Add python wheel egg relationships (#2903)

* add python package relationships

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

* nil for empty relationships collections

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

* new json schema for optional python requiremenets

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

* update format snapshots for python packages

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

* decompose python parsers more + add tests around plural fields

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

* update JSON schema with python dep refs

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

---------

Signed-off-by: Alex Goodman <wagoodman@users.noreply.github.com>
This commit is contained in:
Alex Goodman
2024-05-25 20:11:38 +00:00
committed by GitHub
parent 64e11d53e9
commit 05e8ba948d
42 changed files with 3706 additions and 321 deletions
@@ -4,6 +4,7 @@ import (
"path"
"github.com/anchore/syft/internal/log"
"github.com/anchore/syft/internal/relationship"
"github.com/anchore/syft/internal/sbomsync"
"github.com/anchore/syft/syft/artifact"
"github.com/anchore/syft/syft/file"
@@ -20,22 +21,22 @@ func NewDependencyRelationships(resolver file.Resolver, accessor sbomsync.Access
// 3. craft package-to-package relationships for each binary that represent shared library dependencies
//note: we only care about package-to-package relationships
var relIndex *relationshipIndex
var relIndex *relationship.Index
accessor.ReadFromSBOM(func(s *sbom.SBOM) {
relIndex = newRelationshipIndex(s.Relationships...)
relIndex = relationship.NewIndex(s.Relationships...)
})
return generateRelationships(resolver, accessor, index, relIndex)
}
func generateRelationships(resolver file.Resolver, accessor sbomsync.Accessor, index *sharedLibraryIndex, relIndex *relationshipIndex) []artifact.Relationship {
func generateRelationships(resolver file.Resolver, accessor sbomsync.Accessor, index *sharedLibraryIndex, relIndex *relationship.Index) []artifact.Relationship {
// read all existing dependencyOf relationships
accessor.ReadFromSBOM(func(s *sbom.SBOM) {
for _, r := range s.Relationships {
if r.Type != artifact.DependencyOfRelationship {
continue
}
relIndex.track(r)
relIndex.Track(r)
}
})
@@ -58,7 +59,7 @@ func generateRelationships(resolver file.Resolver, accessor sbomsync.Accessor, i
}
})
return relIndex.newRelationships()
return relIndex.NewRelationships()
}
// PackagesToRemove returns a list of binary packages (resolved by the ELF cataloger) that should be removed from the SBOM
@@ -146,7 +147,7 @@ func getBinaryPackagesToDelete(resolver file.Resolver, s *sbom.SBOM) []artifact.
return pkgsToDelete
}
func populateRelationships(exec file.Executable, parentPkg pkg.Package, resolver file.Resolver, relIndex *relationshipIndex, index *sharedLibraryIndex) {
func populateRelationships(exec file.Executable, parentPkg pkg.Package, resolver file.Resolver, relIndex *relationship.Index, index *sharedLibraryIndex) {
for _, libReference := range exec.ImportedLibraries {
// for each library reference, check s.Artifacts.Packages.Sorted(pkg.BinaryPkg) for a binary package that represents that library
// if found, create a relationship between the parent package and the library package
@@ -166,7 +167,7 @@ func populateRelationships(exec file.Executable, parentPkg pkg.Package, resolver
realBaseName := path.Base(loc.RealPath)
pkgCollection := index.owningLibraryPackage(realBaseName)
if pkgCollection.PackageCount() < 1 {
relIndex.add(
relIndex.Add(
artifact.Relationship{
From: loc.Coordinates,
To: parentPkg,
@@ -175,7 +176,7 @@ func populateRelationships(exec file.Executable, parentPkg pkg.Package, resolver
)
}
for _, p := range pkgCollection.Sorted() {
relIndex.add(
relIndex.Add(
artifact.Relationship{
From: p,
To: parentPkg,
@@ -225,7 +225,6 @@ func TestNewDependencyRelationships(t *testing.T) {
resolver: nil,
coordinateIndex: map[file.Coordinates]file.Executable{},
packages: []pkg.Package{},
want: make([]artifact.Relationship, 0),
},
{
name: "given a package that imports glibc, expect a relationship between the two packages when the package is an executable",
@@ -297,7 +296,6 @@ func TestNewDependencyRelationships(t *testing.T) {
Type: artifact.DependencyOfRelationship,
},
},
want: []artifact.Relationship{},
},
{
name: "given a package that imports a library that is not tracked by the resolver, expect no relationships to be created",
@@ -308,7 +306,6 @@ func TestNewDependencyRelationships(t *testing.T) {
parallelLibCoordinate: syftTestFixtureExecutable2,
