<?php
/**
 * Compares two snapshots and explains each difference where it can.
 *
 * @package OrbitstemSEO
 */

namespace Orbitstem\SEO\Monitor;

use Orbitstem\SEO\Diagnostics\Tables;

defined( 'ABSPATH' ) || exit;

/**
 * Each change: type, severity (error|warning|notice), object_id, url,
 * before, after, explained (bool) and explanation (text).
 *
 * A change is "explained" when, between the two snapshots, Orbitstem's change
 * log recorded a change to that item or to the site setting responsible, or
 * the post itself was edited. Anything else is "unexplained": the cue to look
 * at theme, plugin or server changes. Ranking movements are out of scope:
 * this compares what the site says, not what search engines do.
 */
class Diff {

	const LIMIT = 500;

	/**
	 * Snapshot store.
	 *
	 * @var Snapshotter
	 */
	private $snapshots;

	/**
	 * Constructor.
	 *
	 * @param Snapshotter $snapshots Snapshots.
	 */
	public function __construct( Snapshotter $snapshots ) {
		$this->snapshots = $snapshots;
	}

	/**
	 * Changes from snapshot A to snapshot B.
	 *
	 * @param int $a Older snapshot.
	 * @param int $b Newer snapshot.
	 * @return array{changes:array<int,array>,counts:array}
	 */
	public function compare( $a, $b ) {
		$sa = $this->snapshots->get( $a );
		$sb = $this->snapshots->get( $b );
		if ( ! $sa || ! $sb ) {
			return array(
				'changes' => array(),
				'counts'  => $this->count( array() ),
			);
		}
		// The window starts when the older snapshot finished: anything earlier is already in it.
		$from    = $sa['completed_at'] ? $sa['completed_at'] : $sa['created_at'];
		$to      = $sb['completed_at'] ? $sb['completed_at'] : $sb['created_at'];
		$changes = array_merge(
			$this->site_changes( (array) $sa['site'], (array) $sb['site'] ),
			$this->item_changes( $a, $b ),
			$this->rendered_changes( $a, $b )
		);
		$changes = $this->explain( $changes, $from, $to );
		return array(
			'changes' => $changes,
			'counts'  => $this->count( $changes ),
		);
	}

	/**
	 * Counts by severity, and unexplained by severity.
	 *
	 * @param array $changes Changes.
	 * @return array
	 */
	public function count( array $changes ) {
		$counts = array(
			'error'       => 0,
			'warning'     => 0,
			'notice'      => 0,
			'unexplained' => array(
				'error'   => 0,
				'warning' => 0,
				'notice'  => 0,
			),
		);
		foreach ( $changes as $c ) {
			++$counts[ $c['severity'] ];
			if ( ! $c['explained'] ) {
				++$counts['unexplained'][ $c['severity'] ];
			}
		}
		return $counts;
	}

	/**
	 * Site-wide changes.
	 *
	 * @param array $a Older site signals.
	 * @param array $b Newer site signals.
	 * @return array
	 */
	private function site_changes( array $a, array $b ) {
		$out = array();
		$get = function ( $arr, $key, $fallback = '' ) {
			return isset( $arr[ $key ] ) ? $arr[ $key ] : $fallback;
		};
		if ( 1 === (int) $get( $a, 'blog_public', 1 ) && 0 === (int) $get( $b, 'blog_public', 1 ) ) {
			$out[] = $this->change( 'site_visibility_off', 'error', 0, '', __( 'Search engines allowed', 'orbitstem-seo' ), __( 'Search engines discouraged (every page now noindex)', 'orbitstem-seo' ) );
		} elseif ( 0 === (int) $get( $a, 'blog_public', 1 ) && 1 === (int) $get( $b, 'blog_public', 1 ) ) {
			$out[] = $this->change( 'site_visibility_on', 'notice', 0, '', __( 'Search engines discouraged', 'orbitstem-seo' ), __( 'Search engines allowed', 'orbitstem-seo' ) );
		}
		if ( $get( $a, 'output_enabled', true ) && ! $get( $b, 'output_enabled', true ) ) {
			$out[] = $this->change( 'site_output_off', 'warning', 0, '', __( 'Orbitstem SEO tags printed', 'orbitstem-seo' ), __( 'Orbitstem SEO tags switched off (another SEO plugin active?)', 'orbitstem-seo' ) );
		}
		$new_plugins = array_diff( (array) $get( $b, 'seo_plugins', array() ), (array) $get( $a, 'seo_plugins', array() ) );
		if ( $new_plugins ) {
			$out[] = $this->change( 'site_seo_plugin', 'warning', 0, '', '', implode( ', ', $new_plugins ) );
		}
		if ( $get( $a, 'permalink_structure' ) !== $get( $b, 'permalink_structure' ) ) {
			$out[] = $this->change( 'site_permalinks', 'warning', 0, '', (string) $get( $a, 'permalink_structure' ), (string) $get( $b, 'permalink_structure' ) );
		}
		if ( $get( $a, 'sitemap_enabled', true ) && ! $get( $b, 'sitemap_enabled', true ) ) {
			$out[] = $this->change( 'site_sitemap_off', 'warning', 0, '', __( 'Sitemap on', 'orbitstem-seo' ), __( 'Sitemap off', 'orbitstem-seo' ) );
		}
		if ( $get( $a, 'theme' ) !== $get( $b, 'theme' ) ) {
			$out[] = $this->change( 'site_theme', 'notice', 0, '', (string) $get( $a, 'theme' ), (string) $get( $b, 'theme' ) );
		}
		if ( $get( $a, 'settings_hash' ) !== $get( $b, 'settings_hash' ) ) {
			$out[] = $this->change( 'site_settings', 'notice', 0, '', '', __( 'Orbitstem SEO settings were saved with different values', 'orbitstem-seo' ) );
		}
		if ( '' !== $get( $a, 'robots_txt_hash' ) && $get( $a, 'robots_txt_hash' ) !== $get( $b, 'robots_txt_hash' ) ) {
			$out[] = $this->change( 'site_robots_txt', 'notice', 0, home_url( '/robots.txt' ), '', __( 'robots.txt content changed', 'orbitstem-seo' ) );
		}
		return $out;
	}

