feat(sync): clone missing repos before pulling updates

Author: vredchenko

packages/smartem-workspace/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "smartem-workspace"
-version = "0.4.0"
+version = "0.5.0"
 description = "CLI tool to automate SmartEM multi-repo workspace setup"
 readme = "README.md"
 license = "Apache-2.0"

packages/smartem-workspace/smartem_workspace/__init__.py

@@ -1,3 +1,3 @@
 """SmartEM workspace setup CLI tool."""
 
-__version__ = "0.4.0"
+__version__ = "0.5.0"

packages/smartem-workspace/smartem_workspace/cli.py

@@ -259,12 +259,20 @@ def sync(
         bool,
         typer.Option("--dry-run", "-n", help="Show what would be done without making changes"),
     ] = False,
+    git_ssh: Annotated[
+        bool,
+        typer.Option("--git-ssh", help="Force SSH URLs for cloning (default: auto-detect)"),
+    ] = False,
+    git_https: Annotated[
+        bool,
+        typer.Option("--git-https", help="Force HTTPS URLs for cloning (default: auto-detect)"),
+    ] = False,
     path: Annotated[
         Path | None,
         typer.Option("--path", "-p", help="Workspace path (auto-detected if not specified)"),
     ] = None,
 ) -> None:
-    """Pull latest changes from all cloned repositories."""
+    """Sync workspace: clone missing repos and pull updates for existing ones."""
     out = get_console()
     workspace_path = path or find_workspace_root()
     if workspace_path is None:
@@ -276,11 +284,17 @@ def sync(
         out.print("[red]Failed to load configuration[/red]")
         raise typer.Exit(1)
 
+    if git_ssh and git_https:
+        out.print("[red]Cannot specify both --git-ssh and --git-https[/red]")
+        raise typer.Exit(1)
+
+    use_ssh: bool | None = True if git_ssh else (False if git_https else None)
+
     out.print("[bold blue]SmartEM Workspace Sync[/bold blue]")
     out.print(f"Workspace: {workspace_path}")
 
-    results = sync_all_repos(workspace_path, config, dry_run=dry_run)
-    print_sync_results(results)
+    results = sync_all_repos(workspace_path, config, out, dry_run=dry_run, use_ssh=use_ssh)
+    print_sync_results(results, out)
 
     errors = sum(1 for r in results if r.status == "error")
     if errors:

packages/smartem-workspace/smartem_workspace/commands/sync.py

@@ -7,24 +7,23 @@
 from rich.console import Console
 from rich.progress import Progress, SpinnerColumn, TextColumn
 
-from smartem_workspace.config.schema import ReposConfig
-from smartem_workspace.setup.repos import get_local_dir
+from smartem_workspace.config.schema import Organization, ReposConfig, Repository
+from smartem_workspace.setup.repos import clone_repo, get_local_dir
 from smartem_workspace.utils.git import (
+    check_github_ssh_auth,
     fetch_remote,
     get_commits_behind,
     get_current_branch,
     has_uncommitted_changes,
     run_git_command,
 )
 
-console = Console()
-
 
 @dataclass
 class SyncResult:
     repo_name: str
     org_name: str
-    status: Literal["updated", "up-to-date", "error", "skipped", "dry-run"]
+    status: Literal["updated", "up-to-date", "error", "skipped", "dry-run", "cloned"]
     message: str
     commits_behind: int = 0
 
@@ -69,41 +68,80 @@ def sync_single_repo(repo_path: Path, dry_run: bool = False) -> SyncResult:
 def sync_all_repos(
     workspace_path: Path,
     config: ReposConfig,
+    console: Console,
     dry_run: bool = False,
+    use_ssh: bool | None = None,
 ) -> list[SyncResult]:
     repos_dir = workspace_path / "repos"
-    results = []
+    results: list[SyncResult] = []
 
     if not repos_dir.exists():
-        console.print("[red]repos directory not found[/red]")
-        return results
+        repos_dir.mkdir(parents=True, exist_ok=True)
+
+    missing_repos: list[tuple[Organization, Repository, Path]] = []
+    existing_repos: list[tuple[str, str, Path]] = []
 
