h4b-image-optim/includes/class-cli-rescue.php

<?php
/**
 * wp h4b-img rescue — regenerate Smush-mangled JPGs from their .webp twins.
 *
 * @package H4B\ImageOptim
 */

namespace H4B\ImageOptim;

use Imagick;
use ImagickException;

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

final class CLI_Rescue {

	/**
	 * Find or fix JPGs damaged by Smush Ultra's grey-wash bug.
	 *
	 * ## OPTIONS
	 *
	 * [--scan]
	 * : Just scan; report counts. Default if neither --scan nor --apply given.
	 *
	 * [--apply]
	 * : Actually regenerate broken JPGs from their .webp siblings.
	 *
	 * [--min-severity=<float>]
	 * : Only rescue files with severity >= this. Default 10.0
	 *
	 * [--limit=<n>]
	 * : Maximum files to process. 0 = unlimited.
	 *
	 * [--csv=<path>]
	 * : Write per-file scan results to CSV.
	 *
	 * [--manifest=<path>]
	 * : Write JSON manifest of rescue actions.
	 *
	 * [--backup-dir=<path>]
	 * : Where to put pre-rescue backups. Default: wp-content/h4b-img-originals/_rescue_<timestamp>/
	 *
	 * ## EXAMPLES
	 *     wp h4b-img rescue --scan
	 *     wp h4b-img rescue --apply --min-severity=20 --limit=10
	 *     wp h4b-img rescue --apply --csv=/tmp/scan.csv --manifest=/tmp/manifest.json
	 */
	public function __invoke( $args, $assoc ): void {
		$apply        = ! empty( $assoc['apply'] );
		$min_sev      = (float) ( $assoc['min-severity'] ?? 10.0 );
		$limit        = (int) ( $assoc['limit'] ?? 0 );
		$csv_path     = $assoc['csv'] ?? null;
		$manifest_path = $assoc['manifest'] ?? null;
		$backup_dir   = $assoc['backup-dir'] ?? null;

		$uploads = wp_get_upload_dir();
		$basedir = trailingslashit( $uploads['basedir'] );
		// Smush's WebP siblings live in wp-content/smush-webp/, NOT inside uploads/.
		// Mirror tree: uploads/2026/02/foo.jpg  →  smush-webp/2026/02/foo.jpg.webp
		$webp_root = trailingslashit( WP_CONTENT_DIR ) . 'smush-webp/';

		if ( ! is_dir( $webp_root ) ) {
			\WP_CLI::error( "WebP root not found: $webp_root — rescue requires Smush's smush-webp/ directory to read clean WebP twins from." );
		}

		// Scan phase
		\WP_CLI::log( "Scanning $basedir for .jpg files…" );
		$jpgs = self::find_jpgs( $basedir );
		\WP_CLI::log( sprintf( 'Found %d JPG files.', count( $jpgs ) ) );

		$progress = \WP_CLI\Utils\make_progress_bar( 'Analysing', count( $jpgs ) );
		$results = [];
		$broken_list = [];
		foreach ( $jpgs as $jpg ) {
			$rel = substr( $jpg, strlen( $basedir ) );
			$webp = $webp_root . $rel . '.webp';
			// Also try the symmetric layout where webp is alongside the jpg
			if ( ! is_readable( $webp ) ) {
				$webp_alt = $jpg . '.webp';
				if ( is_readable( $webp_alt ) ) {
					$webp = $webp_alt;
				}
			}
			$res = Rescue_Detector::analyse( $jpg, $webp );
			$res['path']     = $jpg;
			$res['rel_path'] = $rel;
			$res['has_webp'] = is_readable( $webp );
			$results[] = $res;
			if ( $res['classification'] === 'broken' && $res['severity'] >= $min_sev ) {
				$broken_list[] = $res;
			}
			$progress->tick();
		}
		$progress->finish();

		// Sort broken by severity desc
		usort( $broken_list, fn( $a, $b ) => $b['severity'] <=> $a['severity'] );
		if ( $limit > 0 ) {
			$broken_list = array_slice( $broken_list, 0, $limit );
		}

		$by_class = [];
		foreach ( $results as $r ) {
			$by_class[ $r['classification'] ] = ( $by_class[ $r['classification'] ] ?? 0 ) + 1;
		}

		\WP_CLI::log( '' );
		\WP_CLI::log( 'Classification summary:' );
		foreach ( $by_class as $k => $v ) {
			\WP_CLI::log( sprintf( '  %-35s %5d', $k, $v ) );
		}
		\WP_CLI::log( sprintf( '  → %d to rescue at severity >= %.1f', count( $broken_list ), $min_sev ) );

		// CSV
		if ( $csv_path ) {
			$fh = fopen( $csv_path, 'w' );
			fputcsv( $fh, [ 'path', 'classification', 'severity', 'delta_mean', 'jpg_stddev', 'webp_mean', 'jpg_mean', 'webp_white_pixels', 'has_webp', 'error' ] );
			foreach ( $results as $r ) {
				fputcsv( $fh, [
					$r['rel_path'],
					$r['classification'],
					$r['severity'],
					$r['delta_mean'],
					$r['jpg_stddev'],
					$r['webp_mean'],
					$r['jpg_mean'],
					$r['webp_white_pixels'],
					$r['has_webp'] ? 1 : 0,
					$r['error'],
				] );
			}
			fclose( $fh );
			\WP_CLI::log( "Scan CSV written: $csv_path" );
		}

		if ( ! $apply ) {
			\WP_CLI::success( 'Scan complete. Use --apply to actually rescue.' );
			return;
		}

		if ( empty( $broken_list ) ) {
			\WP_CLI::success( 'Nothing to rescue.' );
			return;
		}

