Merge pull request #847 from newfold-labs/fix/ai-sitegen-sideload-images

Fix AI sitegen sideload images

Author: wpalani

includes/RestApi/AppController.php

@@ -5,6 +5,13 @@
 use NewfoldLabs\WP\Module\Onboarding\Permissions;
 use NewfoldLabs\WP\Module\Onboarding\Services\AppService;
 
+/**
+ * AppController class for handling onboarding application REST API endpoints.
+ *
+ * This controller manages the REST API routes and handlers for the onboarding
+ * application functionality. It provides endpoints for starting and completing
+ * the onboarding process.
+ */
 class AppController {
 
 	/**
@@ -21,6 +28,13 @@ class AppController {
 	 */
 	protected $rest_base = '/app';
 
+	/**
+	 * Register the REST API routes for the onboarding application.
+	 *
+	 * Registers two main endpoints:
+	 * - /app/start: Initiates the onboarding process
+	 * - /app/complete: Completes the onboarding process with selected homepage
+	 */
 	public function register_routes() {
 		\register_rest_route(
 			$this->namespace,
@@ -74,7 +88,7 @@ public function start(): \WP_REST_Response {
 	 */
 	public function complete( \WP_REST_Request $request ): \WP_REST_Response {
 		$data                      = json_decode( $request->get_body(), true );
-		$selected_sitegen_homepage = $data['selected_sitegen_homepage'];
+		$selected_sitegen_homepage = $data['selected_sitegen_homepage'] ?? false;
 		if ( ! $selected_sitegen_homepage ) {
 			return new \WP_REST_Response(
 				array( 'error' => 'Selected sitegen homepage is required.' ),

includes/Services/SiteGenImageService.php

@@ -21,8 +21,8 @@ class SiteGenImageService {
 	 */
 	public static function process_homepage_images_immediate_async( $post_id, $content ) {
 		// Extract image URLs from content
-		preg_match_all( '/<img[^>]+src=["\']([^"\']+)["\']/i', $content, $matches );
-		$image_urls = isset( $matches[1] ) ? $matches[1] : array();
+		$image_urls = self::extract_all_image_urls( $content );
+
 		if ( empty( $image_urls ) ) {
 			return;
 		}
@@ -37,6 +37,33 @@ public static function process_homepage_images_immediate_async( $post_id, $conte
 		}
 	}
 
+	/**
+	 * Extract all image URLs from content specifically targeting Unsplash and patterns.hiive.cloud domains.
+	 *
+	 * @param string $content The content to extract image URLs from.
+	 * @return array Array of unique image URLs.
+	 */
+	private static function extract_all_image_urls( $content ) {
+		$image_urls = array();
+
+		// Extract Unsplash images
+		preg_match_all( '/https?:\/\/([^\/]+\.)?unsplash\.com\/[^\s"\'<>]+/i', $content, $matches );
+		if ( isset( $matches[0] ) ) {
+			$image_urls = array_merge( $image_urls, $matches[0] );
+		}
+
+		// Extract patterns.hiive.cloud images
+		preg_match_all( '/https?:\/\/patterns\.hiive\.cloud\/[^\s"\'<>]+/i', $content, $matches );
+		if ( isset( $matches[0] ) ) {
+			$image_urls = array_merge( $image_urls, $matches[0] );
+		}
+
+		// Decode HTML entities in URLs to ensure proper replacement
+		$image_urls = array_map( 'html_entity_decode', $image_urls );
+
+		return array_values( array_unique( $image_urls ) );
+	}
+
 	/**
 	 * Uploads images to the WordPress media library as attachments.
 	 *
@@ -57,7 +84,7 @@ public static function upload_images_to_wp_media_library( $image_urls, $post_id
 		self::connect_to_filesystem();
 
 		$uploaded_image_urls = array();
-		$total_images = count( $image_urls );
+		$total_images        = count( $image_urls );
 		$successful_uploads  = 0;
 
 		try {
@@ -177,6 +204,19 @@ public static function update_post_content_with_new_image_urls( $post_id, $url_m
 			if ( ! empty( $new_url ) ) {
 				// Use str_replace for exact URL replacement
 				$new_content = str_replace( $original_url, $new_url, $content );
+
+				// If no replacement happened, try with HTML entity encoded version
+				if ( $new_content === $content ) {
+					$encoded_url = htmlspecialchars( $original_url, ENT_QUOTES | ENT_HTML5, 'UTF-8' );
+					$new_content = str_replace( $encoded_url, $new_url, $content );
+				}
+
+				// If still no replacement, try with double-encoded version (common in WordPress)
+				if ( $new_content === $content ) {
+					$double_encoded_url = htmlspecialchars( htmlspecialchars( $original_url, ENT_QUOTES | ENT_HTML5, 'UTF-8' ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
+					$new_content        = str_replace( $double_encoded_url, $new_url, $content );
+				}
+
 				if ( $new_content !== $content ) {
 					$content = $new_content;
 					$updated = true;