<?php
/**
 * Update guard: checks key pages right after theme, plugin and core updates.
 *
 * @package OrbitstemSEO
 */

namespace Orbitstem\SEO\Monitor;

use Orbitstem\SEO\Core\Settings;
use Orbitstem\SEO\Diagnostics\UrlChecker;
use Orbitstem\SEO\Links\Url;
use Orbitstem\SEO\Repair\ChangeLog;

defined( 'ABSPATH' ) || exit;

/**
 * Keeps a baseline of what ten key pages tell search engines (status, title,
 * canonical, noindex, H1 headings, schema), refreshed daily. When a theme,
 * plugin or WordPress itself is updated, switched or (de)activated, the pages
 * are checked again about half a minute later and anything that got worse is
 * reported at once: an admin notice, an SEO issue and, if Monitor alerts are
 * on, an email. Pages edited since the baseline are skipped, because the edit
 * explains their changes. Every such site event also goes into the change
 * log, so Repair > History shows it.
 */
class UpdateGuard {

	const BASELINE      = 'orbitstem_seo_guard_baseline';
	const ALERT         = 'orbitstem_seo_guard_alert';
	const PENDING       = 'orbitstem_seo_guard_pending';
	const HOOK          = 'orbitstem_seo_update_guard';
	const BASELINE_HOOK = 'orbitstem_seo_guard_baseline_build';
	const LOG_ID        = 'site_event';
	const PAGES         = 10;

	/**
	 * Settings.
	 *
	 * @var Settings
	 */
	private $settings;

	/**
	 * Change log.
	 *
	 * @var ChangeLog
	 */
	private $log;

	/**
	 * Constructor.
	 *
	 * @param Settings  $settings Settings.
	 * @param ChangeLog $log      Change log.
	 */
	public function __construct( Settings $settings, ChangeLog $log ) {
		$this->settings = $settings;
		$this->log      = $log;
	}

	/**
	 * Registers hooks.
	 */
	public function register() {
		add_action( 'upgrader_process_complete', array( $this, 'on_upgrade' ), 20, 2 );
		add_action( 'activated_plugin', array( $this, 'on_plugin' ), 20 );
		add_action( 'deactivated_plugin', array( $this, 'on_plugin_off' ), 20 );
		add_action( 'switch_theme', array( $this, 'on_theme' ), 20 );
		add_action( self::HOOK, array( $this, 'check' ) );
		add_action( 'orbitstem_seo_daily', array( $this, 'refresh_baseline' ) );
		add_action( self::BASELINE_HOOK, array( $this, 'refresh_baseline' ) );
		add_action( 'admin_init', array( $this, 'ensure_baseline' ) );
		// Sites run only from WP-CLI or a server cron may never open wp-admin.
		if ( ( defined( 'WP_CLI' ) && WP_CLI ) || wp_doing_cron() ) {
			add_action( 'init', array( $this, 'ensure_baseline' ) );
		}
		add_action( 'update_option_' . Settings::OPTION, array( $this, 'on_settings' ), 10, 2 );
		add_action( 'admin_notices', array( $this, 'notice' ) );
		add_action( 'admin_post_orbitstem_seo_guard_dismiss', array( $this, 'dismiss' ) );
		// Just before a change: if no baseline exists yet (for example right after re-activation,
		// before WP-Cron ran), take it now, while the pages are still as they were.
		add_filter( 'upgrader_pre_install', array( $this, 'before_install' ), 1 );
		add_action( 'activate_plugin', array( $this, 'before_change' ), 1 );
		add_action( 'deactivate_plugin', array( $this, 'before_change' ), 1 );
	}

	/**
	 * Before files are replaced by an update or install.
	 *
	 * @param mixed $response Upgrader response (passed through).
	 * @return mixed
	 */
	public function before_install( $response ) {
		$this->before_change();
		return $response;
	}

	/**
	 * Builds the baseline right away when the guard is on and there is none.
	 */
	public function before_change() {
		if ( ! $this->settings->get( 'update_guard' ) || get_option( self::BASELINE ) ) {
			return;
		}
		$this->refresh_baseline();
		// A baseline scheduled earlier would now run after the change and save the changed pages as normal.
		wp_clear_scheduled_hook( self::BASELINE_HOOK );
	}

