<?php
/**
 * Writes extracted links to the links/instances tables.
 *
 * @package Unbreak
 */

namespace Unbreak\Extract;

use Unbreak\Install\Schema;
use Unbreak\Support\Clock;
use Unbreak\Support\Settings;
use Unbreak\Support\Url;

defined( 'ABSPATH' ) || exit;

/**
 * Re-indexes one source at a time: replaces its instances, upserts links, removes orphaned links.
 */
final class Indexer {

	/**
	 * Post statuses that are scanned.
	 */
	public const POST_STATUSES = array( 'publish', 'private' );

	/**
	 * Rows per multi-row INSERT / IN() list.
	 */
	private const CHUNK = 200;

	/**
	 * Post types whose content is scanned. Filterable.
	 *
	 * @return string[]
	 */
	public static function post_types(): array {
		$types = get_post_types( array( 'public' => true ), 'names' );
		unset( $types['attachment'] );
		/**
		 * Filters the post types Unbreak scans.
		 *
		 * @param string[] $types Post type names.
		 */
		$types = apply_filters( 'unbreak_post_types', array_values( $types ) );
		return array_values( array_filter( array_map( 'strval', (array) $types ) ) );
	}

	/**
	 * Whether a post should be scanned.
	 *
	 * @param \WP_Post $post Post.
	 */
	public static function is_eligible( \WP_Post $post ): bool {
		return in_array( $post->post_status, self::POST_STATUSES, true )
			&& in_array( $post->post_type, self::post_types(), true )
			&& ! wp_is_post_revision( $post )
			&& ! wp_is_post_autosave( $post );
	}

	/**
	 * Re-index a post (post content + Elementor data). Removes its instances if it is no longer eligible.
	 *
	 * @param int $post_id Post ID.
	 * @return int Number of instances stored.
	 */
	public function index_post( int $post_id ): int {
		$post    = get_post( $post_id );
		$entries = array();

		if ( $post instanceof \WP_Post && self::is_eligible( $post ) ) {
			$base   = (string) get_permalink( $post );
			$base   = '' !== $base ? $base : home_url( '/' );
			$images = (bool) Settings::get( 'include_images' );

			$elementor_enabled = Settings::source_enabled( ElementorExtractor::SOURCE_TYPE );
			$built_with_el     = $elementor_enabled && ElementorExtractor::is_built_with_elementor( $post_id );

			// For Elementor pages, post_content is only a rendered copy of the Elementor data.
			if ( Settings::source_enabled( PostContentExtractor::SOURCE_TYPE ) && ! $built_with_el ) {
				$entries[] = array(
					'source_type' => PostContentExtractor::SOURCE_TYPE,
					'source_id'   => $post_id,
					'links'       => PostContentExtractor::extract( (string) $post->post_content, $base, $images ),
				);
			}
			if ( $elementor_enabled ) {
				$entries[] = array(
					'source_type' => ElementorExtractor::SOURCE_TYPE,
					'source_id'   => $post_id,
					'links'       => ElementorExtractor::extract_post( $post_id, $base, $images ),
				);
			}
		}

		return $this->replace(
			array( PostContentExtractor::SOURCE_TYPE, ElementorExtractor::SOURCE_TYPE ),
			$post_id,
			$entries
		);
	}

	/**
	 * Re-index every custom menu link.
	 *
	 * @return int Number of instances stored.
	 */
	public function index_menus(): int {
		$entries = array();
		if ( Settings::source_enabled( MenuExtractor::SOURCE_TYPE ) ) {
			foreach ( MenuExtractor::extract_all() as $item_id => $links ) {
				$entries[] = array(
					'source_type' => MenuExtractor::SOURCE_TYPE,
					'source_id'   => $item_id,
					'links'       => $links,
				);
			}
		}
		return $this->replace( array( MenuExtractor::SOURCE_TYPE ), null, $entries );
	}

	/**
	 * Re-index every active widget.
	 *
	 * @return int Number of instances stored.
	 */
	public function index_widgets(): int {
		$entries = array();
		if ( Settings::source_enabled( WidgetExtractor::SOURCE_TYPE ) ) {
			foreach ( WidgetExtractor::extract_all( (bool) Settings::get( 'include_images' ) ) as $widget ) {
				$entries[] = array(
					'source_type' => WidgetExtractor::SOURCE_TYPE,
					'source_id'   => $widget['source_id'],
					'links'       => $widget['links'],
				);
			}
		}
		return $this->replace( array( WidgetExtractor::SOURCE_TYPE ), null, $entries );
	}

