<?php
/**
 * Known active-theme images only: no directory crawl or theme-file writes.
 *
 * @package AffStackDevImageManager
 */

namespace AffStackDev\ImageManager;

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * Pictures of the active theme and of the other picture folders: found on scanned public pages, indexed by their files,
 * and delivered with their copies on public pages; theme files are never changed.
 */
final class Theme {
	/**
	 * Whether a source row is a theme picture.
	 *
	 * @param array $row The source row.
	 * @return bool True when it is.
	 */
	public static function is_source( array $row ): bool {
		return 'theme' === ( $row['details']['kind'] ?? '' );
	}
	/**
	 * The folders theme pictures may come from: the parent and child theme, and the other picture folders chosen in
	 * Settings.
	 *
	 * @return array Folder key => its real path and address (in the home address's scheme).
	 */
	private static function roots(): array {
		if ( ! function_exists( 'get_template_directory' ) || ! function_exists( 'get_stylesheet_directory' ) ) {
			return array();
		}
		$out   = array();
		$pairs = array( array( get_template_directory(), get_template_directory_uri() ), array( get_stylesheet_directory(), get_stylesheet_directory_uri() ) );
		foreach ( self::folders() as $folder ) {
			$pairs[] = array( WP_CONTENT_DIR . '/' . $folder, content_url( $folder ) );
		}
		foreach ( $pairs as  $pair ) {
			list( $dir, $uri ) = $pair;
			$root              = realpath( $dir );
			// 1.9.0 (F-IM-28): the root keeps the home address's scheme, so a worker started from WP-CLI (no HTTPS request,
			// content_url() says http) finds the same roots and keys as a visitor's HTTPS request.
			$uri = set_url_scheme( rtrim( $uri, '/' ), (string) wp_parse_url( home_url( '/' ), PHP_URL_SCHEME ) );
			if ( $root && ! is_link( $dir ) && Urls::origin( $uri ) === Urls::origin( home_url( '/' ) ) ) {
				$out[ hash( 'sha256', $uri ) ] = array(
					'dir' => $root,
					'uri' => $uri,
				);
			}
		}
		return $out;
	}
	/**
	 * 1.9.0 (D-500 gap 11): the folders inside wp-content the store owner named (Settings), each a real folder there that is
	 * not the uploads folder or inside it, not a link, and not Image Manager's own.
	 *
	 * @param array|null $names Folder names to check instead of the saved ones.
	 * @return array Relative folder names.
	 */
	public static function folders( ?array $names = null ): array {
		$content = realpath( WP_CONTENT_DIR );
		$uploads = realpath( (string) wp_get_upload_dir()['basedir'] );
		$own     = realpath( DIR );
		$out     = array();
		foreach ( array_slice( (array) ( $names ?? ( Settings::get()['picture_folders'] ?? array() ) ), 0, 10 ) as $name ) {
			$name = trim( (string) $name, " \t/" );
			if ( '' === $name || strlen( $name ) > 200 || ! preg_match( '~^[A-Za-z0-9._-]+(?:/[A-Za-z0-9._-]+)*$~D', $name ) || preg_match( '~(?:^|/)\.\.?(?:/|$)~', $name ) ) {
				continue;
			}
			$dir  = WP_CONTENT_DIR . '/' . $name;
			$real = realpath( $dir );
			if ( ! $content || ! $real || is_link( $dir ) || ! is_dir( $real ) || ! str_starts_with( $real, $content . '/' ) ) {
				continue;
			}
			if ( ( $uploads && ( $real === $uploads || str_starts_with( $real, $uploads . '/' ) || str_starts_with( $uploads, $real . '/' ) ) ) || ( $own && ( $real === $own || str_starts_with( $real, $own . '/' ) ) ) ) {
				continue;
			}
			$out[] = $name;
		}
		return array_values( array_unique( $out ) );
	}
	/**
	 * Returns a verified local image descriptor, never trusting a submitted filesystem path.
	 *
	 * @param string $url The picture's address on a page.
