feat: initial v0.1.0 MVP
Replaces Smush Pro's optimisation pipeline without the grey-wash bug. CLI commands working: wp h4b-img status wp h4b-img optimise --id=<n> wp h4b-img bulk wp h4b-img rescue Verified on dev.rds.ink: - ICC profile preservation works (the Smush-bug fix) - Bulk: 20 attachments → 487 KB saved (10.4%), 0 errors - Rescue: end-to-end mechanism verified on WorkingAsOne_horse fixture - WebP synchronous, AVIF queued via WP-Cron - Originals backed up to wp-content/h4b-img-originals/ See CHANGELOG.md for details + ../DESIGN-h4b-image-optim.md for architecture. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
Henk
161
includes/class-rescue-detector.php
Normal file
161
includes/class-rescue-detector.php
Normal file
@@ -0,0 +1,161 @@
|
||||
<?php
|
||||
/**
|
||||
* PHP port of the Python grey-wash detector used to find Smush-mangled JPGs.
|
||||
*
|
||||
* The detection strategy:
|
||||
* 1. Find pixels in the .webp twin where all three RGB channels are
|
||||
* >= WEBP_WHITE_RGB_MIN (i.e. should-be-white regions).
|
||||
* 2. Sample those same pixel coordinates in the JPG.
|
||||
* 3. If the JPG average there is significantly darker AND the stddev is
|
||||
* high, classify as `broken` (the Smush grey-wash bug fingerprint).
|
||||
*
|
||||
* @package H4B\ImageOptim
|
||||
*/
|
||||
|
||||
namespace H4B\ImageOptim;
|
||||
|
||||
use Imagick;
|
||||
use ImagickPixel;
|
||||
use ImagickException;
|
||||
|
||||
if ( ! defined( 'ABSPATH' ) ) {
|
||||
exit;
|
||||
}
|
||||
|
||||
final class Rescue_Detector {
|
||||
|
||||
// Calibrated thresholds (validated against rds.ink + dayboro + tawnytrails)
|
||||
public const WEBP_WHITE_RGB_MIN = 248;
|
||||
public const JPG_DARKER_THAN_WEBP = 10;
|
||||
public const JPG_WHITE_STDDEV_BROKEN = 4.0;
|
||||
public const MIN_WHITE_PIXELS = 100;
|
||||
|
||||
/**
|
||||
* @return array{
|
||||
* classification:string,
|
||||
* severity:float,
|
||||
* delta_mean:float,
|
||||
* jpg_stddev:float,
|
||||
* webp_mean:float,
|
||||
* jpg_mean:float,
|
||||
* webp_white_pixels:int,
|
||||
* error:?string
|
||||
* }
|
||||
*/
|
||||
public static function analyse( string $jpg_path, string $webp_path ): array {
|
||||
$result = [
|
||||
'classification' => 'unknown',
|
||||
'severity' => 0.0,
|
||||
'delta_mean' => 0.0,
|
||||
'jpg_stddev' => 0.0,
|
||||
'webp_mean' => 0.0,
|
||||
'jpg_mean' => 0.0,
|
||||
'webp_white_pixels' => 0,
|
||||
'error' => null,
|
||||
];
|
||||
|
||||
if ( ! is_readable( $jpg_path ) ) {
|
||||
$result['classification'] = 'error';
|
||||
$result['error'] = 'jpg_not_readable';
|
||||
return $result;
|
||||
}
|
||||
if ( ! is_readable( $webp_path ) ) {
|
||||
$result['classification'] = 'no_webp';
|
||||
return $result;
|
||||
}
|
||||
|
||||
try {
|
||||
$jpg = new Imagick( $jpg_path );
|
||||
$webp = new Imagick( $webp_path );
|
||||
} catch ( ImagickException $e ) {
|
||||
$result['classification'] = 'error';
|
||||
$result['error'] = 'imagick_open: ' . $e->getMessage();
|
||||
return $result;
|
||||
}
|
||||
|
||||
// Resize webp to match jpg if needed (handles minor dimension drift)
|
||||