	/**
	 * Changes to published items (matched by post ID, so a new slug is a change, not a removal).
	 *
	 * @param int $a Older snapshot.
	 * @param int $b Newer snapshot.
	 * @return array
	 */
	private function item_changes( $a, $b ) {
		global $wpdb;
		$items = Tables::snapshot_items();
		$out   = array();

		// phpcs:disable WordPress.DB.DirectDatabaseQuery
		$changed = $wpdb->get_results(
			$wpdb->prepare(
				'SELECT a.object_id, a.url AS a_url, b.url AS b_url, a.noindex AS a_noindex, b.noindex AS b_noindex, a.canonical AS a_canonical, b.canonical AS b_canonical,
					a.sitemap AS a_sitemap, b.sitemap AS b_sitemap, a.title AS a_title, b.title AS b_title, a.desc_hash AS a_desc, b.desc_hash AS b_desc, a.inbound AS a_inbound, b.inbound AS b_inbound
				FROM %i a INNER JOIN %i b ON b.snapshot_id = %d AND b.kind = 1 AND b.object_id = a.object_id
				WHERE a.snapshot_id = %d AND a.kind = 1 AND ( a.url <> b.url OR a.noindex <> b.noindex OR a.canonical <> b.canonical OR a.sitemap <> b.sitemap OR a.title <> b.title OR a.desc_hash <> b.desc_hash OR ( a.inbound > 0 AND b.inbound = 0 ) )
				LIMIT %d',
				$items,
				$items,
				$b,
				$a,
				self::LIMIT * 4
			),
			ARRAY_A
		);
		$removed = $wpdb->get_results(
			$wpdb->prepare(
				'SELECT a.object_id, a.url FROM %i a LEFT JOIN %i b ON b.snapshot_id = %d AND b.kind = 1 AND b.object_id = a.object_id WHERE a.snapshot_id = %d AND a.kind = 1 AND b.id IS NULL AND a.noindex = 0 LIMIT %d',
				$items,
				$items,
				$b,
				$a,
				self::LIMIT
			),
			ARRAY_A
		);
		$added   = (int) $wpdb->get_var(
			$wpdb->prepare(
				'SELECT COUNT(*) FROM %i b LEFT JOIN %i a ON a.snapshot_id = %d AND a.kind = 1 AND a.object_id = b.object_id WHERE b.snapshot_id = %d AND b.kind = 1 AND a.id IS NULL',
				$items,
				$items,
				$a,
				$b
			)
		);
		// phpcs:enable