	 * @return array|null The folder it is in, its path inside it, its address without query and its real path, or null
	 *          when it is no picture of those folders.
	 */
	public static function resolve( string $url ): ?array {
		if ( ! Settings::get()['theme_images'] || strlen( $url ) > 2048 ) {
			return null;
		}
		$p = wp_parse_url( $url );
		if ( ! is_array( $p ) || ! empty( $p['user'] ) || ! empty( $p['pass'] ) || ! empty( $p['fragment'] ) || ! Urls::safe_path( $p['path'] ?? '' ) ) {
			return null;
		}
		if ( ! empty( $p['query'] ) ) {
			parse_str( $p['query'], $q );
			if ( array_diff( array_keys( $q ), array( 'ver', 'version' ) ) ) {
				return null;
			}
			foreach ( $q as  $v ) {
				if ( ! is_string( $v ) || ! preg_match( '/^[a-zA-Z0-9._-]{1,64}$/D', $v ) ) {
					return null;
				}
			}
		}
		$clean = strtok( $url, '?' );
		foreach ( self::roots() as  $key => $root ) {
			if ( ! str_starts_with( $clean, $root['uri'] . '/' ) ) {
				continue;
			}
			$relative = rawurldecode( substr( $clean, strlen( $root['uri'] ) + 1 ) );
			if ( ! preg_match( '/\.(?:jpe?g|png|gif|webp)$/iD', $relative ) || str_contains( $relative, '..' ) || preg_match( '~[\\\\\x00-\x1f:]~', $relative ) ) {
				return null;
			}
			$path = $root['dir'];
			foreach ( explode( '/', $relative ) as  $piece ) {
				if ( '' === $piece || '.' === $piece ) {
					return null;
				}
				$path .= '/' . $piece;
				if ( is_link( $path ) ) {
					return null;
				}
			}
			$real = realpath( $path );
			if ( ! $real || ! str_starts_with( $real, $root['dir'] . '/' ) || ! is_file( $real ) || ! is_readable( $real ) ) {
				return null;
			}
			return array(
				'root'     => $key,
				'relative' => $relative,
				'url'      => $clean,
				'path'     => $real,
			);
		}
		return null;
	}
	/**
	 * The theme source an address is indexed under.
	 *
	 * @param string $url The picture's address (its query is ignored).
	 * @return array|null The source row, or null.
	 */
	public static function find( string $url ): ?array {
		global $wpdb;
		$clean = strtok( $url, '?' );
        // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Read/update the plugin theme-source URL index and collision records; current mappings are required during rediscovery.
		$id  = (int) $wpdb->get_var( $wpdb->prepare( 'SELECT source_id FROM %i WHERE url_key=%s', Repository::table( 'urls' ), hash( 'sha256', 'theme-url:' . $clean ) ) );
		$row = $id ? Repository::row( $id ) : null;
		return $row && self::is_source( $row ) ? $row : null;
	}
	/**
	 * Group only responsive candidates encountered on the same actual HTML element.
	 *
	 * @param array  $urls     The addresses one element offers (src and srcset).
	 * @param string $page     The page they were found on.
	 * @param int    $priority The source's priority.
	 * @return array The addresses handled.
	 */
	public static function discover( array $urls, string $page, int $priority ): array {
		global $wpdb;
		$files   = array();
		$row     = null;
		$handled = array();
		foreach ( array_slice( array_unique( $urls ), 0, 32 ) as  $url ) {
			$r = self::resolve( $url );
			if ( ! $r ) {
				continue;
			}
			$info = wp_getimagesize( $r['path'] );
			if ( ! $info || ! in_array( $info['mime'] ?? '', array( 'image/jpeg', 'image/png', 'image/gif', 'image/webp' ), true ) ) {
				continue;
			}
			$known = self::find( $url );
			if ( $known && $row && (int) $known['id'] !== (int) $row['id'] ) {
				continue;
			}
			if ( $known ) {
				$row = $known;
			}
			$key           = $r['root'] . '/' . $r['relative'];
			$files[ $key ] = array(
				'root'     => $r['root'],
				'relative' => $r['relative'],
				'url'      => $r['url'],
				'width'    => (int) $info[0],
				'height'   => (int) $info[1],
			);
			$handled[]     = $url;
		}
		if ( ! $files ) {
			return array();
		}
		$first = reset( $files );
		if ( ! $row ) {
			// Reuse an earlier unmapped row for the main file rather than duplicating the pilot inventory.
