Draft: wire selective sync execution to pull-files/pull-db (guessed Reprint interface)

Author: epeicher

apps/cli/commands/pull-reprint.ts

@@ -670,9 +670,10 @@ export async function runCommand(
 		}
 
 		if ( ! hasPullCompletedStage( studioMetadata, 'completed' ) ) {
-			// Fetch the deferred media/uploads. The selector's media choice is not
-			// applied yet (see runFullPull) — wired into pull-files in the follow-up.
-			if ( hasSkippedFiles( studioMetadata.stateDirectory ) ) {
+			// Fetch the deferred media/uploads unless the user excluded the media
+			// library. Core, plugins, and themes already came down in the
+			// essential-files pass; this second pass is just the media library.
+			if ( ! studioMetadata.skipUploads && hasSkippedFiles( studioMetadata.stateDirectory ) ) {
 				await downloadSkippedFiles(
 					getSiteRuntime( site ),
 					studioMetadata,
@@ -1017,26 +1018,27 @@ function readPullMetadata( metadataPath: string ): PullSessionMetadata | null {
 }
 
 /**
- * Run reprint's composite `pull` command: the whole site-clone
- * pipeline (preflight → files-pull → db-pull → db-apply →
- * flat-docroot → apply-runtime) in a single child process, with
- * reprint owning the stage ordering and, when the prior pull already
- * completed, resetting its own sub-command state for a delta re-pull
- * via prepare_repull().
+ * Run the site-clone pipeline as three high-level reprint commands so the
+ * selective-sync choice maps cleanly onto them:
  *
- * The SQLite target geometry:
- *   - If preflight exposed the remote `wp-content` (contentDir set),
- *     the database lands under `rawDirectory + contentDir`, an
- *     already-mounted host path that flat-docroot later symlinks into
- *     the flattened site.
- *   - Otherwise it falls back to `sitePath/wp-content`.
+ *   1. `pull-files` — files + flattened layout (`--only` restricts to the
+ *      chosen wp-content folders).
+ *   2. `pull-db`    — database download + import. **Skipped entirely** when the
+ *      user excluded the database.
+ *   3. `apply-runtime` — server config, run last so it picks up the DB
+ *      credentials `pull-db` wrote to state.
  *
- * The flattened site (`--flatten-to`) and runtime output
- * (`--output-dir`) directories are mounted up front so the single
- * fork can write them onto the host filesystem.  `ensurePort` must
- * run first so `--new-site-url` points at the local server.
+ * DRAFT — `pull-files`/`pull-db` are high-level commands Reprint is adding to
+ * hide the low-level orchestration (files-index/files-pull/flat-docroot and
+ * db-pull/db-apply). Names, arguments, and which command owns flatten/runtime
+ * are **guesses** (marked `TODO(reprint pull-files/pull-db)`); reconcile once
+ * the real interface lands. Optimized for simplicity and host-agnostic reuse:
+ * Studio just picks files vs db and passes `--only`; the commands own the rest.
  *
- * Advances the pull stage to 'pulled'.
+ * SQLite target geometry and the `--flatten-to`/`--output-dir` mounts carry
+ * over from the previous composite-`pull` call. Advances the stage to 'pulled';
+ * the three commands are individually resumable/idempotent, so a crash between
+ * them safely re-runs the lot on the next pass.
  */
 export async function runFullPull(
 	runtime: SiteRuntime,
@@ -1050,41 +1052,67 @@ export async function runFullPull(
 		? `${ metadata.rawDirectory }${ contentDir }/database/.ht.sqlite`
 		: `${ metadata.sitePath }/wp-content/database/.ht.sqlite`;
 	const reprintRuntime = runtime === SITE_RUNTIME_NATIVE_PHP ? 'nginx-fpm' : 'playground-cli';
+	const onlyArgs = ( metadata.fileOnlyPaths ?? [] ).map( ( onlyPath ) => `--only=${ onlyPath }` );
+	const mounts = [
+		{ hostPath: metadata.sitePath, vfsPath: metadata.sitePath },
+		{ hostPath: metadata.runtimeDirectory, vfsPath: metadata.runtimeDirectory },
+	];
+
+	const runStep = ( progressLabel: string, args: string[] ) =>
+		runReprintCommandUntilComplete(
+			metadata.stateDirectory,
+			metadata.rawDirectory,
+			args,
+			( progress ) => logger.reportProgress( progress ),
+			{ progressLabel, mounts, verboseCommands: verbose, runtime }
+		);
 
