Merge c4301b7a44 into abfd1713e0

## Changes

Users can configure the bundle to not synchronize any files with:
```yaml
sync:
  paths: []
```

If it is explicitly configured as an empty list, the validate command
must not warn about not having any files to synchronize. The warning
exists to alert users who are unintentionally not synchronizing any
files (they might have a `.gitignore` pattern that matches everything).

Closes #1663.

## Tests

* New unit test.

bundle/config/mutator/override_compute.go

@@ -38,18 +38,27 @@ func overrideJobCompute(j *resources.Job, compute string) {
 }
 
 func (m *overrideCompute) Apply(ctx context.Context, b *bundle.Bundle) diag.Diagnostics {
-	if b.Config.Bundle.Mode != config.Development {
+	var diags diag.Diagnostics
+
+	if b.Config.Bundle.Mode == config.Production {
 		if b.Config.Bundle.ClusterId != "" {
-			return diag.Errorf("cannot override compute for an target that does not use 'mode: development'")
+			// Overriding compute via a command-line flag for production works, but is not recommended.
+			diags = diags.Extend(diag.Warningf("overriding compute for a target that uses 'mode: production' is not recommended"))
+		}
+		if env.Get(ctx, "DATABRICKS_CLUSTER_ID") != "" {
+			// The DATABRICKS_CLUSTER_ID may be set by accident when doing a production deploy.
+			// Overriding the cluster in production is almost always a mistake since customers
+			// want consistency in production and not compute that is different each deploy.
+			// For this reason we log a warning and ignore the environment variable.
+			return diag.Warningf("the DATABRICKS_CLUSTER_ID variable is set but is ignored since the current target uses 'mode: production'")
 		}
-		return nil
 	}
 	if v := env.Get(ctx, "DATABRICKS_CLUSTER_ID"); v != "" {
 		b.Config.Bundle.ClusterId = v
 	}
 
 	if b.Config.Bundle.ClusterId == "" {
-		return nil
+		return diags
 	}
 
 	r := b.Config.Resources
@@ -57,5 +66,5 @@ func (m *overrideCompute) Apply(ctx context.Context, b *bundle.Bundle) diag.Diag
 		overrideJobCompute(r.Jobs[i], b.Config.Bundle.ClusterId)
 	}
 
-	return nil
+	return diags
 }

bundle/config/mutator/override_compute_test.go

@@ -14,7 +14,7 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-func TestOverrideDevelopment(t *testing.T) {
+func TestOverrideComputeModeDevelopment(t *testing.T) {
 	t.Setenv("DATABRICKS_CLUSTER_ID", "")
 	b := &bundle.Bundle{
 		Config: config.Root{
@@ -62,10 +62,13 @@ func TestOverrideDevelopment(t *testing.T) {
 	assert.Empty(t, b.Config.Resources.Jobs["job1"].Tasks[3].JobClusterKey)
 }
 
-func TestOverrideDevelopmentEnv(t *testing.T) {
+func TestOverrideComputeModeDefault(t *testing.T) {
 	t.Setenv("DATABRICKS_CLUSTER_ID", "newClusterId")
 	b := &bundle.Bundle{
 		Config: config.Root{
+			Bundle: config.Bundle{
+				Mode: "",
+			},
 			Resources: config.Resources{
 				Jobs: map[string]*resources.Job{
 					"job1": {JobSettings: &jobs.JobSettings{
@@ -86,11 +89,12 @@ func TestOverrideDevelopmentEnv(t *testing.T) {
 
 	m := mutator.OverrideCompute()
 	diags := bundle.Apply(context.Background(), b, m)
-	require.NoError(t, diags.Error())
+	require.Empty(t, diags)
-	assert.Equal(t, "cluster2", b.Config.Resources.Jobs["job1"].Tasks[1].ExistingClusterId)
+	assert.Equal(t, "newClusterId", b.Config.Resources.Jobs["job1"].Tasks[0].ExistingClusterId)
+	assert.Equal(t, "newClusterId", b.Config.Resources.Jobs["job1"].Tasks[1].ExistingClusterId)
 }
 
-func TestOverridePipelineTask(t *testing.T) {
+func TestOverrideComputePipelineTask(t *testing.T) {
 	t.Setenv("DATABRICKS_CLUSTER_ID", "newClusterId")
 	b := &bundle.Bundle{
 		Config: config.Root{
@@ -115,7 +119,7 @@ func TestOverridePipelineTask(t *testing.T) {
 	assert.Empty(t, b.Config.Resources.Jobs["job1"].Tasks[0].ExistingClusterId)
 }
 