			$row = Repository::discover(
				0,
				$priority,
				array(
					'url'     => $first['url'],
					'page'    => $page,
					'surface' => 'active_theme',
					'mapping' => 'verified_theme_file',
				)
			);
		}
		$d                = $row['details'];
		$before           = $d['theme_files'] ?? array();
		$d['kind']        = 'theme';
		$d['surface']     = 'active_theme';
		$d['mapping']     = 'verified_theme_file';
		$d['theme_files'] = array_slice( array_merge( $before, $files ), 0, 32, true );
		$d['pages']       = array_slice( array_values( array_unique( array_merge( $d['pages'] ?? array(), array( $page ) ) ) ), 0, 12 );
		$d['urls']        = array_slice( array_values( array_unique( array_merge( $d['urls'] ?? array(), $handled ) ) ), 0, 32 );
		$values           = array(
			'details'  => $d,
			'priority' => min( (int) $row['priority'], $priority ),
		);
		$changed          = $before !== $d['theme_files'];
		if ( 'unmapped' === $row['state'] || ( $changed && ! $row['excluded'] ) ) {
			$values['state']  = 'discovered';
			$values['reason'] = '';
		}
		if ( $changed && ! empty( $row['manifest']['schema'] ) ) {
			$m                  = $row['manifest'];
			$m['meta_hash']     = hash( 'sha256', wp_json_encode( $d['theme_files'] ) );
			$values['manifest'] = $m;
		}
		Repository::update( (int) $row['id'], $values );
		foreach ( $files as  $f ) {
			$key = hash( 'sha256', 'theme-url:' . $f['url'] );
            // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Read/update the plugin theme-source URL index and collision records; current mappings are required during rediscovery.
			Reviewed_Store::database_query( $wpdb->prepare( 'INSERT IGNORE INTO %i (url_key,attachment_id,source_id) VALUES (%s,0,%d)', Repository::table( 'urls' ), $key, (int) $row['id'] ) );
			// Retire only duplicate unmapped inventory entries: no files or prior optimization records are deleted.
            // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Read/update the plugin theme-source URL index and collision records; current mappings are required during rediscovery.
			$duplicate = $wpdb->get_var( $wpdb->prepare( "SELECT id FROM %i WHERE source_key=%s AND state='unmapped' AND id<>%d", Repository::table( 'sources' ), hash( 'sha256', 'unmapped:' . $f['url'] ), (int) $row['id'] ) );
            // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery,WordPress.DB.DirectDatabaseQuery.NoCaching -- Read/update the plugin theme-source URL index and collision records; current mappings are required during rediscovery.
			if ( $duplicate ) {
				Reviewed_Store::database_delete(
					Repository::table( 'sources' ),
					array(
						'id'    => (int) $duplicate,
						'state' => 'unmapped',
					)
				);
			}
		}
		return $handled;
	}
	/**
	 * A theme source's files that still resolve to the same file.
	 *
	 * @param array $row The source row.
	 * @return array Path key => relative path, real path, width, height and name.
	 */
	public static function files( array $row ): array {
		if ( ! self::is_source( $row ) || ! Settings::get()['theme_images'] ) {
			return array();
		}
		$out = array();
		foreach ( $row['details']['theme_files'] ?? array() as  $key => $f ) {
			$r = self::resolve( $f['url'] ?? '' );
			if ( ! $r || $r['root'] !== $f['root'] || $r['relative'] !== $f['relative'] ) {
				continue;
			}
			$out[ $key ] = array(
				'relative' => $key,
				'path'     => $r['path'],
				'width'    => (int) $f['width'],
				'height'   => (int) $f['height'],
				'name'     => basename( $f['relative'] ),
			);
		}
		return $out;
	}
	/**
	 * Whether a theme source may be delivered: found on a public page, not printed inside a picture element, its files
	 * still there.
