perf: build lower layer squashes only when a scan reads them

The image source provider now reads images with `image.WithLazyLayerSquashes`: the image squash is built during the read, and each lower layer's squash only when something reads it. A squashed scan never does, so it skips a full file tree copy per layer.

- squashed scans: faster, much less memory on images with many layers
- `scan`/`attest` with all-layers or deep-squashed stay eager (an internal ctx flag), since eager squashing overlaps with fetching and indexing and is faster when every layer gets read
- library callers going through `syft.GetSource` get lazy for every scope, since the scope isn't known when the image is read. All-layers is still correct, just slower
- no public API change

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>

cmd/syft/internal/commands/scan.go

@@ -260,6 +260,11 @@ func getSource(ctx context.Context, opts *options.Catalog, userInput string, sou
 		cfg = cfg.WithDigestAlgorithms(hashers...)
 	}
 
+	if source.ParseScope(opts.Scope) != source.SquashedScope {
+		// every other scope reads each layer's squash, which is cheaper to build while the image is read
+		ctx = internal.WithEagerLayerSquashes(ctx)
+	}
+
 	src, err := syft.GetSource(ctx, userInput, cfg)
 	if err != nil {
 		return nil, fmt.Errorf("could not determine source: %w", err)

internal/layer_squashes.go

@@ -0,0 +1,17 @@
+package internal
+
+import "context"
+
+type eagerLayerSquashesKey struct{}
+
+// WithEagerLayerSquashes has the image source provider build every layer's squashed tree while it reads the
+// image, instead of the first time each one is used. Set it when the scan will read lower layers (any scope but
+// squashed): eager squashing overlaps with fetching and indexing the layers, so those scans finish sooner.
+func WithEagerLayerSquashes(ctx context.Context) context.Context {
+	return context.WithValue(ctx, eagerLayerSquashesKey{}, true)
+}
+
+func EagerLayerSquashes(ctx context.Context) bool {
+	v, _ := ctx.Value(eagerLayerSquashesKey{}).(bool)
+	return v
+}

syft/source/stereoscopesource/image_source_provider.go

@@ -6,6 +6,7 @@ import (
 	"github.com/anchore/go-collections"
 	"github.com/anchore/stereoscope"
 	"github.com/anchore/stereoscope/pkg/image"
+	"github.com/anchore/syft/internal"
 	"github.com/anchore/syft/syft/source"
 )
 
@@ -29,6 +30,10 @@ func (l stereoscopeImageSourceProvider) Name() string {
 }
 
 func (l stereoscopeImageSourceProvider) Provide(ctx context.Context) (source.Source, error) {
+	if !internal.EagerLayerSquashes(ctx) {
+		// lower layer squashes are only built if something reads them, so a squashed scan never pays for them
+		ctx = image.WithLazyLayerSquashes(ctx)
+	}
 	img, err := l.stereoscopeProvider.Provide(ctx)
 	if err != nil {
 		return nil, err

syft/source/stereoscopesource/image_source_provider_test.go

@@ -0,0 +1,76 @@
+package stereoscopesource
+
+import (
+	"context"
+	"path/filepath"
+	"testing"
+
+	"github.com/google/go-containerregistry/pkg/name"
+	"github.com/google/go-containerregistry/pkg/v1/random"
+	"github.com/google/go-containerregistry/pkg/v1/tarball"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/anchore/stereoscope"
+	"github.com/anchore/stereoscope/pkg/filetree"
+	"github.com/anchore/stereoscope/pkg/image"
+	"github.com/anchore/syft/internal"
+	"github.com/anchore/syft/syft/source"
+)
+
+func Test_stereoscopeImageSourceProvider_layerSquashes(t *testing.T) {
+	img, err := random.Image(1024, 3)
+	require.NoError(t, err)
+	tag, err := name.NewTag("localhost/layer-squashes:latest")
+	require.NoError(t, err)
+	archive := filepath.Join(t.TempDir(), "image.tar")
+	require.NoError(t, tarball.WriteToFile(archive, tag, img))
+
+	provide := func(t *testing.T, ctx context.Context) *stereoscopeImageSource {
+		providers := Providers(ProviderConfig{
+			StereoscopeImageProviderConfig: stereoscope.ImageProviderConfig{UserInput: archive},
+		})
+		for _, p := range providers {
+			if p.Value.Name() != image.DockerTarballSource {
+				continue
+			}
+			src, err := p.Value.Provide(ctx)
+			require.NoError(t, err)
+			t.Cleanup(func() { require.NoError(t, src.Close()) })
+			return src.(*stereoscopeImageSource)
+		}
+		t.Fatalf("no %s provider", image.DockerTarballSource)
+		return nil
+	}
+
+	lazy := provide(t, context.Background())
+	eager := provide(t, internal.WithEagerLayerSquashes(context.Background()))
+
+	// lower layers are built on first use unless eager was asked for, the top is the image squash either way
+	require.Len(t, lazy.image.Layers, 3)
+	for idx := range lazy.image.Layers {
+		_, isTree := lazy.image.Layers[idx].SquashedTree.(*filetree.FileTree)
+		assert.Equalf(t, idx == 2, isTree, "lazy layer %d", idx)
+		assert.IsTypef(t, &filetree.FileTree{}, eager.image.Layers[idx].SquashedTree, "eager layer %d", idx)
+	}
+
+	// every scope sees the same files either way
+	for _, scope := range []source.Scope{source.SquashedScope, source.AllLayersScope, source.DeepSquashedScope} {
+		t.Run(string(scope), func(t *testing.T) {
+			files := func(src *stereoscopeImageSource) []string {
+				res, err := src.FileResolver(scope)
+				require.NoError(t, err)
+				locs, err := res.FilesByGlob("**/*")
+				require.NoError(t, err)
+				var out []string
+				for _, l := range locs {
+					out = append(out, l.RealPath+"@"+l.FileSystemID)
+				}
+				return out
+			}
+			want := files(eager)
+			require.NotEmpty(t, want)
+			assert.ElementsMatch(t, want, files(lazy))
+		})
+	}
+}