	/**
	 * Builds the baseline soon after activation, an upgrade or switching the
	 * guard on, so the first update afterwards is compared with a page known
	 * to be fine (not with the page as the update left it).
	 */
	public function ensure_baseline() {
		self::schedule_baseline( $this->settings );
	}

	/**
	 * Schedules a baseline build when the guard is on and there is none yet.
	 * Cheap enough for activation: it only schedules, the pages are fetched by
	 * WP-Cron a few seconds later.
	 *
	 * @param Settings $settings Settings.
	 */
	public static function schedule_baseline( Settings $settings ) {
		if ( $settings->get( 'update_guard' ) && ! get_option( self::BASELINE ) && ! wp_next_scheduled( self::BASELINE_HOOK ) ) {
			wp_schedule_single_event( time() + 5, self::BASELINE_HOOK );
		}
	}

	/**
	 * The guard was switched on: the old baseline may be weeks old.
	 *
	 * @param mixed $old_value Old settings.
	 * @param mixed $value     New settings.
	 */
	public function on_settings( $old_value, $value ) {
		$was = is_array( $old_value ) && ! empty( $old_value['update_guard'] );
		$is  = is_array( $value ) && ! empty( $value['update_guard'] );
		if ( $is && ! $was ) {
			delete_option( self::BASELINE );
			wp_clear_scheduled_hook( self::BASELINE_HOOK );
			wp_schedule_single_event( time() + 5, self::BASELINE_HOOK );
		}
	}

	/**
	 * Plugins, themes, translations or WordPress updated or installed.
	 *
	 * @param \WP_Upgrader $upgrader Upgrader.
	 * @param array        $extra    Details: type, action, plugins, themes.
	 */
	public function on_upgrade( $upgrader, $extra ) {
		$type   = isset( $extra['type'] ) ? (string) $extra['type'] : '';
		$action = isset( $extra['action'] ) ? (string) $extra['action'] : '';
		if ( 'translation' === $type || ! in_array( $action, array( 'update', 'install' ), true ) ) {
			return;
		}
		$names = array();
		if ( 'plugin' === $type ) {
			$list = isset( $extra['plugins'] ) ? (array) $extra['plugins'] : ( isset( $extra['plugin'] ) ? array( $extra['plugin'] ) : array() );
			foreach ( $list as $file ) {
				$names[] = $this->plugin_name( $file );
			}
		} elseif ( 'theme' === $type ) {
			$list = isset( $extra['themes'] ) ? (array) $extra['themes'] : ( isset( $extra['theme'] ) ? array( $extra['theme'] ) : array() );
			foreach ( $list as $slug ) {
				$theme   = wp_get_theme( $slug );
				$names[] = $theme->exists() ? $theme->get( 'Name' ) . ' ' . $theme->get( 'Version' ) : $slug;
			}
		} elseif ( 'core' === $type ) {
			$names[] = 'WordPress ' . get_bloginfo( 'version' );
		}
		$labels = array(
			/* translators: %s: plugin names */
			'plugin' => 'update' === $action ? __( 'Plugin updated: %s', 'orbitstem-seo' ) : __( 'Plugin installed: %s', 'orbitstem-seo' ),
			/* translators: %s: theme names */
			'theme'  => 'update' === $action ? __( 'Theme updated: %s', 'orbitstem-seo' ) : __( 'Theme installed: %s', 'orbitstem-seo' ),
			/* translators: %s: WordPress version */
			'core'   => __( 'WordPress updated: %s', 'orbitstem-seo' ),
		);
		if ( isset( $labels[ $type ] ) && $names ) {
			$this->event( sprintf( $labels[ $type ], implode( ', ', $names ) ) );
		}
	}

	/**
	 * A plugin was activated.
	 *
	 * @param string $file Plugin file.
	 */
	public function on_plugin( $file ) {
		if ( false !== strpos( (string) $file, 'orbitstem-seo/' ) ) {
			return;
		}
		/* translators: %s: plugin name */
		$this->event( sprintf( __( 'Plugin activated: %s', 'orbitstem-seo' ), $this->plugin_name( $file ) ) );
	}

	/**
	 * A plugin was deactivated.
	 *
	 * @param string $file Plugin file.
	 */
	public function on_plugin_off( $file ) {
		if ( false !== strpos( (string) $file, 'orbitstem-seo/' ) ) {
			return;
		}
		/* translators: %s: plugin name */
		$this->event( sprintf( __( 'Plugin deactivated: %s', 'orbitstem-seo' ), $this->plugin_name( $file ) ) );
	}