	 *
	 * @param array $row The source row.
	 * @return bool True when it may.
	 */
	public static function eligible( array $row ): bool {
		if ( ! self::is_source( $row ) || ! Settings::get()['theme_images'] || empty( $row['details']['pages'] ) || ! empty( $row['details']['native_picture'] ) || ! self::files( $row ) ) {
			return false;
		}
		/**
		 * Filters whether a theme image carried over from an earlier release's record (its copies keep their earlier names) may still be treated as public.
		 *
		 * @since 1.7.0.4 Earliest release in the source history.
		 *
		 * @param bool  $allowed True to allow.
		 * @param array $row     Image Manager's index row of the theme image.
		 */
		if ( ! empty( $row['details']['legacy_copy_prefix'] ) && true !== apply_filters( 'afsd_im_public_theme_image_allowed', true, $row ) ) {
			return false;
		}
		/**
		 * Filters whether a theme image seen on public pages is eligible for optimization and delivery.
		 *
		 * @since 1.7.0.4 Earliest release in the source history.
		 *
		 * @param bool  $allowed   True to allow.
		 * @param int   $source_id Image Manager's source ID of the theme image.
		 * @param array $details   The source's recorded details (its theme files, the public pages and addresses it was seen on).
		 */
		return true === apply_filters( 'afsd_im_theme_source_allowed', true, (int) $row['id'], $row['details'] );
	}
	/**
	 * The theme picture pass on a public page (not a product page), where delivery is allowed: a pass on the page's
	 * whole HTML through WordPress's template enhancement output buffer (Hooks::PAGE_HTML).
	 */
	public static function start_buffer(): void {
		if ( ! Settings::get()['theme_images'] || ! Delivery::allowed() || ( function_exists( 'is_product' ) && is_product() ) ) {
			return;
		}
		// The response is bounded, read-only, and only indexed active-theme images can change.
		add_filter( Hooks::PAGE_HTML, array( self::class, 'render' ), Hooks::PASS_THEME );
	}
	/**
	 * The buffered page after the theme picture pass; the page as it was when anything goes wrong.
	 *
	 * @param string $html The page.
	 * @return string The page.
	 */
	public static function render( string $html ): string {
		try {
			return self::render_images( $html );
		} catch ( \Throwable $e ) {
			return $html;
		} // Never turn a delivery integration failure into a broken public response.
	}
	/**
	 * Wrap the indexed theme pictures of a page in picture elements with their copies.
	 *
	 * @param string $html The page.
	 * @return string The page.
	 */
	private static function render_images( string $html ): string {
		if ( strlen( $html ) > 2097152 || ! Delivery::allowed() || ( function_exists( 'is_product' ) && is_product() ) || ! Settings::get()['theme_images'] ) {
			return $html;
		}
		$p = new \WP_HTML_Tag_Processor( $html );
		if ( ! method_exists( $p, 'set_attribute' ) || ! method_exists( $p, 'is_tag_closer' ) ) {
			return $html;
		}
		$blocked = 0;
		$serial  = bin2hex( random_bytes( 12 ) );
		$wraps   = array();
		$n       = 0;
		while ( $n++ < 8000 && $p->next_tag( array( 'tag_closers' => 'visit' ) ) ) {
			$tag = $p->get_tag();
			if ( in_array( $tag, array( 'PICTURE', 'NOSCRIPT', 'SVG', 'VIDEO', 'AUDIO' ), true ) ) {
				$blocked = $p->is_tag_closer() ? max( 0, $blocked - 1 ) : $blocked + 1;
				continue;
			}
			if ( $blocked || 'IMG' !== $tag || $p->is_tag_closer() || count( $wraps ) >= 128 ) {
				continue;
			}
			$source = self::source_markup( $p );
			if ( ! $source ) {
				continue;
			}
			$marker           = $serial . '-' . count( $wraps );
			$wraps[ $marker ] = $source;
			$p->set_attribute( 'data-afsd-im-wrap', $marker );
		}
		if ( ! $wraps ) {
			return $html;
		}
		$marked = $p->get_updated_html();
		$output = preg_replace_callback(
			'~<img\b(?:[^>\x22\x27]|\x22[^\x22]*\x22|\x27[^\x27]*\x27)*>~i',
			static function ( $m ) use ( $wraps ) {
				if ( ! preg_match( '~\sdata-afsd-im-wrap="([a-f0-9]+-[0-9]+)"~', $m[0], $tag ) || ! isset( $wraps[ $tag[1] ] ) ) {
					return $m[0];
				}
				$img = str_replace( $tag[0], '', $m[0] );
				return '<picture class="afsd-im-picture">' . $wraps[ $tag[1] ] . $img . '</picture>';
			},
			$marked
		);
		return is_string( $output ) ? $output : $html;
	}
	/**
	 * The source elements for the img the processor is on, or nothing when it is scripted or lazy-loaded by another
	 * plugin.