		foreach ( $changed as $r ) {
			$id  = (int) $r['object_id'];
			$url = $r['b_url'];
			if ( $r['a_url'] !== $r['b_url'] ) {
				$out[] = $this->change( 'url_changed', 'notice', $id, $url, $r['a_url'], $r['b_url'] );
			}
			if ( ! (int) $r['a_noindex'] && (int) $r['b_noindex'] ) {
				$out[] = $this->change( 'noindex_on', 'error', $id, $url, 'index', 'noindex' );
			} elseif ( (int) $r['a_noindex'] && ! (int) $r['b_noindex'] ) {
				$out[] = $this->change( 'noindex_off', 'notice', $id, $url, 'noindex', 'index' );
			}
			if ( $r['a_canonical'] !== $r['b_canonical'] && $r['a_url'] === $r['b_url'] ) {
				$out[] = $this->change( 'canonical_changed', $r['b_canonical'] !== $r['b_url'] ? 'warning' : 'notice', $id, $url, $r['a_canonical'], $r['b_canonical'] );
			}
			if ( (int) $r['a_sitemap'] && ! (int) $r['b_sitemap'] && ! ( ! (int) $r['a_noindex'] && (int) $r['b_noindex'] ) ) {
				$out[] = $this->change( 'sitemap_out', 'warning', $id, $url, __( 'In sitemap', 'orbitstem-seo' ), __( 'Not in sitemap', 'orbitstem-seo' ) );
			}
			if ( $r['a_title'] !== $r['b_title'] ) {
				$out[] = $this->change( 'title_changed', 'notice', $id, $url, $r['a_title'], $r['b_title'] );
			}
			if ( $r['a_desc'] !== $r['b_desc'] ) {
				$out[] = $this->change( 'description_changed', 'notice', $id, $url, '', __( 'Meta description changed', 'orbitstem-seo' ) );
			}
			if ( (int) $r['a_inbound'] > 0 && 0 === (int) $r['b_inbound'] ) {
				/* translators: %d: number of posts */
				$out[] = $this->change( 'inbound_lost', 'notice', $id, $url, sprintf( _n( 'Linked from %d post', 'Linked from %d posts', (int) $r['a_inbound'], 'orbitstem-seo' ), (int) $r['a_inbound'] ), __( 'No longer linked from any post', 'orbitstem-seo' ) );
			}
		}
		foreach ( $removed as $r ) {
			$out[] = $this->change( 'removed', 'warning', (int) $r['object_id'], $r['url'], __( 'Published and indexable', 'orbitstem-seo' ), __( 'No longer published (deleted, trashed or unpublished)', 'orbitstem-seo' ) );
		}
		if ( $added ) {
			/* translators: %d: number of items */
			$out[] = $this->change( 'added', 'notice', 0, '', '', sprintf( _n( '%d new published item', '%d new published items', $added, 'orbitstem-seo' ), $added ), true );
		}
		return array_slice( $out, 0, self::LIMIT * 2 );
	}

	/**
	 * Changes in rendered sample pages (matched by URL).
	 *
	 * @param int $a Older snapshot.
	 * @param int $b Newer snapshot.
	 * @return array
	 */
	private function rendered_changes( $a, $b ) {
		global $wpdb;
		$items = Tables::snapshot_items();
		// phpcs:ignore WordPress.DB.DirectDatabaseQuery
		$rows = $wpdb->get_results(
			$wpdb->prepare(
				'SELECT b.object_id, b.url, a.status AS a_status, b.status AS b_status, a.noindex AS a_noindex, b.noindex AS b_noindex, a.canonical AS a_canonical, b.canonical AS b_canonical, a.title AS a_title, b.title AS b_title, a.duplicates AS a_dup, b.duplicates AS b_dup
				FROM %i a INNER JOIN %i b ON b.snapshot_id = %d AND b.kind = 2 AND b.url_hash = a.url_hash
				WHERE a.snapshot_id = %d AND a.kind = 2',
				$items,
				$items,
				$b,
				$a
			),
			ARRAY_A
		);
		$out  = array();
		foreach ( $rows as $r ) {
			$id = (int) $r['object_id'];
			if ( (int) $r['a_status'] !== (int) $r['b_status'] ) {
				$out[] = $this->change( 'page_status', 200 === (int) $r['a_status'] ? 'error' : 'warning', $id, $r['url'], 'HTTP ' . $r['a_status'], -1 === (int) $r['b_status'] ? __( 'No response', 'orbitstem-seo' ) : 'HTTP ' . $r['b_status'] );
			}
			if ( ! (int) $r['a_noindex'] && (int) $r['b_noindex'] ) {
				$out[] = $this->change( 'page_noindex_on', 'error', $id, $r['url'], 'index', 'noindex' );
			}
			// A noindex page prints no canonical: that is part of the noindex change reported above.
			if ( $r['a_canonical'] !== $r['b_canonical'] && ! ( (int) $r['b_noindex'] && '' === (string) $r['b_canonical'] ) ) {
				$out[] = $this->change( 'page_canonical', 'warning', $id, $r['url'], $r['a_canonical'], $r['b_canonical'] );
			}
			if ( $r['a_title'] !== $r['b_title'] ) {
				$out[] = $this->change( 'page_title', 'notice', $id, $r['url'], $r['a_title'], $r['b_title'] );
			}
			if ( ! (int) $r['a_dup'] && (int) $r['b_dup'] ) {
				$out[] = $this->change( 'page_duplicates', 'warning', $id, $r['url'], '', __( 'Duplicate SEO tags appeared', 'orbitstem-seo' ) );
			}
		}
		return $out;
	}