	/**
	 * The theme was switched.
	 *
	 * @param string $name New theme name.
	 */
	public function on_theme( $name ) {
		/* translators: %s: theme name */
		$this->event( sprintf( __( 'Theme switched to %s', 'orbitstem-seo' ), $name ) );
	}

	/**
	 * Records a site event and schedules a check.
	 *
	 * @param string $summary What happened.
	 */
	private function event( $summary ) {
		if ( \Orbitstem\SEO\Diagnostics\Tables::v3_ready() ) {
			$this->log->record( self::LOG_ID, 'site', 0, $summary, null, array( 'kind' => 'site_event' ) );
		}
		if ( ! $this->settings->get( 'update_guard' ) ) {
			return;
		}
		$pending   = (array) get_option( self::PENDING, array() );
		$pending[] = $summary;
		update_option( self::PENDING, array_slice( $pending, -20 ), false );
		if ( ! wp_next_scheduled( self::HOOK ) ) {
			wp_schedule_single_event( time() + 30, self::HOOK );
		}
	}

	/**
	 * Plugin name from its file.
	 *
	 * @param string $file Plugin file.
	 * @return string
	 */
	private function plugin_name( $file ) {
		$path = WP_PLUGIN_DIR . '/' . $file;
		if ( file_exists( $path ) && function_exists( 'get_plugin_data' ) ) {
			$data = get_plugin_data( $path, false, false );
			if ( ! empty( $data['Name'] ) ) {
				return $data['Name'] . ( ! empty( $data['Version'] ) ? ' ' . $data['Version'] : '' );
			}
		}
		return (string) $file;
	}

	/**
	 * The key pages: homepage, posts page, newest posts and pages.
	 *
	 * @return array<string,int> URL => post ID (0 for the homepage listing).
	 */
	public function key_pages() {
		$pages = array( home_url( '/' ) => (int) get_option( 'page_on_front' ) );
		$blog  = (int) get_option( 'page_for_posts' );
		if ( $blog ) {
			$pages[ (string) get_permalink( $blog ) ] = $blog;
		}
		foreach ( array( 'post', 'page' ) as $type ) {
			$ids = get_posts(
				array(
					'post_type'      => $type,
					'post_status'    => 'publish',
					'has_password'   => false,
					'posts_per_page' => 4,
					'orderby'        => 'post' === $type ? 'date' : 'modified',
					'fields'         => 'ids',
					'no_found_rows'  => true,
				)
			);
			foreach ( $ids as $id ) {
				$pages[ (string) get_permalink( $id ) ] = (int) $id;
			}
		}
		return array_slice( $pages, 0, self::PAGES, true );
	}

	/**
	 * What one page tells search engines right now.
	 *
	 * @param string $url Page URL.
	 * @return array{status:int,title:bool,canonical:string,noindex:bool,h1:int,schema:bool}
	 */
	public function signals( $url ) {
		$response = wp_remote_get( $url, UrlChecker::request_args( 20 ) );
		$status   = is_wp_error( $response ) ? 0 : (int) wp_remote_retrieve_response_code( $response );
		$html     = is_wp_error( $response ) ? '' : (string) wp_remote_retrieve_body( $response );
		$head     = false !== stripos( $html, '</head>' ) ? substr( $html, 0, stripos( $html, '</head>' ) ) : $html;
		$header   = is_wp_error( $response ) ? '' : strtolower( (string) wp_remote_retrieve_header( $response, 'x-robots-tag' ) );
		$canon    = preg_match( '#<link\b[^>]*\brel=["\']canonical["\'][^>]*\bhref=["\']([^"\']+)["\']#i', $head, $m ) ? Url::normalize( html_entity_decode( $m[1] ), $url ) : '';
		$robots   = preg_match_all( '#<meta\b[^>]*\bname=["\']robots["\'][^>]*\bcontent=["\']([^"\']*)["\']#i', $head, $r ) ? strtolower( implode( ',', $r[1] ) ) : '';
		return array(
			'status'    => $status,
			'title'     => (bool) preg_match( '#<title[^>]*>\s*[^<\s][^<]*</title>#i', $head ),
			'canonical' => $canon,
			'noindex'   => false !== strpos( $robots, 'noindex' ) || false !== strpos( $header, 'noindex' ),
			'h1'        => (int) preg_match_all( '#<h1[\s>]#i', $html ),
			'schema'    => false !== stripos( $html, 'application/ld+json' ),
		);
	}

