Add SBOM Discovery with Rekor (#1157)

This PR adds the ability to discover build-time SBOMs from binaries with the Rekor transparency log.
It does this by creating external document references for them in SPDX JSON.

Explained in more detail in syft issue #1159

Signed-off-by: Christopher Phillips <christopher.phillips@anchore.com>

README.md

@@ -177,6 +177,7 @@ registry:yourrepo/yourimage:tag          pull image directly from a registry (no
 
 #### Non Default:
 - cargo-auditable-binary
+- rekor
 
 ### Excluding file paths
 
@@ -663,3 +664,12 @@ The following checks were performed on each of these signatures:
 ```
 
 Consumers of your image can now trust that the SBOM associated with your image is correct and from a trusted source.
+
+## Discovery of SBOMs on Rekor (experimental)
+Syft can search the Rekor transparency log for SBOM attestations of binaries it finds while scanning and incorporate the results into the SBOMs it produces. This allows the use of SBOMs produced at build time (such as by a trusted builder), to augment binary metadata in the resultant SBOM. 
+
+The rekor-cataloger searches Rekor by hash for binaries and performs verification to ensure that the SBOMs and attestations have not been tampered with. It verifies the log entry has been signed by Rekor's public key, the certificate associated with the log entry chains back to the Fulcio root certificate, the log entry timestamp lies in the period of validity of the certificate, and other verifications. In the SBOM that Syft produces, the information is represented as an external document reference containing the URI and hash of the SBOM.
+
+This is an experimental feature. It uses external sources, a functionality that is new to Syft. The use of trusted builders to produce SBOMs has not yet been fully established, and more consideration of what external sources to trust is necessary. Currently, Syft accepts any SBOM attestation that has a valid certificate issued by Fulcio.   
+
+To enable the rekor-cataloger, use the flag ``` --catalogers rekor ```.

go.mod

@@ -57,6 +57,7 @@ require (
 require (
 	github.com/Masterminds/sprig/v3 v3.2.2
 	github.com/docker/docker v20.10.17+incompatible
+	github.com/go-openapi/runtime v0.24.1
 	github.com/google/go-containerregistry v0.11.0
 	github.com/in-toto/in-toto-golang v0.3.4-0.20220709202702-fa494aaa0add
 	github.com/knqyf263/go-rpmdb v0.0.0-20220629110411-9a3bd2ebb923
@@ -140,7 +141,6 @@ require (
 	github.com/go-openapi/jsonpointer v0.19.5 // indirect
 	github.com/go-openapi/jsonreference v0.20.0 // indirect
 	github.com/go-openapi/loads v0.21.1 // indirect
-	github.com/go-openapi/runtime v0.24.1 // indirect
 	github.com/go-openapi/spec v0.20.6 // indirect
 	github.com/go-openapi/strfmt v0.21.3 // indirect
 	github.com/go-openapi/swag v0.21.1 // indirect

internal/formats/spdx22json/model/doc_element_id.go

@@ -0,0 +1,12 @@
+package model
+
+import "github.com/anchore/syft/internal/formats/common/spdxhelpers"
+
+// DocElementID represents the identifier string portion of an SPDX document
+// identifier. It should be to used to reference any other SPDX document.
+// DocElementIDs should NOT contain the 'DOCUMENTREF-' portion.
+type DocElementID string
+
+func (d DocElementID) String() string {
+	return "DocumentRef-" + spdxhelpers.SanitizeElementID(string(d))
+}

internal/formats/spdx22json/model/external_document_ref.go

@@ -4,6 +4,6 @@ type ExternalDocumentRef struct {
 	// externalDocumentId is a string containing letters, numbers, ., - and/or + which uniquely identifies an external document within this document.
 	ExternalDocumentID string   `json:"externalDocumentId"`
 	Checksum           Checksum `json:"checksum"`
-	// SPDX ID for SpdxDocument.  A propoerty containing an SPDX document.
+	// SPDX ID for SpdxDocument.  A property containing an SPDX document.
 	SpdxDocument string `json:"spdxDocument"`
 }

internal/formats/spdx22json/to_format_model.go

@@ -1,6 +1,7 @@
 package spdx22json
 
 import (
+	"errors"
 	"fmt"
 	"sort"
 	"strings"
@@ -14,6 +15,7 @@ import (
 	"github.com/anchore/syft/syft/artifact"
 	"github.com/anchore/syft/syft/file"
 	"github.com/anchore/syft/syft/pkg"
+	"github.com/anchore/syft/syft/rekor"
 	"github.com/anchore/syft/syft/sbom"
 	"github.com/anchore/syft/syft/source"
 )
@@ -37,14 +39,55 @@ func toFormatModel(s sbom.SBOM) *model.Document {
 			},
 			LicenseListVersion: spdxlicense.Version,
 		},
-		DataLicense:       "CC0-1.0",