	/**
	 * Re-index one comment.
	 *
	 * @param int $comment_id Comment ID.
	 * @return int Number of instances stored.
	 */
	public function index_comment( int $comment_id ): int {
		$entries = array();
		if ( Settings::source_enabled( CommentExtractor::SOURCE_TYPE ) ) {
			$links = CommentExtractor::extract( $comment_id, (bool) Settings::get( 'include_images' ) );
			if ( null !== $links ) {
				$entries[] = array(
					'source_type' => CommentExtractor::SOURCE_TYPE,
					'source_id'   => $comment_id,
					'links'       => $links,
				);
			}
		}
		return $this->replace( array( CommentExtractor::SOURCE_TYPE ), $comment_id, $entries );
	}

	/**
	 * Remove all instances of a source type (used when a source is disabled in settings).
	 *
	 * @param string $source_type Source type.
	 */
	public function remove_source_type( string $source_type ): void {
		$this->replace( array( $source_type ), null, array() );
	}

	/**
	 * Replace the instances in a scope with freshly extracted ones.
	 *
	 * @param string[]                                                                                                                                                 $types     Source types in scope.
	 * @param int|null                                                                                                                                                 $source_id Source ID in scope, or null for every ID of those types.
	 * @param array<int,array{source_type:string,source_id:int,links:array<int,array{url:string,element:string,anchor_text:string,context:string,source_key:string}>}> $entries   New instances.
	 * @return int Number of instances stored.
	 */
	private function replace( array $types, ?int $source_id, array $entries ): int {
		global $wpdb;
		$instances = Schema::table( 'instances' );

		$type_in = implode( ',', array_fill( 0, count( $types ), '%s' ) );
		// "( %d = 0 OR source_id = %d )": the first value is 0 when the scope covers every ID of these types.
		$params = array_merge( array( $instances ), $types, array( null === $source_id ? 0 : 1, (int) $source_id ) );

		// phpcs:disable WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare,WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber -- $type_in only holds %s placeholders.
		$old_ids = array_map( 'intval', (array) $wpdb->get_col( $wpdb->prepare( "SELECT DISTINCT link_id FROM %i WHERE source_type IN ({$type_in}) AND ( %d = 0 OR source_id = %d )", $params ) ) );
		$wpdb->query( $wpdb->prepare( "DELETE FROM %i WHERE source_type IN ({$type_in}) AND ( %d = 0 OR source_id = %d )", $params ) );
		// phpcs:enable

		// Admin/auth URLs of this site (dashboard, login, nonce links) are never stored or checked.
		foreach ( $entries as $i => $entry ) {
			$entries[ $i ]['links'] = array_values(
				array_filter(
					$entry['links'],
					static function ( array $link ): bool {
						return ! Url::is_wp_admin( $link['url'] );
					}
				)
			);
		}

		$urls = array();
		foreach ( $entries as $entry ) {
			foreach ( $entry['links'] as $link ) {
				$urls[ $link['url'] ] = true;
			}
		}
		$ids = $this->ensure_links( array_keys( $urls ) );

		$rows = array();
		foreach ( $entries as $entry ) {
			foreach ( $entry['links'] as $link ) {
				if ( ! isset( $ids[ $link['url'] ] ) ) {
					continue;
				}
				$rows[] = array(
					$ids[ $link['url'] ],
					$entry['source_type'],
					$entry['source_id'],
					'' !== $link['source_key'] ? $link['source_key'] : null,
					'' !== $link['anchor_text'] ? $link['anchor_text'] : null,
					$link['element'],
					'' !== $link['context'] ? $link['context'] : null,
				);
			}
		}
		$this->insert_instances( $rows );

		$removed = array_diff( $old_ids, array_values( $ids ) );
		if ( ! empty( $removed ) ) {
			$this->delete_orphans( $removed );
		}

		return count( $rows );
	}

	/**
	 * Make sure a link row exists for each URL; returns url => id.
	 *
	 * @param string[] $urls Normalized URLs.
	 * @return array<string,int>
	 */
	public function ensure_links( array $urls ): array {
		global $wpdb;
		$table = Schema::table( 'links' );
		$ids   = array();
		if ( empty( $urls ) ) {
			return $ids;
		}

		$by_hash = array();
		foreach ( $urls as $url ) {
			$by_hash[ Url::hash( $url ) ] = $url;
		}

		foreach ( array_chunk( array_keys( $by_hash ), self::CHUNK ) as $chunk ) {
			$ids += $this->lookup_ids( $chunk, $by_hash );

			$missing = array();
			foreach ( $chunk as $hash ) {
				if ( ! isset( $ids[ $by_hash[ $hash ] ] ) ) {
					$missing[] = $hash;
				}
			}
			if ( empty( $missing ) ) {
				continue;
			}