	/**
	 * Saves the current signals as the new baseline.
	 *
	 * @return array Baseline.
	 */
	public function refresh_baseline() {
		$baseline = array(
			'time'  => time(),
			'pages' => array(),
		);
		foreach ( $this->key_pages() as $url => $id ) {
			$signals = $this->signals( $url );
			if ( 200 === $signals['status'] ) {
				$baseline['pages'][ $url ] = $signals + array( 'id' => $id );
			}
		}
		update_option( self::BASELINE, $baseline, false );
		return $baseline;
	}

	/**
	 * Compares the key pages with the baseline after a site event.
	 *
	 * @return array Problems found (also stored as the alert).
	 */
	public function check() {
		$pending  = (array) get_option( self::PENDING, array() );
		$baseline = get_option( self::BASELINE );
		delete_option( self::PENDING );
		if ( ! is_array( $baseline ) || empty( $baseline['pages'] ) ) {
			$this->refresh_baseline();
			return array();
		}
		$problems = array();
		foreach ( $baseline['pages'] as $url => $before ) {
			$id = isset( $before['id'] ) ? (int) $before['id'] : 0;
			if ( $id && strtotime( (string) get_post_field( 'post_modified_gmt', $id ) . ' UTC' ) > (int) $baseline['time'] ) {
				continue; // Edited since the baseline: the edit explains changes.
			}
			$now = $this->signals( $url );
			foreach ( $this->compare( $before, $now ) as $text ) {
				$problems[] = array(
					'url'  => $url,
					'text' => $text,
				);
			}
		}
		if ( $problems ) {
			update_option(
				self::ALERT,
				array(
					'time'     => time(),
					'events'   => $pending,
					'problems' => $problems,
				),
				false
			);
			$this->email( $pending, $problems );
		}
		$this->refresh_baseline();
		return $problems;
	}

	/**
	 * What got worse on one page, in plain words.
	 *
	 * @param array $before Baseline signals.
	 * @param array $now    Current signals.
	 * @return string[]
	 */
	public function compare( array $before, array $now ) {
		$out = array();
		if ( 200 !== $now['status'] ) {
			/* translators: %d: HTTP status */
			$out[] = 0 === $now['status'] ? __( 'The page no longer answers.', 'orbitstem-seo' ) : sprintf( __( 'The page now answers with HTTP %d instead of 200.', 'orbitstem-seo' ), $now['status'] );
			return $out;
		}
		if ( $before['title'] && ! $now['title'] ) {
			$out[] = __( 'The page lost its title tag.', 'orbitstem-seo' );
		}
		if ( ! $before['noindex'] && $now['noindex'] ) {
			$out[] = __( 'The page now tells search engines not to index it (noindex).', 'orbitstem-seo' );
		}
		if ( '' !== $before['canonical'] && $before['canonical'] !== $now['canonical'] ) {
			/* translators: %s: new canonical URL or "none" */
			$out[] = sprintf( __( 'The canonical URL changed to %s.', 'orbitstem-seo' ), '' !== $now['canonical'] ? $now['canonical'] : __( 'none', 'orbitstem-seo' ) );
		}
		if ( $before['h1'] <= 1 && $now['h1'] > 1 ) {
			/* translators: %d: number of H1 headings */
			$out[] = sprintf( __( 'The page now has %d H1 headings instead of one.', 'orbitstem-seo' ), $now['h1'] );
		} elseif ( 1 === $before['h1'] && 0 === $now['h1'] ) {
			$out[] = __( 'The page lost its H1 heading.', 'orbitstem-seo' );
		}
		if ( $before['schema'] && ! $now['schema'] ) {
			$out[] = __( 'The page lost its structured data (schema).', 'orbitstem-seo' );
		}
		return $out;
	}

