rpm: implement an RPM filescanner to discern RPM filepaths

file.go

@@ -5,6 +5,7 @@ type FileKind string
 
 const (
 	FileKindWhiteout = FileKind("whiteout")
+	FileKindRPM      = FileKind("rpm")
 )
 
 // File represents interesting files that are found in the layer.

indexer/controller/coalesce.go

@@ -128,7 +128,7 @@
 		}
 
 		for k, v := range ir.Files {
-			source.Files[k] = v
+			source.Files[k] = append(source.Files[k], v...)
 		}
 	}
 	return source

indexer/controller/controller.go

@@ -41,7 +41,7 @@ func New(options *indexer.Options) *Controller {
 		Environments:  map[string][]*claircore.Environment{},
 		Distributions: map[string]*claircore.Distribution{},
 		Repositories:  map[string]*claircore.Repository{},
-		Files:         map[string]claircore.File{},
+		Files:         map[string][]claircore.File{},
 	}
 
 	s := &Controller{

indexreport.go

@@ -34,7 +34,7 @@ type IndexReport struct {
 	// an error string in the case the index did not succeed
 	Err string `json:"err"`
 	// Files doesn't end up in the json report but needs to be available at post-coalesce
-	Files map[string]File `json:"-"`
+	Files map[string][]File `json:"-"`
 }
 
 // IndexRecords returns a list of IndexRecords derived from the IndexReport

libindex/libindex.go

@@ -110,6 +110,7 @@ func New(ctx context.Context, opts *Options, cl *http.Client) (*Libindex, error)
 	opts.Ecosystems = append(opts.Ecosystems, whiteout.NewEcosystem(ctx))
 	opts.Resolvers = []indexer.Resolver{
 		&whiteout.Resolver{},
+		&rhel.Resolver{},
 	}
 
 	if cl == nil {

linux/coalescer.go

@@ -48,6 +48,7 @@
 		if len(artifacts.Pkgs) == 0 {
 			continue
 		}
+		lh := artifacts.Hash.String()
 
 		tmp := make(map[string][]*claircore.Package)
 		for _, pkg := range artifacts.Pkgs {
@@ -58,6 +59,12 @@
 		for db, pkgs := range tmp {
 			dbs[db] = pkgs
 		}
+		for _, f := range artifacts.Files {
+			if c.ir.Files == nil {
+				c.ir.Files = make(map[string][]claircore.File)
+			}
+			c.ir.Files[lh] = append(c.ir.Files[lh], f)
+		}
 	}
 
 	for db, packages := range dbs {

rhel/resolver.go

@@ -0,0 +1,57 @@
+package rhel
+
+import (
+	"context"
+
+	"github.com/quay/zlog"
+
+	"github.com/quay/claircore"
+	"github.com/quay/claircore/indexer"
+)
+
+var (
+	_ indexer.Resolver = (*Resolver)(nil)
+)
+
+type Resolver struct{}
+
+// Resolve takes a claircore.IndexReport and uses the rpm Files
+// to determine if the package originated from an RPM. If the package
+// was deemed to have been installed via RPM, it isn't included in the
+// final report.
+func (r *Resolver) Resolve(ctx context.Context, ir *claircore.IndexReport, layers []*claircore.Layer) *claircore.IndexReport {
+	for pkgID, pkg := range ir.Packages {
+		isRPMPackage := false
+	envLoop:
+		for _, env := range ir.Environments[pkgID] {
+			if env == nil {
+				continue
+			}
+			for _, rID := range env.RepositoryIDs {
+				r := ir.Repositories[rID]
+				if r.Key == repositoryKey {
+					isRPMPackage = true
+					break envLoop
+				}
+			}
+		}
+		if !isRPMPackage {
+		filesLoop:
+			for _, fs := range ir.Files {
+				for _, f := range fs {
+					if f.Kind == claircore.FileKindRPM && f.Path == pkg.Filepath {
+						zlog.Debug(ctx).
+							Str("package name", pkg.Name).
+							Str("package file", pkg.Filepath).
+							Str("rpm file", f.Path).
+							Msg("package determined to have come from RPM, deleting")
+						delete(ir.Packages, pkgID)
+						delete(ir.Environments, pkgID)
+						break filesLoop
+					}
+				}
+			}
+		}
+	}
+	return ir
+}