-func TestOverrideForEachTask(t *testing.T) {
+func TestOverrideComputeForEachTask(t *testing.T) {
 	t.Setenv("DATABRICKS_CLUSTER_ID", "newClusterId")
 	b := &bundle.Bundle{
 		Config: config.Root{
@@ -140,10 +144,11 @@ func TestOverrideForEachTask(t *testing.T) {
 	assert.Empty(t, b.Config.Resources.Jobs["job1"].Tasks[0].ForEachTask.Task)
 }
 
-func TestOverrideProduction(t *testing.T) {
+func TestOverrideComputeModeProduction(t *testing.T) {
 	b := &bundle.Bundle{
 		Config: config.Root{
 			Bundle: config.Bundle{
+				Mode:      config.Production,
 				ClusterId: "newClusterID",
 			},
 			Resources: config.Resources{
@@ -166,13 +171,18 @@ func TestOverrideProduction(t *testing.T) {
 
 	m := mutator.OverrideCompute()
 	diags := bundle.Apply(context.Background(), b, m)
-	require.True(t, diags.HasError())
+	require.Len(t, diags, 1)
+	assert.Equal(t, "overriding compute for a target that uses 'mode: production' is not recommended", diags[0].Summary)
+	assert.Equal(t, "newClusterID", b.Config.Resources.Jobs["job1"].Tasks[0].ExistingClusterId)
 }
 
-func TestOverrideProductionEnv(t *testing.T) {
+func TestOverrideComputeModeProductionIgnoresVariable(t *testing.T) {
 	t.Setenv("DATABRICKS_CLUSTER_ID", "newClusterId")
 	b := &bundle.Bundle{
 		Config: config.Root{
+			Bundle: config.Bundle{
+				Mode: config.Production,
+			},
 			Resources: config.Resources{
 				Jobs: map[string]*resources.Job{
 					"job1": {JobSettings: &jobs.JobSettings{
@@ -193,5 +203,7 @@ func TestOverrideProductionEnv(t *testing.T) {
 
 	m := mutator.OverrideCompute()
 	diags := bundle.Apply(context.Background(), b, m)
-	require.NoError(t, diags.Error())
+	require.Len(t, diags, 1)
+	assert.Equal(t, "the DATABRICKS_CLUSTER_ID variable is set but is ignored since the current target uses 'mode: production'", diags[0].Summary)
+	assert.Equal(t, "cluster2", b.Config.Resources.Jobs["job1"].Tasks[1].ExistingClusterId)
 }

bundle/config/mutator/translate_paths.go

@@ -126,7 +126,33 @@ func (t *translateContext) rewritePath(
 func (t *translateContext) translateNotebookPath(literal, localFullPath, localRelPath, remotePath string) (string, error) {
 	nb, _, err := notebook.DetectWithFS(t.b.SyncRoot, filepath.ToSlash(localRelPath))
 	if errors.Is(err, fs.ErrNotExist) {
-		return "", fmt.Errorf("notebook %s not found", literal)
+		if filepath.Ext(localFullPath) != notebook.ExtensionNone {
+			return "", fmt.Errorf("notebook %s not found", literal)
+		}
+
+		extensions := []string{
+			notebook.ExtensionPython,
+			notebook.ExtensionR,
+			notebook.ExtensionScala,
+			notebook.ExtensionSql,
+			notebook.ExtensionJupyter,
+		}
+
+		// Check whether a file with a notebook extension already exists. This
+		// way we can provide a more targeted error message.
+		for _, ext := range extensions {
+			literalWithExt := literal + ext
+			localRelPathWithExt := filepath.ToSlash(localRelPath + ext)
+			if _, err := fs.Stat(t.b.SyncRoot, localRelPathWithExt); err == nil {
+				return "", fmt.Errorf(`notebook %s not found. Did you mean %s?
+Local notebook references are expected to contain one of the following
+file extensions: [%s]`, literal, literalWithExt, strings.Join(extensions, ", "))
+			}
+		}
+
+		// Return a generic error message if no matching possible file is found.
+		return "", fmt.Errorf(`notebook %s not found. Local notebook references are expected
+to contain one of the following file extensions: [%s]`, literal, strings.Join(extensions, ", "))
 	}
 	if err != nil {
 		return "", fmt.Errorf("unable to determine if %s is a notebook: %w", localFullPath, err)

bundle/config/mutator/translate_paths_test.go

@@ -2,6 +2,7 @@ package mutator_test
 
 import (
 	"context"
+	"fmt"
 	"os"
 	"path/filepath"
 	"runtime"
@@ -508,6 +509,59 @@ func TestPipelineNotebookDoesNotExistError(t *testing.T) {
 	assert.EqualError(t, diags.Error(), "notebook ./doesnt_exist.py not found")
 }
 
+func TestPipelineNotebookDoesNotExistErrorWithoutExtension(t *testing.T) {
+	for _, ext := range []string{
+		".py",
+		".r",
+		".scala",
+		".sql",
+		".ipynb",
+		"",
+	} {
+		t.Run("case_"+ext, func(t *testing.T) {
+			dir := t.TempDir()
+
+			if ext != "" {
+				touchEmptyFile(t, filepath.Join(dir, "foo"+ext))
+			}
+
+			b := &bundle.Bundle{
+				SyncRootPath: dir,
+				SyncRoot:     vfs.MustNew(dir),
+				Config: config.Root{
+					Resources: config.Resources{
+						Pipelines: map[string]*resources.Pipeline{
+							"pipeline": {
+								PipelineSpec: &pipelines.PipelineSpec{
+									Libraries: []pipelines.PipelineLibrary{
+										{
+											Notebook: &pipelines.NotebookLibrary{
+												Path: "./foo",
+											},
+										},
+									},
+								},
+							},
+						},
+					},
+				},
+			}
+
+			bundletest.SetLocation(b, ".", []dyn.Location{{File: filepath.Join(dir, "fake.yml")}})
+			diags := bundle.Apply(context.Background(), b, mutator.TranslatePaths())
+
+			if ext == "" {
+				assert.EqualError(t, diags.Error(), `notebook ./foo not found. Local notebook references are expected
+to contain one of the following file extensions: [.py, .r, .scala, .sql, .ipynb]`)
+			} else {
+				assert.EqualError(t, diags.Error(), fmt.Sprintf(`notebook ./foo not found. Did you mean ./foo%s?
+Local notebook references are expected to contain one of the following
+file extensions: [.py, .r, .scala, .sql, .ipynb]`, ext))
+			}
+		})
+	}
+}
+
 func TestPipelineFileDoesNotExistError(t *testing.T) {
 	dir := t.TempDir()
 

bundle/config/validate/files_to_sync.go

@@ -21,6 +21,12 @@ func (v *filesToSync) Name() string {
 }
 
 func (v *filesToSync) Apply(ctx context.Context, rb bundle.ReadOnlyBundle) diag.Diagnostics {
+	// The user may be intentional about not synchronizing any files.
+	// In this case, we should not show any warnings.
+	if len(rb.Config().Sync.Paths) == 0 {
+		return nil
+	}
+
 	sync, err := files.GetSync(ctx, rb)
 	if err != nil {
 		return diag.FromErr(err)
@@ -31,6 +37,7 @@ func (v *filesToSync) Apply(ctx context.Context, rb bundle.ReadOnlyBundle) diag.
 		return diag.FromErr(err)
 	}
 
+	// If there are files to sync, we don't need to show any warnings.
 	if len(fl) != 0 {
 		return nil
 	}