		// Rescue phase
		$timestamp = date( 'Ymd_His' );
		$backup_dir = $backup_dir
			?: ( $basedir . Optimizer::ORIGINALS_DIRNAME . '/_rescue_' . $timestamp );
		wp_mkdir_p( $backup_dir );
		\WP_CLI::log( "Backup dir: $backup_dir" );

		$progress = \WP_CLI\Utils\make_progress_bar( 'Rescuing', count( $broken_list ) );
		$actions = [];
		$done = 0;
		$errored = 0;
		foreach ( $broken_list as $r ) {
			$jpg = $r['path'];
			// Re-resolve the WebP using the same fallback logic
			$webp = $webp_root . $r['rel_path'] . '.webp';
			if ( ! is_readable( $webp ) ) {
				$webp_alt = $jpg . '.webp';
				if ( is_readable( $webp_alt ) ) {
					$webp = $webp_alt;
				}
			}
			$backup = trailingslashit( $backup_dir ) . $r['rel_path'];
			$res = self::rescue_one( $jpg, $webp, $backup );
			$res['path']     = $jpg;
			$res['severity'] = $r['severity'];
			$actions[] = $res;
			if ( $res['status'] === 'done' ) {
				$done++;
			} else {
				$errored++;
			}
			$progress->tick();
		}
		$progress->finish();

		if ( $manifest_path ) {
			file_put_contents( $manifest_path, wp_json_encode( [
				'timestamp'  => $timestamp,
				'backup_dir' => $backup_dir,
				'min_severity' => $min_sev,
				'total'      => count( $actions ),
				'done'       => $done,
				'errored'    => $errored,
				'actions'    => $actions,
			], JSON_PRETTY_PRINT ) );
			\WP_CLI::log( "Manifest: $manifest_path" );
		}

		\WP_CLI::success( sprintf(
			'Rescued %d / %d (errors: %d). Backups in %s',
			$done, count( $broken_list ), $errored, $backup_dir
		) );
	}

	/**
	 * @return string[]
	 */
	private static function find_jpgs( string $root ): array {
		$out = [];
		$it = new \RecursiveIteratorIterator(
			new \RecursiveDirectoryIterator( $root, \FilesystemIterator::SKIP_DOTS )
		);
		foreach ( $it as $entry ) {
			/** @var \SplFileInfo $entry */
			if ( ! $entry->isFile() ) {
				continue;
			}
			$path = $entry->getPathname();
			// Skip our own working directories
			if ( strpos( $path, '/' . Optimizer::ORIGINALS_DIRNAME . '/' ) !== false
				|| strpos( $path, '/smush-webp/' ) !== false ) {
				continue;
			}
			$ext = strtolower( $entry->getExtension() );
			if ( $ext === 'jpg' || $ext === 'jpeg' ) {
				$out[] = $path;
			}
		}
		return $out;
	}

	/**
	 * Rescue a single broken JPG by re-encoding from its clean WebP twin.
	 *
	 * @return array{status:string, new_size:int, original_size:int, error:?string, backup_path:string}
	 */
	private static function rescue_one( string $jpg, string $webp, string $backup_path ): array {
		$res = [
			'status'        => 'pending',
			'new_size'      => 0,
			'original_size' => is_readable( $jpg ) ? filesize( $jpg ) : 0,
			'error'         => null,
			'backup_path'   => $backup_path,
		];

		if ( ! is_readable( $webp ) ) {
			$res['status'] = 'error';
			$res['error']  = 'webp_not_readable';
			return $res;
		}

		// Backup first
		wp_mkdir_p( dirname( $backup_path ) );
		if ( ! @copy( $jpg, $backup_path ) ) {
			$res['status'] = 'error';
			$res['error']  = 'backup_failed';
			return $res;
		}
		if ( filesize( $backup_path ) !== $res['original_size'] ) {
			$res['status'] = 'error';
			$res['error']  = 'backup_size_mismatch';
			return $res;
		}

		// Decode WebP → re-encode as JPEG q=90, 4:4:4, sRGB, ICC preserved
		try {
			$img = new Imagick( $webp );
			$img->setImageFormat( 'jpeg' );
			$img->setImageCompressionQuality( 90 );
			$img->setSamplingFactors( [ '1x1', '1x1', '1x1' ] ); // 4:4:4 — no chroma loss
			$img->setInterlaceScheme( Imagick::INTERLACE_NO );
			$img->setImageColorspace( Imagick::COLORSPACE_SRGB );

			// Ensure an ICC profile is attached
			ICC_Profile::preserve_or_inject( $img );

			// Atomic write
			$tmp = $jpg . '.h4b.tmp';
			$img->writeImage( $tmp );
			$img->clear();

			$stat = stat( $jpg );
			if ( $stat !== false ) {
				@chown( $tmp, $stat['uid'] );
				@chgrp( $tmp, $stat['gid'] );
				@chmod( $tmp, $stat['mode'] & 0777 );
			}
			rename( $tmp, $jpg );
		} catch ( ImagickException $e ) {
			$res['status'] = 'error';
			$res['error']  = 'reencode_failed: ' . $e->getMessage();
			// Best-effort restore from backup
			@copy( $backup_path, $jpg );
			return $res;
		}

		clearstatcache( true, $jpg );
		$res['new_size'] = filesize( $jpg );
		$res['status']   = 'done';
		return $res;
	}
}