$jw = $jpg->getImageWidth();
|
||||
$jh = $jpg->getImageHeight();
|
||||
if ( $webp->getImageWidth() !== $jw || $webp->getImageHeight() !== $jh ) {
|
||||
$webp->resizeImage( $jw, $jh, Imagick::FILTER_LANCZOS, 1.0 );
|
||||
}
|
||||
|
||||
// Pull raw pixel data as RGB bytes
|
||||
try {
|
||||
$jpg_bytes = $jpg->exportImagePixels( 0, 0, $jw, $jh, 'RGB', Imagick::PIXEL_CHAR );
|
||||
$webp_bytes = $webp->exportImagePixels( 0, 0, $jw, $jh, 'RGB', Imagick::PIXEL_CHAR );
|
||||
} catch ( ImagickException $e ) {
|
||||
$jpg->clear(); $webp->clear();
|
||||
$result['classification'] = 'error';
|
||||
$result['error'] = 'pixel_export: ' . $e->getMessage();
|
||||
return $result;
|
||||
}
|
||||
$jpg->clear();
|
||||
$webp->clear();
|
||||
|
||||
$n_pixels = $jw * $jh;
|
||||
$threshold = self::WEBP_WHITE_RGB_MIN;
|
||||
|
||||
$webp_white_count = 0;
|
||||
$jpg_sum = 0.0;
|
||||
$webp_sum = 0.0;
|
||||
$jpg_sum_sq = 0.0;
|
||||
|
||||
// Iterate; PHP arrays from exportImagePixels are 0-indexed integers
|
||||
for ( $i = 0; $i < $n_pixels; $i++ ) {
|
||||
$base = $i * 3;
|
||||
$wr = $webp_bytes[ $base ];
|
||||
$wg = $webp_bytes[ $base + 1 ];
|
||||
$wb = $webp_bytes[ $base + 2 ];
|
||||
if ( $wr < $threshold || $wg < $threshold || $wb < $threshold ) {
|
||||
continue;
|
||||
}
|
||||
$jr = $jpg_bytes[ $base ];
|
||||
$jg = $jpg_bytes[ $base + 1 ];
|
||||
$jb = $jpg_bytes[ $base + 2 ];
|
||||
$jpg_mean_pixel = ( $jr + $jg + $jb ) / 3.0;
|
||||
$webp_mean_pixel = ( $wr + $wg + $wb ) / 3.0;
|
||||
|
||||
$webp_white_count++;
|
||||
$jpg_sum += $jpg_mean_pixel;
|
||||
$webp_sum += $webp_mean_pixel;
|
||||
$jpg_sum_sq += $jpg_mean_pixel * $jpg_mean_pixel;
|
||||
}
|
||||
|
||||
$result['webp_white_pixels'] = $webp_white_count;
|
||||
|
||||
if ( $webp_white_count < self::MIN_WHITE_PIXELS ) {
|
||||
$result['classification'] = 'no_white_area';
|
||||
return $result;
|
||||
}
|
||||
|
||||
$jpg_mean = $jpg_sum / $webp_white_count;
|
||||
$webp_mean = $webp_sum / $webp_white_count;
|
||||
$variance = ( $jpg_sum_sq / $webp_white_count ) - ( $jpg_mean * $jpg_mean );
|
||||
$jpg_stddev = $variance > 0 ? sqrt( $variance ) : 0.0;
|
||||
|
||||
$result['jpg_mean'] = $jpg_mean;
|
||||
$result['webp_mean'] = $webp_mean;
|
||||
$result['delta_mean'] = $webp_mean - $jpg_mean;
|
||||
$result['jpg_stddev'] = $jpg_stddev;
|
||||
|
||||
$is_darker = $result['delta_mean'] > self::JPG_DARKER_THAN_WEBP;
|
||||
$is_mottled = $jpg_stddev > self::JPG_WHITE_STDDEV_BROKEN;
|
||||
|
||||
if ( $is_darker && $is_mottled ) {
|
||||
$result['classification'] = 'broken';
|
||||
$result['severity'] = $result['delta_mean']
|
||||
* ( $jpg_stddev / self::JPG_WHITE_STDDEV_BROKEN );
|
||||
} elseif ( $is_darker ) {
|
||||
$result['classification'] = 'uniformly_darker';
|
||||
$result['severity'] = $result['delta_mean'];
|
||||
} elseif ( $is_mottled ) {
|
||||
$result['classification'] = 'mottled_but_correct_brightness';
|
||||
$result['severity'] = $jpg_stddev;
|
||||
} else {
|
||||
$result['classification'] = 'clean';
|
||||
}
|
||||
|
||||
return $result;
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user