-    repo_paths = []
     for org in config.organizations:
         local_dir = get_local_dir(org)
         org_dir = repos_dir / local_dir
 
         for repo in org.repos:
             repo_path = org_dir / repo.name
             if repo_path.exists():
-                repo_paths.append((org.name, repo.name, repo_path))
-
-    if not repo_paths:
-        console.print("[yellow]No cloned repositories found[/yellow]")
+                existing_repos.append((org.name, repo.name, repo_path))
+            else:
+                missing_repos.append((org, repo, repo_path))
+
+    if missing_repos:
+        github_ssh_ok: bool | None = None
+        if use_ssh is None:
+            has_github = any(org.provider == "github" for org, _, _ in missing_repos)
+            if has_github:
+                github_ssh_ok = check_github_ssh_auth()
+
+        action = "Would clone" if dry_run else "Cloning"
+        console.print(f"\n[bold]{action} {len(missing_repos)} missing repositories...[/bold]\n")
+
+        with Progress(
+            SpinnerColumn(),
+            TextColumn("[progress.description]{task.description}"),
+            console=console,
+            transient=True,
+        ) as progress:
+            task = progress.add_task("Starting...", total=len(missing_repos))
+
+            for org, repo, repo_path in missing_repos:
+                progress.update(task, description=f"{org.name}/{repo.name}")
+
+                if dry_run:
+                    results.append(SyncResult(repo.name, org.name, "dry-run", "Would clone"))
+                else:
+                    success = clone_repo(repo, org, repos_dir, use_ssh, github_ssh_ok)
+                    if success:
+                        results.append(SyncResult(repo.name, org.name, "cloned", "Cloned successfully"))
+                        existing_repos.append((org.name, repo.name, repo_path))
+                    else:
+                        results.append(SyncResult(repo.name, org.name, "error", "Clone failed"))
+
+                progress.advance(task)
+
+    if not existing_repos:
+        if not missing_repos:
+            console.print("[yellow]No repositories configured[/yellow]")
         return results
 
     action = "Checking" if dry_run else "Syncing"
-    console.print(f"\n[bold]{action} {len(repo_paths)} repositories...[/bold]\n")
+    console.print(f"\n[bold]{action} {len(existing_repos)} repositories...[/bold]\n")
 
     with Progress(
         SpinnerColumn(),
         TextColumn("[progress.description]{task.description}"),
         console=console,
         transient=True,
     ) as progress:
-        task = progress.add_task("Starting...", total=len(repo_paths))
+        task = progress.add_task("Starting...", total=len(existing_repos))
 
-        for org_name, repo_name, repo_path in repo_paths:
+        for org_name, repo_name, repo_path in existing_repos:
             progress.update(task, description=f"{org_name}/{repo_name}")
             result = sync_single_repo(repo_path, dry_run=dry_run)
             results.append(result)
@@ -112,7 +150,8 @@ def sync_all_repos(
     return results
 
 
-def print_sync_results(results: list[SyncResult]) -> None:
+def print_sync_results(results: list[SyncResult], console: Console) -> None:
+    cloned = sum(1 for r in results if r.status == "cloned")
     updated = sum(1 for r in results if r.status == "updated")
     up_to_date = sum(1 for r in results if r.status == "up-to-date")
     skipped = sum(1 for r in results if r.status == "skipped")
@@ -122,7 +161,9 @@ def print_sync_results(results: list[SyncResult]) -> None:
     for result in results:
         full_name = f"{result.org_name}/{result.repo_name}"
 
-        if result.status == "updated":
+        if result.status == "cloned":
+            console.print(f"  [green]+[/green] {full_name}: {result.message}")
+        elif result.status == "updated":
             console.print(f"  [green]\u2713[/green] {full_name}: {result.message}")
         elif result.status == "up-to-date":
             console.print(f"  [dim]\u2713 {full_name}: {result.message}[/dim]")
@@ -135,6 +176,8 @@ def print_sync_results(results: list[SyncResult]) -> None:
 
     console.print()
     parts = []
+    if cloned:
+        parts.append(f"[green]{cloned} cloned[/green]")
     if updated:
         parts.append(f"[green]{updated} updated[/green]")
     if would_update: