Fix canonical Data Liberation bundle verification

.buildkite/commands/run-data-liberation-checks.sh

@@ -18,6 +18,10 @@ bash .buildkite/commands/install-node-dependencies.sh
 # show up as untracked/deleted files, not just modified ones — check the
 # whole dist/ status, not only the diff.
 echo '--- :package: Verify data-liberation plugin bundles are fresh'
+# The vendored plugin commits bundles generated from its own lockfile. Studio's
+# workspace can hoist compatible but different transitive dependencies, which
+# changes otherwise canonical esbuild output.
+npm ci --prefix packages/data-liberation-agent --workspaces=false --ignore-scripts --no-audit --no-fund
 npm -w data-liberation run build:mcp-bundle
 if [[ -n "$(git status --porcelain -- packages/data-liberation-agent/dist)" ]]; then
   git status --porcelain -- packages/data-liberation-agent/dist
@@ -32,7 +36,7 @@ fi
 # missing from SKILL_DRIVERS would fail only at plugin-user runtime — catch
 # the drift here instead.
 echo '--- :mag: Verify skill-invoked drivers have committed bundles'
-missing_drivers=$(grep -rhoE 'run\.mjs [a-z0-9_-]+' packages/data-liberation-agent/skills packages/data-liberation-agent/commands 2>/dev/null | awk '{print $2}' | sort -u | while read -r name; do
+missing_drivers=$({ grep -rhoE 'run\.mjs [a-z0-9_-]+' packages/data-liberation-agent/skills packages/data-liberation-agent/commands 2>/dev/null || true; } | awk '{print $2}' | sort -u | while read -r name; do
   [[ -f "packages/data-liberation-agent/dist/scripts/${name}.mjs" ]] || echo "$name"
 done)
 if [[ -n "$missing_drivers" ]]; then

apps/cli/commands/site/create.ts

@@ -72,6 +72,7 @@ import {
 } from 'cli/lib/cli-config/core';
 import { removeSiteFromConfig } from 'cli/lib/cli-config/sites';
 import { connectToDaemon, disconnectFromDaemon, emitCliEvent } from 'cli/lib/daemon-client';
+import { liberateWebsite } from 'cli/lib/data-liberation-client';
 import {
 	getAiInstructionsPath,
 	getWordPressVersionPath,
@@ -88,7 +89,10 @@ import { getPreferredSiteLanguage } from 'cli/lib/site-language';
 import { generateSiteName } from 'cli/lib/site-name';
 import { getDefaultSitePath } from 'cli/lib/site-paths';
 import { logSiteDetails, openSiteInBrowser, setupCustomDomain } from 'cli/lib/site-utils';
-import { keepSqliteIntegrationUpdated } from 'cli/lib/sqlite-integration';
+import {
+	isSqliteIntegrationAvailable,
+	keepSqliteIntegrationUpdated,
+} from 'cli/lib/sqlite-integration';
 import { getTracksOrigin, recordTracksEvent, TRACKS_EVENTS } from 'cli/lib/tracks';
 import { StatsGroup } from 'cli/lib/types/bump-stats';
 import { untildify } from 'cli/lib/utils';
@@ -517,7 +521,71 @@ function staticSiteImportReceiptError( receipt: Record< string, unknown > | unde
 	}
 	const code = ( error as Record< string, unknown > ).code;
 	const message = ( error as Record< string, unknown > ).message;
-	return [ code, message ].filter( ( value ) => typeof value === 'string' && value ).join( ': ' );
+	const detail = [ code, message ]
+		.filter( ( value ) => typeof value === 'string' && value )
+		.join( ': ' );
+	if ( code === 'static_site_importer_quality_failed' ) {
+		return sprintf(
+			/* translators: %s: Static Site Importer quality validation detail */
+			__(
+				'Static Site Importer materialized a preview but did not accept it. The site and staged request were preserved: %s'
+			),
+			detail || __( 'Quality validation failed.' )
+		);
+	}
+	return detail;
+}
+
+function staticSiteImportQualityFailure(
+	receipt: Record< string, unknown > | undefined
+): string | undefined {
+	const response = receipt?.response;
+	if ( ! response || typeof response !== 'object' || Array.isArray( response ) ) {
+		return undefined;
+	}
+	const result = ( response as Record< string, unknown > ).result;
+	const importResult =
+		result && typeof result === 'object' && ! Array.isArray( result )
+			? ( result as Record< string, unknown > )
+			: ( response as Record< string, unknown > );
+	const validation = importResult.import_validation_result;
+	const summary = importResult.import_report_summary;
+	const quality =
+		validation && typeof validation === 'object' && ! Array.isArray( validation )
+			? validation
+			: summary && typeof summary === 'object' && ! Array.isArray( summary )
+			? summary
+			: undefined;
+	if ( ! quality ) {
+		return undefined;
+	}
+	const counts = ( quality as Record< string, unknown > ).counts;
+	const {
+		status,
+		quality_pass: qualityPass,
+		fail_import: failImport,
+		fallback_count: fallbackCount,
+	} = quality as Record< string, unknown >;
+	if ( status !== 'failed' && qualityPass !== false && failImport !== true ) {
+		return undefined;
+	}
+	const fallbackBlocks =
+		counts && typeof counts === 'object' && ! Array.isArray( counts )
+			? ( counts as Record< string, unknown > ).fallback_blocks
+			: fallbackCount;
+	const detail =
+		typeof fallbackBlocks === 'number'
+			? sprintf(
+					/* translators: %d: number of fallback blocks */
+					__( 'SSI reported %d fallback blocks.' ),
+					fallbackBlocks
+			  )
+			: __( 'SSI rejected the imported content.' );
+	return sprintf(
+		/* translators: %s: Static Site Importer validation detail */
+		__( '%s Review the importer diagnostics and retry.' ),
+		detail
+	);
 }
 
