Route REST URL imports through the canonical import-url ability contract

.github/workflows/test.yml

@@ -28,6 +28,11 @@ jobs:
 
       - uses: Extra-Chill/homeboy-action@31e95050206332c4ebd36894387c980c99443ecf  # v2.8.15
         with:
+          # Pin homeboy core to 0.330.0: core 0.331.0+ (PR #11553) validates
+          # the `test.failures` sidecar as a JSON array, but the wordpress
+          # extension's wp-codebox adapter still emits an object sidecar.
+          # Remove this pin once the extension ships an array sidecar.
+          version: 0.330.0
           commands: review lint
           args: --extension wordpress
           php-version: '8.2'
@@ -49,6 +54,8 @@ jobs:
 
       - uses: Extra-Chill/homeboy-action@31e95050206332c4ebd36894387c980c99443ecf  # v2.8.15
         with:
+          # Pin homeboy core to 0.330.0: see lint step for the rationale.
+          version: 0.330.0
           commands: review test
           args: --extension wordpress
           php-version: ${{ matrix.php-version }}

blocks/importer/view.js

@@ -333,8 +333,10 @@
 			const uploadInputs = root.querySelectorAll( '[data-static-site-importer-source-files], [data-static-site-importer-source-directory]' );
 			const provider = root.getAttribute( 'data-static-site-importer-provider' ) || '';
 			const isCurrentSiteImport = root.getAttribute( 'data-static-site-importer-apply-to-current-site' ) === '1';
+			const initialUrl = form ? form.getAttribute( 'data-static-site-importer-default-url' ) || '' : '';
+			const isUrlOnly = '' !== initialUrl && ! html.value && 0 === uploadInputs.length;
 			const source = {
-				url: form ? form.getAttribute( 'data-static-site-importer-default-url' ) || '' : '',
+				url: initialUrl,
 				html: html ? html.value : '',
 				files: await buildFiles( uploadInputs, root ),
 				archive: await buildArchive( uploadInputs, root ),
@@ -352,34 +354,59 @@
 			try {
 				const restUrl = root.getAttribute( 'data-static-site-importer-rest-url' );
 				const nonce = root.getAttribute( 'data-static-site-importer-nonce' );
-				const response = await fetch( restUrl, {
-					method: 'POST',
-					headers: {
-						'Content-Type': 'application/json',
-						'X-WP-Nonce': nonce,
-					},
-					body: JSON.stringify( {
-						provider,
-						source,
+				const postImport = async function ( importSource, importId ) {
+					const body = {
+						source: importSource,
 						apply_to_current_site: isCurrentSiteImport,
 						activate: isCurrentSiteImport,
 						overwrite: isCurrentSiteImport,
-					} ),
-				} );
-				const report = await response.json();
+					};
+					if ( importId ) {
+						body.source = Object.assign( {}, importSource, { import_id: importId } );
+					}
+					if ( ! isUrlOnly && '' !== provider ) {
+						body.provider = provider;
+					}
+					return fetch( restUrl, {
+						method: 'POST',
+						headers: {
+							'Content-Type': 'application/json',
+							'X-WP-Nonce': nonce,
+						},
+						body: JSON.stringify( body ),
+					} );
+				};
+
+				let response = await postImport( source, '' );
+				let report = await response.json();
+
+				if ( isUrlOnly && response.ok && report.continuation === true ) {
+					const maxContinuations = 64;
+					let steps = 0;
+					while ( response.ok && report.continuation === true && ! report.error && steps < maxContinuations ) {
+						steps++;
+						const nextUrl = report.url || source.url;
+						const nextImportId = report.import_id || '';
+						showStatus( root, 'Continuing URL import (' + steps + ')...' );
+						response = await postImport( { url: nextUrl }, nextImportId );
+						report = await response.json();
+					}
+				}
+
 				setReport( root, report );
-				if ( response.ok && isCurrentSiteImport && report.success ) {
+
+				if ( ! response.ok ) {
+					showStatus( root, ( report.error && report.error.message ) ? report.error.message : 'Import request failed.' );
+				} else if ( isCurrentSiteImport && report.success ) {
 					showStatus( root, 'Import complete.' );
-				} else if ( response.ok && report.success && previewUrl( report ) ) {
+				} else if ( report.success && previewUrl( report ) ) {
 					openPreview( report );
 					showStatus( root, 'WordPress Playground opened.' );
-				} else if ( response.ok && previewUrl( report ) ) {
-					openPreview( report );
-					showStatus( root, 'Preview opened.' );
-				} else if ( response.ok && report.preview && report.preview.status === 'unavailable' ) {
-					showStatus( root, report.preview.message || 'Preview unavailable: WP Codebox did not return a preview URL or Playground blueprint URL.' );
+				} else if ( report.success && report.preview && 'unavailable' === report.preview.status ) {
+					const requirement = report.preview.requires_ability_capable_target;
+					showStatus( root, requirement ? ( report.preview.message || 'URL preview needs a disposable WordPress target that exposes the import-url ability.' ) : ( report.preview.message || 'Preview unavailable: WP Codebox did not return a preview URL or Playground blueprint URL.' ) );
 				} else {
-					showStatus( root, response.ok && report.success ? 'Preview request complete.' : 'Preview request failed.' );
+					showStatus( root, report.success ? 'Preview request complete.' : 'Preview request failed.' );
 				}
 			} catch ( error ) {
 				setReport( root, { success: false, error: { message: error.message } } );