},
packages: []pkg.Package{glibCPackage, syftTestFixturePackage},
want: []artifact.Relationship{},
},
}
for _, tt := range tests {
@@ -1,59 +0,0 @@
package binary
import (
"github.com/scylladb/go-set/strset"
"github.com/anchore/syft/syft/artifact"
)
type relationshipIndex struct {
typesByFromTo map[artifact.ID]map[artifact.ID]*strset.Set
additional []artifact.Relationship
}
func newRelationshipIndex(existing ...artifact.Relationship) *relationshipIndex {
r := &relationshipIndex{
typesByFromTo: make(map[artifact.ID]map[artifact.ID]*strset.Set),
additional: make([]artifact.Relationship, 0),
}
for _, rel := range existing {
r.track(rel)
}
return r
}
// track this relationship as "exists" in the index (this is used to prevent duplicate relationships from being added).
// returns true if the relationship is new to the index, false otherwise.
func (i *relationshipIndex) track(r artifact.Relationship) bool {
fromID := r.From.ID()
if _, ok := i.typesByFromTo[fromID]; !ok {
i.typesByFromTo[fromID] = make(map[artifact.ID]*strset.Set)
}
toID := r.To.ID()
if _, ok := i.typesByFromTo[fromID][toID]; !ok {
i.typesByFromTo[fromID][toID] = strset.New()
}
var exists bool
if i.typesByFromTo[fromID][toID].Has(string(r.Type)) {
exists = true
}
i.typesByFromTo[fromID][toID].Add(string(r.Type))
return !exists
}
// add a new relationship to the index, returning true if the relationship is new to the index, false otherwise (thus is a duplicate).
// nolint:unparam
func (i *relationshipIndex) add(r artifact.Relationship) bool {
if i.track(r) {
i.additional = append(i.additional, r)
return true
}
return false
}
func (i *relationshipIndex) newRelationships() []artifact.Relationship {
return i.additional
}
@@ -1,128 +0,0 @@
package binary
import (
"reflect"
"testing"
"github.com/scylladb/go-set/strset"
"github.com/anchore/syft/syft/artifact"
)
func Test_newRelationshipIndex(t *testing.T) {
from := fakeIdentifiable{id: "from"}
to := fakeIdentifiable{id: "to"}
tests := []struct {
name string
given []artifact.Relationship
want *relationshipIndex
}{
{
name: "newRelationshipIndex returns an empty index with no existing relationships",
want: &relationshipIndex{
typesByFromTo: make(map[artifact.ID]map[artifact.ID]*strset.Set),
additional: make([]artifact.Relationship, 0),
},
},
{
name: "newRelationshipIndex returns an index which tracks existing relationships",
given: []artifact.Relationship{
{
From: from,
To: to,
Type: artifact.EvidentByRelationship,
},
},
want: &relationshipIndex{
typesByFromTo: map[artifact.ID]map[artifact.ID]*strset.Set{
"from": {
"to": strset.New("evident-by"),
},
},
additional: make([]artifact.Relationship, 0),
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
if got := newRelationshipIndex(tt.given...); !reflect.DeepEqual(got, tt.want) {
t.Errorf("newRelationshipIndex() = %v, want %v", got, tt.want)
}
})
}
}
func Test_relationshipIndex_track(t *testing.T) {
from := fakeIdentifiable{id: "from"}
to := fakeIdentifiable{id: "to"}
relationship := artifact.Relationship{From: from, To: to, Type: artifact.EvidentByRelationship}
tests := []struct {
name string
existing []artifact.Relationship
given artifact.Relationship
want bool
}{
{
name: "track returns true for a new relationship",
existing: []artifact.Relationship{},
given: relationship,
want: true,
},
{
name: "track returns false for an existing relationship",
existing: []artifact.Relationship{relationship},
given: relationship,
want: false,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
i := newRelationshipIndex(tt.existing...)
if got := i.track(tt.given); got != tt.want {
t.Errorf("track() = %v, want %v", got, tt.want)
}
})
}
}
func Test_relationshipIndex_add(t *testing.T) {
from := fakeIdentifiable{id: "from"}
to := fakeIdentifiable{id: "to"}
relationship := artifact.Relationship{From: from, To: to, Type: artifact.EvidentByRelationship}
tests := []struct {
name string
existing []artifact.Relationship
given artifact.Relationship
want bool
}{
{
name: "add returns true for a new relationship",
existing: []artifact.Relationship{},
given: relationship,
want: true,
},
{
name: "add returns false for an existing relationship",
existing: []artifact.Relationship{relationship},
given: relationship,
want: false,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
i := newRelationshipIndex(tt.existing...)
if got := i.add(tt.given); got != tt.want {
t.Errorf("add() = %v, want %v", got, tt.want)
}
})
}
}
type fakeIdentifiable struct {
id string
}
func (f fakeIdentifiable) ID() artifact.ID {
return artifact.ID(f.id)
}