-		DocumentNamespace: namespace,
-		Packages:          toPackages(s.Artifacts.PackageCatalog, s.Relationships),
-		Files:             toFiles(s),
-		Relationships:     toRelationships(s.Relationships),
+		DataLicense:          "CC0-1.0",
+		ExternalDocumentRefs: toExternalDocumentRefs(s.Relationships),
+		DocumentNamespace:    namespace,
+		Packages:             toPackages(s.Artifacts.PackageCatalog, s.Relationships),
+		Files:                toFiles(s),
+		Relationships:        toRelationships(s.Relationships),
 	}
 }
 
+// isValidExternalRelationshipDocument returns if rel contains an ExternalRef and if it to_format_model know how to handle it.
+// An error is returned if rel contains an ExternalRef, but the rel cannot be handled
+func isValidExternalRelationshipDocument(rel artifact.Relationship) (bool, error) {
+	if _, ok := rel.From.(rekor.ExternalRef); ok {
+		return false, errors.New("syft cannot handle an ExternalRef in the FROM field of a relationship")
+	}
+	if externalRef, ok := rel.To.(rekor.ExternalRef); ok {
+		relationshipType := artifact.DescribedByRelationship
+		if rel.Type == relationshipType && toChecksumAlgorithm(externalRef.SpdxRef.Alg) == "SHA1" {
+			return true, nil
+		}
+		return false, fmt.Errorf("syft cannot handle an ExternalRef with relationship type: %v", relationshipType)
+	}
+	return false, nil
+}
+
+func toExternalDocumentRefs(relationships []artifact.Relationship) []model.ExternalDocumentRef {
+	externalDocRefs := []model.ExternalDocumentRef{}
+	for _, rel := range relationships {
+		valid, err := isValidExternalRelationshipDocument(rel)
+		if err != nil {
+			log.Warnf("dropping relationship %v: %w", rel, err)
+			continue
+		}
+		if valid {
+			externalRef := rel.To.(rekor.ExternalRef)
+			externalDocRef := model.ExternalDocumentRef{
+				ExternalDocumentID: model.DocElementID(rel.To.ID()).String(),
+				Checksum: model.Checksum{
+					Algorithm:     toChecksumAlgorithm(externalRef.SpdxRef.Alg),
+					ChecksumValue: externalRef.SpdxRef.Checksum,
+				},
+				SpdxDocument: externalRef.SpdxRef.URI,
+			}
+			externalDocRefs = append(externalDocRefs, externalDocRef)
+		}
+	}
+	return externalDocRefs
+}
+
 func toPackages(catalog *pkg.Catalog, relationships []artifact.Relationship) []model.Package {
 	packages := make([]model.Package, 0)
 
@@ -216,21 +259,34 @@ func toFileTypes(metadata *source.FileMetadata) (ty []string) {
 	return ty
 }
 
-func toRelationships(relationships []artifact.Relationship) (result []model.Relationship) {
+func toRelationships(relationships []artifact.Relationship) []model.Relationship {
+	result := []model.Relationship{}
 	for _, r := range relationships {
 		exists, relationshipType, comment := lookupRelationship(r.Type)
-
 		if !exists {
 			log.Warnf("unable to convert relationship from SPDX 2.2 JSON, dropping: %+v", r)
 			continue
 		}
 
-		result = append(result, model.Relationship{
-			SpdxElementID:      model.ElementID(r.From.ID()).String(),
-			RelationshipType:   relationshipType,
-			RelatedSpdxElement: model.ElementID(r.To.ID()).String(),
-			Comment:            comment,
-		})
+		rel := model.Relationship{
+			SpdxElementID:    model.ElementID(r.From.ID()).String(),
+			RelationshipType: relationshipType,
+			Comment:          comment,
+		}
+
+		// if this relationship contains an external document ref, we need to use DocElementID instead of ElementID
+		valid, err := isValidExternalRelationshipDocument(r)
+		if err != nil {
+			log.Warnf("dropping relationship %v: %w", rel, err)
+			continue
+		}
+		if valid {
+			rel.RelatedSpdxElement = model.DocElementID(r.To.ID()).String()
+		} else {
+			rel.RelatedSpdxElement = model.ElementID(r.To.ID()).String()
+		}
+
+		result = append(result, rel)
 	}
 	return result
 }
@@ -241,6 +297,10 @@ func lookupRelationship(ty artifact.RelationshipType) (bool, spdxhelpers.Relatio
 		return true, spdxhelpers.ContainsRelationship, ""
 	case artifact.OwnershipByFileOverlapRelationship:
 		return true, spdxhelpers.OtherRelationship, fmt.Sprintf("%s: indicates that the parent package claims ownership of a child package since the parent metadata indicates overlap with a location that a cataloger found the child package by", ty)
+	case artifact.DependencyOfRelationship:
+		return true, spdxhelpers.DependencyOfRelationship, ""
+	case artifact.DescribedByRelationship:
+		return true, spdxhelpers.DescribedByRelationship, ""
 	}
 	return false, "", ""
 }