When transferring all in-content images, ensure any image URLs in the content are updated to the new image

.wp-env.json

@@ -2,7 +2,8 @@
   "core": null,
   "plugins": [
     ".",
-    "./tests/Basic-Auth-master"
+    "./tests/Basic-Auth-master",
+    "https://downloads.wordpress.org/plugin/classic-editor.zip"
   ],
   "env": {
     "tests": {

includes/classes/DistributorPost.php

@@ -626,6 +626,11 @@ protected function get_media() {
 			$raw_media = $this->parse_media_blocks();
 		} else {
 			$raw_media = get_attached_media( get_allowed_mime_types(), $post_id );
+			// Parse images from post content.
+			$parsed_media = $this->parse_images_from_post_content();
+			foreach ( $parsed_media as $media_post ) {
+				$raw_media[ $media_post->ID ] = $media_post;
+			}
 		}
 
 		$featured_image_id = $this->get_post_thumbnail_id();
@@ -652,6 +657,49 @@ protected function get_media() {
 		return $media_array;
 	}
 
+	/**
+	 * Parse the post's content to obtain media items by image tags.
+	 *
+	 * @return array<WP_Post> Array of media posts.
+	 */
+	protected function parse_images_from_post_content() {
+		$processor = new \WP_HTML_Tag_Processor( $this->post->post_content );
+
+		$media = array();
+		while ( $processor->next_tag( 'img' ) ) {
+			$classes = explode( ' ', $processor->get_attribute( 'class' ) ?? ' ' );
+
+			if ( ! is_array( $classes ) ) {
+				continue;
+			}
+
+			// Filter out classes that are not image classes.
+			$classes = array_filter(
+				$classes,
+				function( $class ) {
+					return strpos( $class, 'wp-image-' ) === 0;
+				}
+			);
+
+			if ( empty( $classes ) ) {
+				continue;
+			}
+
+			$image_id   = (int) str_replace( 'wp-image-', '', current( $classes ) );
+			$media_post = get_post( $image_id );
+
+			if ( empty( $media_post ) ) {
+				continue;
+			}
+
+			$media[ $media_post->ID ] = $media_post;
+		}
+
+		$media = array_filter( $media );
+
+		return $media;
+	}
+
 	/**
 	 * Parse the post's content to obtain media items.
 	 *

includes/classes/InternalConnections/NetworkSiteConnection.php

@@ -180,7 +180,7 @@ public function push( $post, $args = array() ) {
 		 */
 		if ( apply_filters( 'dt_push_post_media', true, $new_post_id, $post_media, $post_id, $args, $this ) ) {
 			Utils\set_media( $new_post_id, $post_media, [ 'use_filesystem' => true ] );
-		};
+		}
 
 		$media_errors = get_transient( 'dt_media_errors_' . $new_post_id );
 

includes/utils.php

@@ -736,6 +736,8 @@ function set_media( $post_id, $media, $args = [] ) {
 		$media = ( false !== $featured_key ) ? array( $media[ $featured_key ] ) : array();
 	}
 
+	$image_urls_to_update = [];
+
 	foreach ( $media as $media_item ) {
 		$args['source_file'] = $media_item['source_file'];
 
@@ -790,6 +792,11 @@ function set_media( $post_id, $media, $args = [] ) {
 			set_meta( $image_id, $media_item['meta'] );
 		}
 
+		// Save the images that we need to try updating in the content.
+		if ( 'featured' !== $settings['media_handling'] ) { // TODO: do we want a new setting for this?
+			$image_urls_to_update[ $image_id ] = $media_item;
+		}
+
 		// Transfer post properties
 		wp_update_post(
 			[
@@ -801,6 +808,11 @@ function set_media( $post_id, $media, $args = [] ) {
 		);
 	}
 
+	// Update image URLs in content if needed.
+	if ( ! empty( $image_urls_to_update ) ) {
+		update_content_image_urls( (int) $post_id, $image_urls_to_update );
+	}
+
 	if ( ! $found_featured_image ) {
 		delete_post_meta( $post_id, '_thumbnail_id' );
 	}
@@ -1039,6 +1051,176 @@ function process_media( $url, $post_id, $args = [] ) {
 	return (int) $result;
 }
 
