Plugin Directory: Extract asset image dimensions during plugin import

wordpress.org/public_html/wp-content/plugins/plugin-directory/bin/backfill-asset-dimensions.php

@@ -0,0 +1,171 @@
+<?php
+// phpcs:ignoreFile — CLI script: output not escaped, loop vars shadow WP globals.
+namespace WordPressdotorg\Plugin_Directory;
+
+use WordPressdotorg\Plugin_Directory\CLI\Import;
+
+// This script should only be called in a CLI environment.
+if ( 'cli' != php_sapi_name() ) {
+	die();
+}
+
+$opts = getopt( '', array( 'url:', 'abspath:', 'plugin:', 'limit:', 'dry-run' ) );
+
+if ( empty( $opts['url'] ) ) {
+	$opts['url'] = 'https://wordpress.org/plugins/';
+}
+if ( empty( $opts['abspath'] ) && false !== strpos( __DIR__, 'wp-content' ) ) {
+	$opts['abspath'] = substr( __DIR__, 0, strpos( __DIR__, 'wp-content' ) );
+}
+
+$opts['dry-run'] = isset( $opts['dry-run'] );
+$opts['limit']   = isset( $opts['limit'] ) ? (int) $opts['limit'] : 0;
+
+// Bootstrap WordPress.
+$_SERVER['HTTP_HOST']   = parse_url( $opts['url'], PHP_URL_HOST );
+$_SERVER['REQUEST_URI'] = parse_url( $opts['url'], PHP_URL_PATH );
+
+require rtrim( $opts['abspath'], '/' ) . '/wp-load.php';
+
+if ( ! class_exists( '\WordPressdotorg\Plugin_Directory\Plugin_Directory' ) ) {
+	fwrite( STDERR, "Error! This site doesn't have the Plugin Directory plugin enabled.\n" );
+	if ( defined( 'WPORG_PLUGIN_DIRECTORY_BLOGID' ) ) {
+		fwrite( STDERR, "Run the following command instead:\n" );
+		fwrite( STDERR, "\tphp " . implode( ' ', $argv ) . ' --url ' . get_site_url( WPORG_PLUGIN_DIRECTORY_BLOGID, '/' ) . "\n" );
+	}
+	die();
+}
+
+global $wpdb;
+
+// Be more patient with slow SVN responses during a bulk run than the
+// per-request import path is. Uses http_request_args (not the timeout
+// filter) so the helper's explicit `timeout => 15` is overridden.
+add_filter(
+	'http_request_args',
+	function ( $args ) {
+		$args['timeout'] = 30;
+		return $args;
+	}
+);
+
+$meta_keys = array(
+	'screenshot' => 'assets_screenshots',
+	'banner'     => 'assets_banners',
+	'icon'       => 'assets_icons',
+);
+
+if ( ! empty( $opts['plugin'] ) ) {
+	$plugin = Plugin_Directory::get_plugin_post( $opts['plugin'] );
+	if ( ! $plugin ) {
+		fwrite( STDERR, "Plugin '{$opts['plugin']}' not found.\n" );
+		exit( 1 );
+	}
+	$post_ids = array( $plugin->ID );
+} else {
+	$post_ids = $wpdb->get_col( $wpdb->prepare(
+		"SELECT DISTINCT post_id
+		   FROM {$wpdb->postmeta}
+		  WHERE meta_key IN ( %s, %s, %s )
+		    AND meta_value != ''
+		    AND meta_value != 'a:0:{}'
+		  ORDER BY post_id ASC",
+		$meta_keys['screenshot'],
+		$meta_keys['banner'],
+		$meta_keys['icon']
+	) );
+
+	if ( $opts['limit'] > 0 ) {
+		$post_ids = array_slice( $post_ids, 0, $opts['limit'] );
+	}
+}
+
+$total_plugins   = count( $post_ids );
+$plugins_updated = 0;
+$counts          = array(
+	'reused'    => 0,
+	'extracted' => 0,
+	'failed'    => 0,
+);
+$failed_files    = array();
+
+echo "Scanning {$total_plugins} plugins" . ( $opts['dry-run'] ? ' (dry-run)' : '' ) . "...\n";
+
+foreach ( $post_ids as $i => $post_id ) {
+	$plugin = get_post( $post_id );
+	if ( ! $plugin || 'plugin' !== $plugin->post_type ) {
+		continue;
+	}
+
+	$slug         = $plugin->post_name; // Only used for the failure list.
+	$plugin_dirty = false;
+
+	foreach ( $meta_keys as $type => $meta_key ) {
+		$records = get_post_meta( $post_id, $meta_key, true );
+		if ( ! is_array( $records ) || ! $records ) {
+			continue;
+		}
+
+		$meta_dirty = false;
+
+		foreach ( $records as $filename => $record ) {
+			if ( ! is_array( $record ) || empty( $record['filename'] ) || ! isset( $record['revision'] ) ) {
+				continue;
+			}
+
+			if ( ! empty( $record['width'] ) && ! empty( $record['height'] ) ) {
+				++$counts['reused'];
+				continue;
+			}
+
+			// Pass the existing record as the prior so the helper short-circuits
+			// for cases it already knows the answer for.
+			$updated = Import::enrich_asset_dimensions( $record, $record, $plugin );
+
+			if ( ! empty( $updated['width'] ) && ! empty( $updated['height'] ) ) {
+				++$counts['extracted'];
+			} else {
+				++$counts['failed'];
+				$failed_files[] = "{$slug} {$type} {$record['filename']}";
+			}
+
+			if ( $updated !== $record ) {
+				$records[ $filename ] = $updated;
+				$meta_dirty           = true;
+			}
+		}
+
+		if ( $meta_dirty ) {
+			$plugin_dirty = true;
+			if ( ! $opts['dry-run'] ) {
+				update_post_meta( $post_id, $meta_key, wp_slash( $records ) );
+			}
+		}
+	}
+
+	if ( $plugin_dirty ) {
+		++$plugins_updated;
+	}
+
+	if ( ( $i + 1 ) % 100 === 0 ) {
+		echo sprintf(
+			"  [%d/%d] reused=%d extracted=%d failed=%d\n",
+			$i + 1,
+			$total_plugins,
+			$counts['reused'],
+			$counts['extracted'],
+			$counts['failed']
+		);
+	}
+}
+
+echo "\nDone.\n";
+echo "  Plugins scanned: {$total_plugins}\n";
+echo "  Plugins updated: {$plugins_updated}" . ( $opts['dry-run'] ? ' (would update)' : '' ) . "\n";
+echo "  Assets with cached dimensions reused: {$counts['reused']}\n";
+echo "  Assets newly extracted:               {$counts['extracted']}\n";
+echo "  Assets failed extraction:             {$counts['failed']}\n";
+
+if ( $failed_files ) {
+	echo "\nFailed assets:\n  " . implode( "\n  ", $failed_files ) . "\n";
+}