			$now          = Clock::mysql();
			$placeholders = array();
			$values       = array();
			foreach ( $missing as $hash ) {
				$url            = $by_hash[ $hash ];
				$host           = substr( Url::host( $url ), 0, 191 );
				$placeholders[] = '(%s,%s,%s,%d,%d,%s,%s)';
				array_push( $values, $url, $hash, $host, Url::host_bucket( $host ), Url::is_internal( $url ) ? 1 : 0, 'pending', $now );
			}
			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare,PluginCheck.Security.DirectDB.UnescapedDBParameter -- Placeholders are generated above.
			$wpdb->query( $wpdb->prepare( 'INSERT IGNORE INTO %i (url, url_hash, host, host_bucket, is_internal, status, next_check_at) VALUES ' . implode( ',', $placeholders ), array_merge( array( $table ), $values ) ) );

			$ids += $this->lookup_ids( $missing, $by_hash );
			/**
			 * Fires after new links were discovered and stored as "pending".
			 *
			 * @param int $count Number of new links.
			 */
			do_action( 'unbreak_links_added', count( $missing ) );
		}

		return $ids;
	}

	/**
	 * Look up link IDs by hash.
	 *
	 * @param string[]             $hashes  Hashes.
	 * @param array<string,string> $by_hash hash => url.
	 * @return array<string,int> url => id
	 */
	private function lookup_ids( array $hashes, array $by_hash ): array {
		global $wpdb;
		$table = Schema::table( 'links' );
		$in    = implode( ',', array_fill( 0, count( $hashes ), '%s' ) );
		// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare -- Placeholders are generated above.
		$rows = $wpdb->get_results( $wpdb->prepare( "SELECT id, url_hash FROM %i WHERE url_hash IN ({$in})", array_merge( array( $table ), $hashes ) ) );
		$ids  = array();
		foreach ( (array) $rows as $row ) {
			if ( isset( $by_hash[ $row->url_hash ] ) ) {
				$ids[ $by_hash[ $row->url_hash ] ] = (int) $row->id;
			}
		}
		return $ids;
	}

	/**
	 * Multi-row insert of instance rows.
	 *
	 * @param array<int,array<int,mixed>> $rows Rows: link_id, source_type, source_id, source_key, anchor_text, element, context.
	 */
	private function insert_instances( array $rows ): void {
		global $wpdb;
		$table = Schema::table( 'instances' );
		foreach ( array_chunk( $rows, self::CHUNK ) as $chunk ) {
			$placeholders = array();
			$values       = array();
			foreach ( $chunk as $row ) {
				$formats = array( '%d', '%s', '%d', '%s', '%s', '%s', '%s' );
				$parts   = array();
				foreach ( $formats as $i => $format ) {
					// Nullable columns get a literal NULL instead of an empty string.
					if ( null === $row[ $i ] ) {
						$parts[] = 'NULL';
						continue;
					}
					$parts[]  = $format;
					$values[] = $row[ $i ];
				}
				$placeholders[] = '(' . implode( ',', $parts ) . ')';
			}
			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.PreparedSQL.NotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare,PluginCheck.Security.DirectDB.UnescapedDBParameter -- Placeholders are generated above.
			$wpdb->query( $wpdb->prepare( 'INSERT INTO %i (link_id, source_type, source_id, source_key, anchor_text, element, context) VALUES ' . implode( ',', $placeholders ), array_merge( array( $table ), $values ) ) );
		}
	}

	/**
	 * Delete links that no longer have any instance.
	 *
	 * @param int[] $link_ids Candidate link IDs.
	 */
	public function delete_orphans( array $link_ids ): void {
		global $wpdb;
		$links     = Schema::table( 'links' );
		$instances = Schema::table( 'instances' );
		foreach ( array_chunk( array_values( array_unique( array_map( 'intval', $link_ids ) ) ), self::CHUNK ) as $chunk ) {
			$in = implode( ',', array_fill( 0, count( $chunk ), '%d' ) );
			// phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared,WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching,WordPress.DB.PreparedSQLPlaceholders.UnfinishedPrepare,WordPress.DB.PreparedSQLPlaceholders.ReplacementsWrongNumber -- Placeholders are generated above.
			$wpdb->query( $wpdb->prepare( "DELETE l FROM %i l LEFT JOIN %i i ON i.link_id = l.id WHERE l.id IN ({$in}) AND i.id IS NULL", array_merge( array( $links, $instances ), $chunk ) ) );
		}
	}
}