	 *
	 * @param \WP_HTML_Tag_Processor $p The HTML processor, on an img.
	 * @return string The source elements' HTML, or an empty string.
	 */
	private static function source_markup( \WP_HTML_Tag_Processor $p ): string {
		foreach ( array( 'data-src', 'data-lazy-src', 'data-original', 'data-lazy-original', 'data-srcset', 'data-lazy-srcset', 'data-large_image', 'data-large-image', 'data-afsd-im', 'data-afsd-im-wrap' ) as  $attr ) {
			if ( null !== $p->get_attribute( $attr ) ) {
				return '';
			}
		}
		$src = $p->get_attribute( 'src' );
		if ( ! is_string( $src ) ) {
			return '';
		}
		$row = self::find( $src );
		if ( ! $row || $row['excluded'] || ! $row['delivery'] || ! self::eligible( $row ) ) {
			return '';
		}
		$files = self::files( $row );
		$map   = $row['details']['theme_files'];
		$get   = static function ( string $url, int $width = 0 ) use ( $row, $map, $files ) {
			$r = self::resolve( $url );
			if ( ! $r ) {
				return null;
			}
			$key = $r['root'] . '/' . $r['relative'];
			$v   = $row['manifest']['variants'][ $key ] ?? null;
			if ( ( ! $v && empty( $row['manifest']['responsive'] ) ) || ! isset( $files[ $key ], $map[ $key ] ) || ( $width && $files[ $key ]['width'] !== $width ) ) {
				return null;
			}
			return array(
				'variant'    => $v ?? array(),
				'file'       => $files[ $key ],
				'width'      => $width,
				'native_url' => $url,
			);
		};
		$base  = $get( $src );
		if ( ! $base ) {
			return '';
		}
		$srcset  = $p->get_attribute( 'srcset' );
		$entries = array();
		if ( is_string( $srcset ) && '' !== $srcset ) {
			$candidates = Urls::srcset( $srcset );
			if ( ! $candidates ) {
				return '';
			}
			foreach ( $candidates as  $c ) {
				$item = $get( $c['url'], $c['width'] );
				if ( ! $item ) {
					return '';
				}$entries[] = $item;
			}
		} else {
			$entries[] = $base;
		}
		$sizes = $p->get_attribute( 'sizes' );
		if ( is_string( $sizes ) ) {
			$sizes = preg_replace( '/^auto(?:\s*,\s*|\s*$)/i', '', $sizes );
		}
		if ( count( $entries ) > 1 && ! $sizes ) {
			return '';
		}
		return Delivery::sources_html( Responsive::extend( Delivery::complete_sets( $entries, ! empty( $row['manifest']['responsive'] ) ), $entries, $row, is_string( $sizes ) ? $sizes : '' ), is_string( $sizes ) ? $sizes : '' );
	}
}