 	logger.reportStart( LoggerAction.DOWNLOAD_FILES, __( 'Pulling site…' ) );
-	await runReprintCommandUntilComplete(
-		metadata.stateDirectory,
-		metadata.rawDirectory,
-		[
-			'pull',
+
+	// 1. Files (+ flattened layout). `--only` restricts to the selected folders.
+	// TODO(reprint pull-files): confirm command name; whether it owns flat-docroot;
+	//   and how it expresses "defer uploads" (assumed --filter=essential-files +
+	//   the deferred files-sync pass below).
+	await runStep( __( 'Pulling files' ), [
+		'pull-files',
+		apiUrl,
+		`--secret=${ secret }`,
+		'--filter=essential-files',
+		...onlyArgs,
+		`--flatten-to=${ metadata.sitePath }`,
+		'--no-adaptive',
+		`--state-dir=${ metadata.stateDirectory }`,
+		`--fs-root=${ metadata.rawDirectory }`,
+	] );
+
+	// 2. Database — only when selected. Skipping it leaves the local DB untouched
+	// on a re-pull (apply-runtime keeps the credentials already in state).
+	// TODO(reprint pull-db): confirm command name + target/rewrite args.
+	if ( ! metadata.skipDatabase ) {
+		await runStep( __( 'Pulling database' ), [
+			'pull-db',
 			apiUrl,
 			`--secret=${ secret }`,
-			'--filter=essential-files',
 			'--target-engine=sqlite',
 			`--target-sqlite-path=${ sqlitePath }`,
 			`--new-site-url=${ metadata.localUrl! }`,
-			`--flatten-to=${ metadata.sitePath }`,
-			`--runtime=${ reprintRuntime }`,
-			'--start-runtime=none',
-			`--output-dir=${ metadata.runtimeDirectory }`,
 			'--no-adaptive',
 			`--state-dir=${ metadata.stateDirectory }`,
 			`--fs-root=${ metadata.rawDirectory }`,
-			// NOTE: the interactive selector / `--only` / `--skip-*` choices are
-			// captured in metadata but NOT applied yet — this is a full pull. The
-			// selection is wired into `pull-files`/`pull-db` in the follow-up PR.
-		],
-		( progress ) => logger.reportProgress( progress ),
-		{
-			progressLabel: __( 'Pulling site' ),
-			mounts: [
-				{ hostPath: metadata.sitePath, vfsPath: metadata.sitePath },
-				{ hostPath: metadata.runtimeDirectory, vfsPath: metadata.runtimeDirectory },
-			],
-			verboseCommands: verbose,
-			runtime,
-		}
-	);
+		] );
+	}
+
+	// 3. Runtime config — last, so it embeds the DB credentials pull-db wrote.
+	await runStep( __( 'Preparing runtime' ), [
+		'apply-runtime',
+		'-',
+		`--runtime=${ reprintRuntime }`,
+		`--output-dir=${ metadata.runtimeDirectory }`,
+		`--flat-document-root=${ metadata.sitePath }`,
+		'--no-adaptive',
+		`--state-dir=${ metadata.stateDirectory }`,
+	] );
+
 	logger.reportSuccess( __( 'Site pulled' ) );
 	recordCompletedStage( metadata, 'pulled' );
 }
@@ -1440,7 +1468,7 @@ export function getReprintApiUrlForSite( siteUrl: string ): string {
 }
 
 function buildFilesSyncArgs(
-	metadata: Pick< PullSessionMetadata, 'stateDirectory' | 'rawDirectory' >,
+	metadata: Pick< PullSessionMetadata, 'stateDirectory' | 'rawDirectory' | 'fileOnlyPaths' >,
 	apiUrl: string,
 	secret: string,
 	extraArgs: string[] = []
@@ -1450,6 +1478,10 @@ function buildFilesSyncArgs(
 		apiUrl,
 		`--secret=${ secret }`,
 		...extraArgs,
+		// Carry the same `--only` set as pull-files so the deferred pass stays in
+		// scope (files-sync's index is a union keyed by a fingerprint of the
+		// prefixes and refuses to resume with a different `--only`).
+		...( metadata.fileOnlyPaths ?? [] ).map( ( onlyPath ) => `--only=${ onlyPath }` ),
 		// Per-batch ceiling — one sub-process yields after 30 s and the
 		// client reconnects to continue.  Not a total-time budget; a slow
 		// or high-latency sync just makes more round-trips.  Kept well