bundle/config/validate/files_to_sync_test.go

@@ -0,0 +1,105 @@
+package validate
+
+import (
+	"context"
+	"testing"
+
+	"github.com/databricks/cli/bundle"
+	"github.com/databricks/cli/bundle/config"
+	"github.com/databricks/cli/internal/testutil"
+	"github.com/databricks/cli/libs/diag"
+	"github.com/databricks/cli/libs/vfs"
+	sdkconfig "github.com/databricks/databricks-sdk-go/config"
+	"github.com/databricks/databricks-sdk-go/experimental/mocks"
+	"github.com/databricks/databricks-sdk-go/service/iam"
+	"github.com/databricks/databricks-sdk-go/service/workspace"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/mock"
+	"github.com/stretchr/testify/require"
+)
+
+func TestFilesToSync_NoPaths(t *testing.T) {
+	b := &bundle.Bundle{
+		Config: config.Root{
+			Sync: config.Sync{
+				Paths: []string{},
+			},
+		},
+	}
+
+	ctx := context.Background()
+	rb := bundle.ReadOnly(b)
+	diags := bundle.ApplyReadOnly(ctx, rb, FilesToSync())
+	assert.Empty(t, diags)
+}
+
+func setupBundleForFilesToSyncTest(t *testing.T) *bundle.Bundle {
+	dir := t.TempDir()
+
+	testutil.Touch(t, dir, "file1")
+	testutil.Touch(t, dir, "file2")
+
+	b := &bundle.Bundle{
+		BundleRootPath: dir,
+		BundleRoot:     vfs.MustNew(dir),
+		SyncRootPath:   dir,
+		SyncRoot:       vfs.MustNew(dir),
+		Config: config.Root{
+			Bundle: config.Bundle{
+				Target: "default",
+			},
+			Workspace: config.Workspace{
+				FilePath: "/this/doesnt/matter",
+				CurrentUser: &config.User{
+					User: &iam.User{},
+				},
+			},
+			Sync: config.Sync{
+				// Paths are relative to [SyncRootPath].
+				Paths: []string{"."},
+			},
+		},
+	}
+
+	m := mocks.NewMockWorkspaceClient(t)
+	m.WorkspaceClient.Config = &sdkconfig.Config{
+		Host: "https://foo.com",
+	}
+
+	// The initialization logic in [sync.New] performs a check on the destination path.
+	// Removing this check at initialization time is tbd...
+	m.GetMockWorkspaceAPI().EXPECT().GetStatusByPath(mock.Anything, "/this/doesnt/matter").Return(&workspace.ObjectInfo{
+		ObjectType: workspace.ObjectTypeDirectory,
+	}, nil)
+
+	b.SetWorkpaceClient(m.WorkspaceClient)
+	return b
+}
+
+func TestFilesToSync_EverythingIgnored(t *testing.T) {
+	b := setupBundleForFilesToSyncTest(t)
+
+	// Ignore all files.
+	testutil.WriteFile(t, "*\n.*\n", b.BundleRootPath, ".gitignore")
+
+	ctx := context.Background()
+	rb := bundle.ReadOnly(b)
+	diags := bundle.ApplyReadOnly(ctx, rb, FilesToSync())
+	require.Equal(t, 1, len(diags))
+	assert.Equal(t, diag.Warning, diags[0].Severity)
+	assert.Equal(t, "There are no files to sync, please check your .gitignore", diags[0].Summary)
+}
+
+func TestFilesToSync_EverythingExcluded(t *testing.T) {
+	b := setupBundleForFilesToSyncTest(t)
+
+	// Exclude all files.
+	b.Config.Sync.Exclude = []string{"*"}
+
+	ctx := context.Background()
+	rb := bundle.ReadOnly(b)
+	diags := bundle.ApplyReadOnly(ctx, rb, FilesToSync())
+	require.Equal(t, 1, len(diags))
+	assert.Equal(t, diag.Warning, diags[0].Severity)
+	assert.Equal(t, "There are no files to sync, please check your .gitignore and sync.exclude configuration", diags[0].Summary)
+}

bundle/config/variable/lookup.go

@@ -22,6 +22,8 @@ type Lookup struct {
 
 	Metastore string `json:"metastore,omitempty"`
 
+	NotificationDestination string `json:"notification_destination,omitempty"`
+
 	Pipeline string `json:"pipeline,omitempty"`
 
 	Query string `json:"query,omitempty"`
@@ -63,6 +65,9 @@ func (l *Lookup) constructResolver() (resolver, error) {
 	if l.Metastore != "" {
 		resolvers = append(resolvers, resolveMetastore{name: l.Metastore})
 	}
+	if l.NotificationDestination != "" {
+		resolvers = append(resolvers, resolveNotificationDestination{name: l.NotificationDestination})
+	}
 	if l.Pipeline != "" {
 		resolvers = append(resolvers, resolvePipeline{name: l.Pipeline})
 	}

bundle/config/variable/resolve_notification_destination.go

@@ -0,0 +1,46 @@
+package variable
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/databricks/databricks-sdk-go"
+	"github.com/databricks/databricks-sdk-go/service/settings"
+)
+
+type resolveNotificationDestination struct {
+	name string
+}
+
+func (l resolveNotificationDestination) Resolve(ctx context.Context, w *databricks.WorkspaceClient) (string, error) {
+	result, err := w.NotificationDestinations.ListAll(ctx, settings.ListNotificationDestinationsRequest{
+		// The default page size for this API is 20.
+		// We use a higher value to make fewer API calls.
+		PageSize: 200,
+	})
+	if err != nil {
+		return "", err
+	}
+
+	// Collect all notification destinations with the given name.
+	var entities []settings.ListNotificationDestinationsResult
+	for _, entity := range result {
+		if entity.DisplayName == l.name {
+			entities = append(entities, entity)
+		}
+	}
+
+	// Return the ID of the first matching notification destination.
+	switch len(entities) {
+	case 0:
+		return "", fmt.Errorf("notification destination named %q does not exist", l.name)
+	case 1:
+		return entities[0].Id, nil
+	default:
+		return "", fmt.Errorf("there are %d instances of clusters named %q", len(entities), l.name)
+	}
+}
+
+func (l resolveNotificationDestination) String() string {
+	return fmt.Sprintf("notification-destination: %s", l.name)
+}