+/**
+ * Find and update an image tag.
+ *
+ * @param string $content The post content.
+ * @param array  $media_item The old media item details.
+ * @param int    $image_id The new image ID.
+ * @return string
+ */
+function update_image_tag( string $content, array $media_item, int $image_id ) {
+	$processor = new \WP_HTML_Tag_Processor( $content );
+
+	while ( $processor->next_tag( 'img' ) ) {
+		$classes = explode( ' ', $processor->get_attribute( 'class' ) ?? ' ' );
+
+		// Only process the image that matches the old ID.
+		if (
+			! is_array( $classes ) ||
+			! in_array( 'wp-image-' . $media_item['id'], $classes, true )
+		) {
+			continue;
+		}
+
+		// Try to determine the image size from the size class WordPress adds.
+		$image_size   = 'full';
+		$classes      = explode( ' ', $processor->get_attribute( 'class' ) ?? [] );
+		$size_classes = array_filter(
+			$classes,
+			function ( $image_class ) {
+				return false !== strpos( $image_class, 'size-' );
+			}
+		);
+
+		if ( ! empty( $size_classes ) ) {
+			// If an image happens to have multiple size classes, just use the first.
+			$size_class = reset( $size_classes );
+			$image_size = str_replace( 'size-', '', $size_class );
+		}
+
+		$src = wp_get_attachment_image_url( $image_id, $image_size );
+
+		// If the image size can't be found, try to get the full size.
+		if ( ! $src ) {
+			$src = wp_get_attachment_image_url( $image_id, 'full' );
+
+			// If we still don't have an image, skip this block.
+			if ( ! $src ) {
+				continue;
+			}
+		}
+
+		$processor->set_attribute( 'src', $src );
+		$processor->add_class( 'wp-image-' . $image_id );
+		$processor->remove_class( 'wp-image-' . $media_item['id'] );
+		$processor->remove_attribute( 'srcset' );
+		$processor->remove_attribute( 'sizes' );
+	}
+
+	return $processor->get_updated_html();
+}
+
+/**
+ * Find and update an image block.
+ *
+ * @param array $blocks All blocks in a post.
+ * @param array $media_item The old media item details.
+ * @param int   $image_id The new image ID.
+ * @return array
+ */
+function update_image_block( array $blocks, array $media_item, int $image_id ) {
+	// Find and update all image blocks that match the old image ID.
+	foreach ( $blocks as $key => $block ) {
+		// Recurse into inner blocks.
+		if ( ! empty( $block['innerBlocks'] ) ) {
+			$blocks[ $key ]['innerBlocks'] = update_image_block( $block['innerBlocks'], $media_item, $image_id );
+		}
+
+		// If the block is an image block and the ID matches, update the ID and URL.
+		if ( 'core/image' === $block['blockName'] && (int) $media_item['id'] === (int) $block['attrs']['id'] ) {
+			$image_size = $block['attrs']['sizeSlug'] ?? 'full';
+
+			$blocks[ $key ]['attrs']['id'] = $image_id;
+
+			$processor = new \WP_HTML_Tag_Processor( $blocks[ $key ]['innerHTML'] );
+
+			// Use the HTML API to update the image src and class.
+			if ( $processor->next_tag( 'img' ) ) {
+				$src = wp_get_attachment_image_url( $image_id, $image_size );
+
+				// If the image size can't be found, try to get the full size.
+				if ( ! $src ) {
+					$src = wp_get_attachment_image_url( $image_id, 'full' );
+
+					// If we still don't have an image, skip this block.
+					if ( ! $src ) {
+						continue;
+					}
+				}
+
+				$processor->set_attribute( 'src', $src );
+				$processor->add_class( 'wp-image-' . $image_id );
+				$processor->remove_class( 'wp-image-' . $media_item['id'] );
+
+				$blocks[ $key ]['innerHTML']       = $processor->get_updated_html();
+				$blocks[ $key ]['innerContent'][0] = $processor->get_updated_html();
+			}
+		}
+	}
+
+	return $blocks;
+}
+
+/**
+ * Update all old image URLs with the new ones.
+ *
+ * @param int   $post_id The post ID.
+ * @param array $images  The old image details.
+ */
+function update_content_image_urls( int $post_id, array $images ) {
+	$dt_post = new DistributorPost( $post_id );
+
+	if ( ! $dt_post ) {
+		return;
+	}
+
+	/**
+	 * Filter whether image URLS should be updated in the content.
+	 *
+	 * @since x.x.x
+	 * @hook dt_update_content_image_urls
+	 *
+	 * @param {bool}  true     Whether image URLs should be updated. Default `true`.
+	 * @param {int}   $post_id The post ID.
+	 * @param {array} $images  The old image details.
+	 *
+	 * @return {bool} Whether image URLs should be updated.
+	 */
+	if ( ! apply_filters( 'dt_update_content_image_urls', true, $post_id, $images ) ) {
+		return;
+	}
+
+	$content    = $dt_post->post->post_content;
+	$has_blocks = $dt_post->has_blocks();
+
+	foreach ( $images as $image_id => $media_item ) {
+		// Process block and classic editor content differently.
+		if ( $has_blocks ) {
+			$blocks = parse_blocks( $content );
+
+			// Update the image block attributes.
+			$updated_blocks = update_image_block( $blocks, $media_item, $image_id );
+			$content        = serialize_blocks( $updated_blocks );
+		} else {
+			$content = update_image_tag( $content, $media_item, $image_id );
+		}
+	}
+
+	// No need to update if the content wasn't modified.
+	if ( $content === $dt_post->post->post_content ) {
+		return;
+	}
+
+	// Update the post content.
+	wp_update_post(
+		[
+			'ID'           => $post_id,
+			'post_content' => $content,
+		]
+	);
+}
+
 /**
  * Get existing media ID based on the original source URL and original media ID.
  *

tests/bin/initialize.sh

@@ -3,6 +3,7 @@ set -e
 
 wp-env run tests-wordpress chmod -c ugo+w /var/www/html
 wp-env run tests-cli wp rewrite structure '/%postname%/' --hard
+wp-env run tests-cli wp plugin deactivate classic-editor
 
 status=0
 wp-env run tests-cli wp site list || status=$?