Performance improvements around package ID (#698)

* set package ID in catalogers and improve hashing performance

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

* update setting ID + tests

Signed-off-by: Alex Goodman <alex.goodman@anchore.com>

syft/source/coordinates.go

@@ -27,7 +27,7 @@ func NewCoordinateSet(start ...Coordinates) CoordinateSet {
 }
 
 func (c Coordinates) ID() artifact.ID {
-	f, err := artifact.IDFromHash(c)
+	f, err := artifact.IDByHash(c)
 	if err != nil {
 		// TODO: what to do in this case?
 		log.Warnf("unable to get fingerprint of location coordinate=%+v: %+v", c, err)

syft/source/location.go

@@ -3,21 +3,19 @@ package source
 import (
 	"fmt"
 
-	"github.com/mitchellh/hashstructure/v2"
-
 	"github.com/anchore/stereoscope/pkg/file"
 	"github.com/anchore/stereoscope/pkg/image"
 	"github.com/anchore/syft/internal/log"
 )
 
-var _ hashstructure.Hashable = (*Location)(nil)
-
 // Location represents a path relative to a particular filesystem resolved to a specific file.Reference. This struct is used as a key
 // in content fetching to uniquely identify a file relative to a request (the VirtualPath).
 type Location struct {
 	Coordinates
-	VirtualPath string         // The path to the file which may or may not have hardlinks / symlinks
-	ref         file.Reference // The file reference relative to the stereoscope.FileCatalog that has more information about this location.
+	// note: it is IMPORTANT to ignore anything but the coordinates for a Location when considering the ID (hash value)
+	// since the coordinates are the minimally correct ID for a location (symlinks should not come into play)
+	VirtualPath string         `hash:"ignore"` // The path to the file which may or may not have hardlinks / symlinks
+	ref         file.Reference `hash:"ignore"` // The file reference relative to the stereoscope.FileCatalog that has more information about this location.
 }
 
 // NewLocation creates a new Location representing a path without denoting a filesystem or FileCatalog reference.
@@ -97,9 +95,3 @@ func (l Location) String() string {
 	}
 	return fmt.Sprintf("Location<%s>", str)
 }
-
-func (l Location) Hash() (uint64, error) {
-	// since location is part of the package definition it is important that only coordinates are used during object
-	// hashing. (Location hash should be a pass-through for the coordinates and not include ref or VirtualPath.)
-	return hashstructure.Hash(l.ID(), hashstructure.FormatV2, nil)
-}

syft/source/location_test.go

@@ -0,0 +1,47 @@
+package source
+
+import (
+	"testing"
+
+	"github.com/anchore/stereoscope/pkg/file"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestLocation_ID(t *testing.T) {
+	tests := []struct {
+		name        string
+		coordinates Coordinates
+		virtualPath string
+		ref         file.Reference
+	}{
+		{
+			name: "coordinates should match location hash",
+			coordinates: Coordinates{
+				RealPath:     "path!",
+				FileSystemID: "filesystem!",
+			},
+		},
+		{
+			name: "coordinates should match location hash (with extra fields)",
+			coordinates: Coordinates{
+				RealPath:     "path!",
+				FileSystemID: "filesystem!",
+			},
+			virtualPath: "virtualPath!",
+			ref: file.Reference{
+				RealPath: "other-real-path!",
+			},
+		},
+	}
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			l := Location{
+				Coordinates: test.coordinates,
+				VirtualPath: test.virtualPath,
+				ref:         test.ref,
+			}
+			assert.Equal(t, l.ID(), test.coordinates.ID())
+		})
+	}
+
+}