	/**
	 * Adds explanations from the change log and post edits between the snapshots.
	 *
	 * @param array  $changes Changes.
	 * @param string $from    Window start (GMT).
	 * @param string $to      Window end (GMT).
	 * @return array
	 */
	private function explain( array $changes, $from, $to ) {
		global $wpdb;
		$by_object = array();
		$site      = array();
		if ( Tables::v3_ready() ) {
			// phpcs:ignore WordPress.DB.DirectDatabaseQuery
			$log = $wpdb->get_results( $wpdb->prepare( 'SELECT id, fixer, object_type, object_id, summary, user_id, state_after FROM %i WHERE created_at BETWEEN %s AND %s ORDER BY id ASC', Tables::changes(), $from, $to ), ARRAY_A );
			foreach ( $log as $entry ) {
				$after = json_decode( (string) $entry['state_after'], true );
				$after = is_array( $after ) ? $after : array();
				$kind  = isset( $after['kind'] ) ? $after['kind'] : '';
				if ( 'post' === $entry['object_type'] ) {
					$by_object[ (int) $entry['object_id'] ][] = array(
						'summary' => $entry['summary'],
						'types'   => self::explains( $entry['fixer'], $kind, $after ),
					);
				} elseif ( in_array( $kind, array( 'visibility', 'settings' ), true ) ) {
					$site[ $kind ][] = $entry['summary'];
				}
			}
		}

		foreach ( $changes as &$c ) {
			if ( $c['explained'] ) {
				continue;
			}
			$reasons = array();
			// Only a recorded change of the matching kind explains it: a canonical
			// change does not explain a new noindex, for example.
			foreach ( $c['object_id'] && isset( $by_object[ $c['object_id'] ] ) ? $by_object[ $c['object_id'] ] : array() as $entry ) {
				if ( in_array( $c['type'], $entry['types'], true ) ) {
					$reasons[] = $entry['summary'];
				}
			}
			if ( ! $reasons && 'site_visibility_off' === $c['type'] && isset( $site['visibility'] ) ) {
				$reasons = $site['visibility'];
			}
			if ( ! $reasons && in_array( $c['type'], array( 'noindex_on', 'sitemap_out', 'site_settings', 'site_sitemap_off', 'page_noindex_on' ), true ) && isset( $site['settings'] ) ) {
				$reasons = $site['settings'];
			}
			if ( ! $reasons && $c['object_id'] ) {
				$post = get_post( $c['object_id'] );
				// An ordinary edit explains content changes (title, description, address), not
				// a new noindex, canonical or rendering problem: those settings are recorded
				// separately when changed in WordPress, so a theme or plugin update stays visible.
				$edit_explains = in_array( $c['type'], array( 'url_changed', 'title_changed', 'description_changed', 'page_title' ), true );
				if ( $edit_explains && $post && $post->post_modified_gmt > $from && $post->post_modified_gmt <= $to ) {
					$editor    = get_userdata( (int) get_post_meta( $post->ID, '_edit_last', true ) );
					$reasons[] = sprintf(
						/* translators: 1: date, 2: user name */
						__( 'Post edited on %1$s by %2$s', 'orbitstem-seo' ),
						get_date_from_gmt( $post->post_modified_gmt, get_option( 'date_format' ) . ' ' . get_option( 'time_format' ) ),
						$editor ? $editor->display_name : __( 'unknown user', 'orbitstem-seo' )
					);
				} elseif ( ! $post || 'publish' !== $post->post_status ) {
					if ( 'removed' === $c['type'] && $post ) {
						/* translators: %s: post status */
						$reasons[] = sprintf( __( 'Post status is now "%s"', 'orbitstem-seo' ), $post->post_status );
					}
				}
			}
			$c['explained']   = (bool) $reasons;
			$c['explanation'] = $reasons ? implode( '; ', array_unique( $reasons ) ) : __( 'Not explained by any change recorded here: check recent theme, plugin or server changes.', 'orbitstem-seo' );
		}
		return $changes;
	}