	/**
	 * Emails the admin when Monitor alerts are on.
	 *
	 * @param string[] $events   What changed on the site.
	 * @param array    $problems Problems.
	 */
	private function email( array $events, array $problems ) {
		if ( ! $this->settings->get( 'monitor_email' ) ) {
			return;
		}
		$lines = array(
			/* translators: %s: site name */
			sprintf( __( 'Orbitstem SEO checked key pages of %s right after a change to the site, and something got worse.', 'orbitstem-seo' ), wp_specialchars_decode( get_bloginfo( 'name' ), ENT_QUOTES ) ),
			'',
			__( 'What changed on the site:', 'orbitstem-seo' ),
		);
		foreach ( $events as $event ) {
			$lines[] = '- ' . $event;
		}
		$lines[] = '';
		$lines[] = __( 'What got worse:', 'orbitstem-seo' );
		foreach ( $problems as $problem ) {
			$lines[] = '- ' . $problem['url'] . ': ' . $problem['text'];
		}
		$lines[] = '';
		$lines[] = admin_url( 'admin.php?page=orbitstem-seo-issues' );
		/* translators: %s: site name */
		wp_mail( (string) get_option( 'admin_email' ), sprintf( __( '[%s] A site update affected your SEO', 'orbitstem-seo' ), wp_specialchars_decode( get_bloginfo( 'name' ), ENT_QUOTES ) ), implode( "\n", $lines ) );
	}

	/**
	 * The current alert, if any.
	 *
	 * @return array|null
	 */
	public function alert() {
		$alert = get_option( self::ALERT );
		return is_array( $alert ) && ! empty( $alert['problems'] ) ? $alert : null;
	}

	/**
	 * The alert as an SEO issue.
	 *
	 * @return array<int,array>
	 */
	public function issues() {
		$alert = $this->alert();
		if ( ! $alert ) {
			return array();
		}
		$rows = array();
		foreach ( $alert['problems'] as $problem ) {
			$rows[] = array(
				'url'     => $problem['url'],
				'problem' => $problem['text'],
			);
		}
		return array(
			array(
				'id'          => 'update_guard',
				'severity'    => 'error',
				'title'       => __( 'A site update changed what pages tell search engines', 'orbitstem-seo' ),
				/* translators: %s: list of updates */
				'explanation' => sprintf( __( 'Right after this change to the site, key pages got worse: %s', 'orbitstem-seo' ), implode( '; ', (array) $alert['events'] ) ),
				'fix'         => __( 'Check the theme or plugin that was just updated (its settings, or roll back to the previous version), then dismiss this alert on the dashboard notice.', 'orbitstem-seo' ),
				'count'       => count( $rows ),
				'columns'     => array(
					'url'     => __( 'Page', 'orbitstem-seo' ),
					'problem' => __( 'What got worse', 'orbitstem-seo' ),
				),
				'rows'        => $rows,
			),
		);
	}

	/**
	 * Admin notice while an alert is open.
	 */
	public function notice() {
		$alert = $this->alert();
		if ( ! $alert || ! current_user_can( 'manage_options' ) ) {
			return;
		}
		$dismiss = wp_nonce_url( admin_url( 'admin-post.php?action=orbitstem_seo_guard_dismiss' ), 'orbitstem_seo_guard_dismiss' );
		printf(
			'<div class="notice notice-error"><p><strong>%1$s</strong> %2$s <a href="%3$s">%4$s</a> · <a href="%5$s">%6$s</a></p></div>',
			esc_html__( 'Orbitstem SEO update guard:', 'orbitstem-seo' ),
			/* translators: %d: number of problems */
			esc_html( sprintf( _n( 'right after a site update, %d problem appeared on key pages.', 'right after a site update, %d problems appeared on key pages.', count( $alert['problems'] ), 'orbitstem-seo' ), count( $alert['problems'] ) ) ),
			esc_url( admin_url( 'admin.php?page=orbitstem-seo-issues#update_guard' ) ),
			esc_html__( 'See what changed', 'orbitstem-seo' ),
			esc_url( $dismiss ),
			esc_html__( 'Dismiss', 'orbitstem-seo' )
		);
	}

	/**
	 * Dismisses the alert.
	 */
	public function dismiss() {
		if ( ! current_user_can( 'manage_options' ) ) {
			wp_die( esc_html__( 'You do not have permission to do this.', 'orbitstem-seo' ), 403 );
		}
		check_admin_referer( 'orbitstem_seo_guard_dismiss' );
		delete_option( self::ALERT );
		wp_safe_redirect( wp_get_referer() ? wp_get_referer() : admin_url( 'admin.php?page=orbitstem-seo' ) );
		exit;
	}
}