wordpress.org/public_html/wp-content/plugins/plugin-directory/cli/class-import.php

@@ -827,6 +827,14 @@ protected function export_and_parse_plugin( $plugin_slug ) {
 			'blueprint'  => 100 * KB_IN_BYTES,
 		);
 
+		// Previously-imported asset metadata, used to skip re-reading any file whose
+		// SVN revision hasn't changed.
+		$prior_assets = array(
+			'screenshot' => get_post_meta( $this->plugin->ID, 'assets_screenshots', true ) ?: array(),
+			'banner'     => get_post_meta( $this->plugin->ID, 'assets_banners',     true ) ?: array(),
+			'icon'       => get_post_meta( $this->plugin->ID, 'assets_icons',       true ) ?: array(),
+		);
+
 		$svn_blueprints_folder = null;
 		$svn_assets_folder = SVN::ls( self::PLUGIN_SVN_BASE . "/{$plugin_slug}/assets/", true /* verbose */ );
 		if ( $svn_assets_folder ) { // /assets/ may not exist.
@@ -862,7 +870,15 @@ protected function export_and_parse_plugin( $plugin_slug ) {
 					$resolution = preg_replace( '/[^0-9]/u', 'x', $resolution );
 				}
 
-				$assets[ $type ][ $asset['filename'] ] = compact( 'filename', 'revision', 'resolution', 'location', 'locale' );
+				$record = compact( 'filename', 'revision', 'resolution', 'location', 'locale' );
+
+				$record = self::enrich_asset_dimensions(
+					$record,
+					$prior_assets[ $type ][ $filename ] ?? null,
+					$this->plugin
+				);
+
+				$assets[ $type ][ $asset['filename'] ] = $record;
 			}
 		}
 
@@ -921,12 +937,21 @@ protected function export_and_parse_plugin( $plugin_slug ) {
 				continue;
 			}
 
-			$assets['screenshot'][ $filename ] = array(
+			$record = array(
 				'filename'   => $filename,
 				'revision'   => $svn_export['revision'],
 				'resolution' => $screenshot_id,
 				'location'   => 'plugin',
 			);
+
+			$record = self::enrich_asset_dimensions(
+				$record,
+				$prior_assets['screenshot'][ $filename ] ?? null,
+				$this->plugin,
+				$plugin_screenshot
+			);
+
+			$assets['screenshot'][ $filename ] = $record;
 		}
 
 		if ( 'trunk' === $stable_tag ) {
@@ -1068,6 +1093,84 @@ function( $block ) {
 		);
 	}
 
+	/**
+	 * Populate `width` and `height` on an asset record, reusing the prior
+	 * import's values when the SVN revision hasn't changed.
+	 *
+	 * @param array       $record The asset record.
+	 * @param array|null  $prior  Matching record from the prior import.
+	 * @param \WP_Post    $post   The plugin post.
+	 * @param string|null $local  Optional local path to read instead of fetching from SVN.
+	 * @return array
+	 */
+	public static function enrich_asset_dimensions( $record, $prior, $post, $local = null ) {
+		if (
+			is_array( $prior ) &&
+			isset( $prior['revision'], $prior['width'], $prior['height'] ) &&
+			(string) $prior['revision'] === (string) $record['revision'] &&
+			$prior['width'] > 0 && $prior['height'] > 0
+		) {
+			$record['width']  = (int) $prior['width'];
+			$record['height'] = (int) $prior['height'];
+
+			return $record;
+		}
+
+		$size = false;
+
+		if ( $local && file_exists( $local ) ) {
+			$size = wp_getimagesize( $local );
+		}
+
+		if ( ! $size ) {
+			$url       = Template::get_asset_url( $post, $record, false /* no CDN */ );
+			$temp_file = wp_tempnam( $record['filename'] );
+
+			// Range the first read to 128 KB — enough for the headers of
+			// most images. Fall back to a full read only when the prefix
+			// isn't enough to decode the header — the falsy `$size` at
+			// the bottom of the loop is the implicit retry. Transport
+			// errors / non-2xx intentionally bail out via `break`: those
+			// failure modes won't be helped by re-requesting the same
+			// URL without Range.
+			foreach ( array( 128 * KB_IN_BYTES, 0 ) as $limit ) {
+				$args = array(
+					'timeout'  => 15,
+					'stream'   => true,
+					'filename' => $temp_file,
+				);
+				if ( $limit > 0 ) {
+					$args['headers']             = array( 'Range' => 'bytes=0-' . ( $limit - 1 ) );
+					$args['limit_response_size'] = $limit;
+				}
+
+				$response = wp_safe_remote_get( $url, $args );
+				$code     = wp_remote_retrieve_response_code( $response );
+				if ( is_wp_error( $response ) || ( 200 !== $code && 206 !== $code ) ) {
+					break;
+				}
+
+				if ( ! file_exists( $temp_file ) || 0 === filesize( $temp_file ) ) {
+					break;
+				}
+
+				$size = wp_getimagesize( $temp_file );
+				if ( $size ) {
+					break;
+				}
+			}
+
+			unlink( $temp_file );
+		}
+
+		if ( $size && ! empty( $size[0] ) && ! empty( $size[1] ) ) {
+			$record['width']  = (int) $size[0];
+			$record['height'] = (int) $size[1];
+		}
+
+		return $record;
+	}
+
 	/**
 	 * Find the plugin readme file.
 	 *