homeboy-test-manifest.json

@@ -32,6 +32,7 @@
     "tests/smoke-template-chrome-dedupe.php": { "environment": "standalone-php" },
     "tests/smoke-template-part-shell-dedupe.php": { "environment": "standalone-php" },
     "tests/smoke-url-import-runtime.php": { "environment": "standalone-php" },
+    "tests/smoke-rest-url-import-helpers.php": { "environment": "standalone-php" },
     "tests/smoke-url-concurrent-fetcher.php": { "environment": "standalone-php" },
     "tests/smoke-validation-runtime-diagnostics.php": { "environment": "standalone-php" },
     "tests/smoke-visual-repair-css.php": { "environment": "standalone-php" },

includes/class-static-site-importer-url-import-runtime.php

@@ -87,35 +87,6 @@ public static function import_url( array $input ) {
 		return Static_Site_Importer_Theme_Generator::import_website_artifact( $runtime['artifact'], $args );
 	}
 
-	/**
-	 * Resolve a URL into a website artifact without importing it.
-	 *
-	 * @param array<string,mixed> $input Ability-style input.
-	 * @return array<string,mixed>|WP_Error Runtime output containing artifact/source_metadata/provider.
-	 */
-	public static function website_artifact_from_url( array $input ) {
-		$url = isset( $input['url'] ) ? Static_Site_Importer_URL_Fetcher::normalize_url( (string) $input['url'] ) : '';
-		if ( '' === $url ) {
-			return new WP_Error( 'static_site_importer_missing_url', 'The url input is required.' );
-		}
-
-		$input['url'] = $url;
-		$request      = self::provider_request( $url, $input );
-		$runtime      = self::resolve_provider( $request );
-		if ( is_wp_error( $runtime ) ) {
-			return $runtime;
-		}
-
-		$artifact = isset( $runtime['artifact'] ) && is_array( $runtime['artifact'] ) ? $runtime['artifact'] : array();
-		if ( empty( $artifact ) ) {
-			return new WP_Error( 'static_site_importer_url_provider_missing_artifact', 'The URL import provider did not return a website artifact.' );
-		}
-
-		$runtime['artifact'] = $artifact;
-
-		return $runtime;
-	}
-
 	/**
 	 * Build a provider request envelope.
 	 *
@@ -133,23 +104,6 @@ private static function provider_request( string $url, array $input ): array {
 		);
 	}
 
-	/**
-	 * Resolve the provider output.
-	 *
-	 * Providers return an array with an `artifact` key containing a website
-	 * artifact, plus optional `source_metadata` and `provider` fields.
-	 *
-	 * @param array<string,mixed> $request Provider request envelope.
-	 * @return array<string,mixed>|WP_Error
-	 */
-	private static function resolve_provider( array $request ) {
-		$provider_output = self::provider_output( $request );
-		if ( is_wp_error( $provider_output ) || is_array( $provider_output ) ) {
-			return $provider_output;
-		}
-		return self::fetch_public_url_provider( $request );
-	}
-
 	/** @return null|array<string,mixed>|WP_Error */
 	private static function provider_output( array $request ) {
 		/**
@@ -165,53 +119,6 @@ private static function provider_output( array $request ) {
 		return apply_filters( 'static_site_importer_url_import_provider', null, $request );
 	}
 
-	/**
-	 * Built-in generic public URL provider.
-	 *
-	 * @param array<string,mixed> $request Provider request envelope.
-	 * @return array<string,mixed>|WP_Error
-	 */
-	private static function fetch_public_url_provider( array $request ) {
-		$provider_args = isset( $request['provider_args'] ) && is_array( $request['provider_args'] ) ? $request['provider_args'] : array();
-		if ( ! empty( $provider_args['collect_site'] ) ) {
-			$provider_args['require_complete_collection'] = true;
-			return Static_Site_Importer_URL_Site_Collector::collect( (string) $request['url'], $provider_args );
-		}
-
-		$fetch = Static_Site_Importer_URL_Fetcher::fetch_to_work_dir(
-			(string) $request['url'],
-			(string) $request['work_dir'],
-			$provider_args
-		);
-		if ( is_wp_error( $fetch ) ) {
-			return $fetch;
-		}
-
-		$html = file_get_contents( $fetch['html_path'] ); // phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents -- Reads the importer-owned fetched HTML artifact.
-		if ( false === $html ) {
-			return new WP_Error( 'static_site_importer_url_artifact_read_failed', 'Failed to read fetched URL HTML.' );
-		}
-		$normalized                    = Static_Site_Importer_Source_Normalizer::normalize_html( $html, (string) $request['url'], $provider_args );
-		$html                          = $normalized['html'];
-		$metadata                      = $fetch['metadata'];
-		$metadata['source_exclusions'] = $normalized['exclusions'];
-		$metadata['diagnostics']       = array_merge( is_array( $metadata['diagnostics'] ?? null ) ? $metadata['diagnostics'] : array(), $normalized['diagnostics'] );
-
-		return array(
-			'provider'        => 'public-url-fetcher',
-			'artifact'        => array(
-				'schema' => 'blocks-engine/php-transformer/site-artifact/v1',
-				'files'  => array(
-					array(
-						'path'    => 'website/index.html',
-						'content' => $html,
-					),
-				),
-			),
-			'source_metadata' => $metadata,
-		);
-	}
-
 	/**
 	 * Build import args for the normal website artifact importer.
 	 *