim( (string) $prefix );
			if ( '' === $prefix || preg_match( '#[/\\\\]|^[A-Za-z]:#', $prefix ) || ! in_array( (string) $key, $known, true ) ) {
				continue;
			}
			$out[ $prefix ] = (string) $key;
		}
		uksort(
			$out,
			static function ( $a, $b ) {
				return strlen( (string) $b ) <=> strlen( (string) $a );
			}
		);
		return $out;
	}

	/**
	 * Owner key for an option name. Option-prefix overrides from Settings win over everything else.
	 *
	 * @param string               $name Option.
	 * @param array<string,string> $plugins slug => owner key (installed plugins/themes).
	 * @param array<string,string> $prefixes Option-prefix overrides (prefix => owner key), longest first.
	 */
	public function owner_for_option( string $name, array $plugins, array $prefixes = array() ): string {
		$bare = $name;
		foreach ( array( '_transient_timeout_', '_transient_', '_site_transient_timeout_', '_site_transient_' ) as $t ) {
			if ( 0 === strpos( $bare, $t ) ) {
				$bare = substr( $bare, strlen( $t ) );
				break;
			}
		}
		foreach ( $prefixes as $prefix => $key ) {
			if ( 0 === strpos( $bare, (string) $prefix ) || 0 === strpos( $name, (string) $prefix ) ) {
				return $key;
			}
		}
		if ( in_array( $bare, self::CORE, true ) || 0 === strpos( $bare, 'theme_mods_' ) || 0 === strpos( $bare, 'widget_' ) || 0 === strpos( $bare, '_wp_' ) || ( 0 === strpos( $bare, 'wp_' ) && ! isset( $plugins['wp'] ) && $this->is_core_wp_option( $bare ) ) ) {
			if ( 0 === strpos( $bare, 'theme_mods_' ) ) {
				$sheet = substr( $bare, strlen( 'theme_mods_' ) );
				if ( isset( $plugins[ $sheet ] ) ) {
					return $plugins[ $sheet ];
				}
			}
			return Owner::CORE;
		}
		$slug = PrefixRegistry::match( $name );
		if ( '' !== $slug ) {
			if ( 'core' === $slug ) {
				return Owner::CORE;
			}
			if ( isset( $plugins[ $slug ] ) ) {
				return $plugins[ $slug ];
			}
			return 'plugin:' . $slug; // Known plugin, currently inactive/uninstalled.
		}
		// Fuzzy: compare the option's leading tokens with each installed slug's tokens. The longest
		// match wins; on a tie a slug matched in full beats one matched only by its first word, so
		// example_reviews_cache belongs to example-reviews, not to example-slider. Two slugs that
		// match equally well, and neither in full, make the option ambiguous: it stays unknown.
		$tokens = array_values( array_filter( (array) preg_split( '/[_\-]+/', strtolower( $bare ) ) ) );
		if ( $tokens ) {
			$best      = '';
			$best_len  = 0;
			$best_full = false;
			$tied      = false;
			foreach ( $plugins as $slug => $key ) {
				list( $len, $full ) = self::slug_match( $tokens, (string) $slug );
				if ( $len <= 0 ) {
					continue;
				}
				if ( $len > $best_len || ( $len === $best_len && $full && ! $best_full ) ) {
					$best      = $key;
					$best_len  = $len;
					$best_full = $full;
					$tied      = false;
				} elseif ( $len === $best_len && $full === $best_full && $key !== $best ) {
					$tied = true;
				}
			}
			if ( '' !== $best && ! ( $tied && ! $best_full ) ) {
				return $best;
			}
		}
		return Owner::UNKNOWN;
	}

	/**
	 * How well an option name's tokens match a plugin or theme slug.
	 *
	 * Returns the number of matched characters and whether the whole slug was matched: the option's
	 * leading tokens repeat the slug's tokens (heavy_init_x for heavy-init; only heavy_x matches it
	 * partly), its first token is the slug without separators (heavyinit_x) or the start of it
	 * (woocom_x for woocommerce, at least four characters, partly), or the slug's initials
	 * (acf_x for advanced-custom-fields). The option's first token needs at least three characters.
	 *
	 * @param string[] $tokens Lower-case tokens of the option name.
	 * @param string   $slug Plugin or theme directory slug.
	 * @return array{0:int,1:bool}
	 */
	private static function slug_match( array $tokens, string $slug ): array {
		$slug_tokens = array_values( array_filter( (array) preg_split( '/[_\-]+/', strtolower( $slug ) ) ) );
		$first       = (string) ( $tokens[0] ?? '' );
		if ( ! $slug_tokens || strlen( $first ) < 3 ) {
			return array( 0, false );
		}
		$len = 0;
		$i   = 0;
		while ( isset( $tokens[ $i ], $slug_tokens[ $i ] ) && $tokens[ $i ] === $slug_tokens[ $i ] ) {
			$len += strlen( (string) $tokens[ $i ] );
			++$i;
		}
		if ( $len > 0 ) {
			return array( $len, count( $slug_tokens ) === $i );
		}
		$compact = implode( '', $slug_tokens );
		if ( $first === $compact ) {
			return array( strlen( $first ), true );
		}
		if ( strlen( $first ) >= 4 && 0 === strpos( $compact, $first ) ) {
			return array( strlen( $first ), false );
		}
		if ( count( $slug_tokens ) > 1 ) {
			$initials = '';
			foreach ( $slug_tokens as $t ) {
				$initials .= $t[0];
			}
			if ( $first === $initials ) {
				return array( strlen( $first ), true );
			}
		}
		return array( 0, false );
	}

	/**
	 * Core options that start with wp_ (the rest of that prefix is up for grabs by plugins).
	 *
	 * @param string $name Option.
	 */
	private function is_core_wp_option( string $name ): bool {
		return in_array( $name, array( 'wp_user_roles', 'wp_page_for_privacy_policy', 'wp_force_deactivated_plugins', 'wp_attachment_pages_enabled', 'wp_calendar_block_has_published_posts' ), true );
	}
}