bundle/config/variable/resolve_notification_destination_test.go

@@ -0,0 +1,82 @@
+package variable
+
+import (
+	"context"
+	"fmt"
+	"testing"
+
+	"github.com/databricks/databricks-sdk-go/experimental/mocks"
+	"github.com/databricks/databricks-sdk-go/service/settings"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/mock"
+	"github.com/stretchr/testify/require"
+)
+
+func TestResolveNotificationDestination_ResolveSuccess(t *testing.T) {
+	m := mocks.NewMockWorkspaceClient(t)
+
+	api := m.GetMockNotificationDestinationsAPI()
+	api.EXPECT().
+		ListAll(mock.Anything, mock.Anything).
+		Return([]settings.ListNotificationDestinationsResult{
+			{Id: "1234", DisplayName: "destination"},
+		}, nil)
+
+	ctx := context.Background()
+	l := resolveNotificationDestination{name: "destination"}
+	result, err := l.Resolve(ctx, m.WorkspaceClient)
+	require.NoError(t, err)
+	assert.Equal(t, "1234", result)
+}
+
+func TestResolveNotificationDestination_ResolveError(t *testing.T) {
+	m := mocks.NewMockWorkspaceClient(t)
+
+	api := m.GetMockNotificationDestinationsAPI()
+	api.EXPECT().
+		ListAll(mock.Anything, mock.Anything).
+		Return(nil, fmt.Errorf("bad"))
+
+	ctx := context.Background()
+	l := resolveNotificationDestination{name: "destination"}
+	_, err := l.Resolve(ctx, m.WorkspaceClient)
+	assert.ErrorContains(t, err, "bad")
+}
+
+func TestResolveNotificationDestination_ResolveNotFound(t *testing.T) {
+	m := mocks.NewMockWorkspaceClient(t)
+
+	api := m.GetMockNotificationDestinationsAPI()
+	api.EXPECT().
+		ListAll(mock.Anything, mock.Anything).
+		Return([]settings.ListNotificationDestinationsResult{}, nil)
+
+	ctx := context.Background()
+	l := resolveNotificationDestination{name: "destination"}
+	_, err := l.Resolve(ctx, m.WorkspaceClient)
+	require.Error(t, err)
+	assert.ErrorContains(t, err, `notification destination named "destination" does not exist`)
+}
+
+func TestResolveNotificationDestination_ResolveMultiple(t *testing.T) {
+	m := mocks.NewMockWorkspaceClient(t)
+
+	api := m.GetMockNotificationDestinationsAPI()
+	api.EXPECT().
+		ListAll(mock.Anything, mock.Anything).
+		Return([]settings.ListNotificationDestinationsResult{
+			{Id: "1234", DisplayName: "destination"},
+			{Id: "5678", DisplayName: "destination"},
+		}, nil)
+
+	ctx := context.Background()
+	l := resolveNotificationDestination{name: "destination"}
+	_, err := l.Resolve(ctx, m.WorkspaceClient)
+	require.Error(t, err)
+	assert.ErrorContains(t, err, `there are 2 instances of clusters named "destination"`)
+}
+
+func TestResolveNotificationDestination_String(t *testing.T) {
+	l := resolveNotificationDestination{name: "name"}
+	assert.Equal(t, "notification-destination: name", l.String())
+}

internal/init_test.go

@@ -97,7 +97,7 @@ func TestAccBundleInitOnMlopsStacks(t *testing.T) {
 	require.NoError(t, err)
 	job, err := w.Jobs.GetByJobId(context.Background(), batchJobId)
 	assert.NoError(t, err)
-	assert.Equal(t, fmt.Sprintf("dev-%s-batch-inference-job", projectName), job.Settings.Name)
+	assert.Contains(t, job.Settings.Name, fmt.Sprintf("dev-%s-batch-inference-job", projectName))
 }
 
 func TestAccBundleInitHelpers(t *testing.T) {