	/**
	 * Change types a change-log entry can explain.
	 *
	 * @param string $fixer Fixer id.
	 * @param string $kind  Guardian kind ('' for other entries).
	 * @param array  $after State after.
	 * @return string[]
	 */
	private static function explains( $fixer, $kind, array $after ) {
		$by_kind = array(
			'slug'           => array( 'url_changed', 'sitemap_out', 'page_status', 'canonical_changed', 'page_canonical' ),
			'post_noindex'   => array( 'noindex_on', 'noindex_off', 'page_noindex_on', 'sitemap_out' ),
			'post_canonical' => array( 'canonical_changed', 'page_canonical', 'sitemap_out' ),
			'trash'          => array( 'removed', 'page_status' ),
			'freed'          => array( 'page_status', 'url_changed' ),
		);
		if ( isset( $by_kind[ $kind ] ) ) {
			return $by_kind[ $kind ];
		}
		if ( 'csv_import' === $fixer ) {
			$by_field = array(
				'title'       => array( 'title_changed', 'page_title' ),
				'description' => array( 'description_changed' ),
				'noindex'     => $by_kind['post_noindex'],
				'canonical'   => $by_kind['post_canonical'],
			);
			$types    = array();
			foreach ( array_keys( $after ) as $field ) {
				if ( isset( $by_field[ $field ] ) ) {
					$types = array_merge( $types, $by_field[ $field ] );
				}
			}
			return $types;
		}
		// Link repairs and other content edits change text, not indexing.
		return array( 'title_changed', 'description_changed', 'page_title' );
	}

	/**
	 * One change record.
	 *
	 * @param string $type      Type.
	 * @param string $severity  Severity.
	 * @param int    $object_id Post ID or 0.
	 * @param string $url       URL or ''.
	 * @param string $before    Before.
	 * @param string $after     After.
	 * @param bool   $explained Pre-explained (informational).
	 * @return array
	 */
	private function change( $type, $severity, $object_id, $url, $before, $after, $explained = false ) {
		return array(
			'type'        => $type,
			'severity'    => $severity,
			'object_id'   => (int) $object_id,
			'url'         => (string) $url,
			'before'      => (string) $before,
			'after'       => (string) $after,
			'explained'   => $explained,
			'explanation' => $explained ? __( 'New content', 'orbitstem-seo' ) : '',
		);
	}

	/**
	 * Readable label for a change type.
	 *
	 * @param string $type Type.
	 * @return string
	 */
	public static function label( $type ) {
		$labels = array(
			'site_visibility_off' => __( 'Whole site hidden from search engines', 'orbitstem-seo' ),
			'site_visibility_on'  => __( 'Site visible to search engines again', 'orbitstem-seo' ),
			'site_output_off'     => __( 'Orbitstem SEO output switched off', 'orbitstem-seo' ),
			'site_seo_plugin'     => __( 'Another SEO plugin became active', 'orbitstem-seo' ),
			'site_permalinks'     => __( 'Permalink structure changed', 'orbitstem-seo' ),
			'site_sitemap_off'    => __( 'Sitemap switched off', 'orbitstem-seo' ),
			'site_theme'          => __( 'Theme changed', 'orbitstem-seo' ),
			'site_settings'       => __( 'Orbitstem SEO settings changed', 'orbitstem-seo' ),
			'site_robots_txt'     => __( 'robots.txt changed', 'orbitstem-seo' ),
			'url_changed'         => __( 'Address changed', 'orbitstem-seo' ),
			'noindex_on'          => __( 'Became noindex', 'orbitstem-seo' ),
			'noindex_off'         => __( 'No longer noindex', 'orbitstem-seo' ),
			'canonical_changed'   => __( 'Canonical changed', 'orbitstem-seo' ),
			'sitemap_out'         => __( 'Dropped out of the sitemap', 'orbitstem-seo' ),
			'title_changed'       => __( 'SEO title changed', 'orbitstem-seo' ),
			'description_changed' => __( 'Meta description changed', 'orbitstem-seo' ),
			'inbound_lost'        => __( 'Lost all internal links', 'orbitstem-seo' ),
			'removed'             => __( 'No longer published', 'orbitstem-seo' ),
			'added'               => __( 'New content', 'orbitstem-seo' ),
			'page_status'         => __( 'Page status changed (rendered)', 'orbitstem-seo' ),
			'page_noindex_on'     => __( 'Page renders noindex', 'orbitstem-seo' ),
			'page_canonical'      => __( 'Rendered canonical changed', 'orbitstem-seo' ),
			'page_title'          => __( 'Rendered title changed', 'orbitstem-seo' ),
			'page_duplicates'     => __( 'Duplicate tags appeared on page', 'orbitstem-seo' ),
		);
		return isset( $labels[ $type ] ) ? $labels[ $type ] : $type;
	}
}