 async function runStaticSiteImport(
@@ -583,6 +651,13 @@ async function runStaticSiteImport(
 			new Error( receiptError || stdout.trim() || __( 'The importer did not return a receipt.' ) )
 		);
 	}
+	const qualityFailure = staticSiteImportQualityFailure( receipt );
+	if ( qualityFailure ) {
+		throw new LoggerError(
+			__( 'Static site import failed quality validation' ),
+			new Error( qualityFailure )
+		);
+	}
 
 	const finalizationStartedAt = Date.now();
 	logger.reportProgress( staticSiteImportProgressMessage( 'finalization', 0 ) );
@@ -1111,7 +1186,10 @@ function coerceSiteId( value: string ) {
 	return value;
 }
 
-export const registerCommand = ( yargs: StudioArgv ) => {
+export const registerCommand = (
+	yargs: StudioArgv,
+	dependencies: { liberate?: typeof liberateWebsite } = {}
+) => {
 	return yargs.command( {
 		command: 'create',
 		describe: __( 'Create a new site' ),
@@ -1463,11 +1541,41 @@ export const registerCommand = ( yargs: StudioArgv ) => {
 			};
 
 			try {
-				const importSource = argv.from;
-				// Remote URLs are rendered into a local source by Data Liberation before they
-				// reach SSI; until that path exists here, `resolveStaticSiteImporterSource`
-				// rejects them. `sourceUrl` still carries provenance for local captures.
+				let importSource = argv.from;
 				const sourceUrl = importSource && isUrl( importSource ) ? importSource : undefined;
+				let liberationOutputDir: string | undefined;
+				if ( sourceUrl ) {
+					if ( ! ( await isSqliteIntegrationAvailable() ) ) {
+						throw new LoggerError(
+							__(
+								'Cannot set up WordPress. Bundled SQLite integration files not found. Please reinstall Studio.'
+							)
+						);
+					}
+					let lastProgressAt = 0;
+					liberationOutputDir = path.join(
+						path.dirname( sitePath ),
+						`${ path.basename( sitePath ) }-source`
+					);
+					defaultLogger.reportStart(
+						LoggerAction.IMPORT_SITE,
+						__( 'Preparing source website with Data Liberation…' )
+					);
+					importSource = await ( dependencies.liberate ?? liberateWebsite )(
+						sourceUrl,
+						liberationOutputDir,
+						{
+							onProgress: ( message ) => {
+								const now = Date.now();
+								if ( now - lastProgressAt >= STATIC_SITE_IMPORT_PROGRESS_INTERVAL_MS ) {
+									lastProgressAt = now;
+									defaultLogger.reportProgress( message );
+								}
+							},
+						}
+					);
+					defaultLogger.reportSuccess( __( 'Source website prepared' ) );
+				}
 
 				if ( importSource ) {
 					config.blueprint = buildCreateFromSourceBlueprint(
@@ -1506,6 +1614,11 @@ export const registerCommand = ( yargs: StudioArgv ) => {
 
 				try {
 					await runCommand( sitePath, config );
+					if ( sourceUrl && liberationOutputDir ) {
+						await fs.promises
+							.rm( liberationOutputDir, { recursive: true, force: true } )
+							.catch( () => {} );
+					}
 				} finally {
 					const bundlePath = config.blueprint?.staticSiteImport?.bundlePath;
 					if ( bundlePath ) {