expand the resolver to include content requests for a single reference

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

syft/cataloger/catalog.go

@@ -43,15 +43,16 @@ func Catalog(resolver scope.Resolver, catalogers ...Cataloger) (*pkg.Catalog, er
 	// perform analysis, accumulating errors for each failed analysis
 	var errs error
 	for _, theCataloger := range catalogers {
-		// TODO: check for multiple rounds of analyses by Iterate error
 		packages, err := theCataloger.Catalog(resolver)
 		if err != nil {
 			errs = multierror.Append(errs, err)
 			continue
 		}
 
-		log.Debugf("cataloger '%s' discovered '%d' packages", theCataloger.Name(), len(packages))
+		catalogedPackages := len(packages)
-		packagesDiscovered.N += int64(len(packages))
+
+		log.Debugf("cataloger '%s' discovered '%d' packages", theCataloger.Name(), catalogedPackages)
+		packagesDiscovered.N += int64(catalogedPackages)
 
 		for _, p := range packages {
 			catalog.Add(p)

syft/cataloger/cataloger.go

@@ -24,10 +24,8 @@ import (
 type Cataloger interface {
 	// Name returns a string that uniquely describes a cataloger
 	Name() string
-	// Catalog is given an object to resolve file references and content, this function should return any discovered Packages after analyzing the catalog source.
+	// Catalog is given an object to resolve file references and content, this function returns any discovered Packages after analyzing the catalog source.
 	Catalog(resolver scope.Resolver) ([]pkg.Package, error)
-	// TODO: add "IterationNeeded" error to indicate to the driver to continue with another Select/Catalog pass
-	// TODO: we should consider refactoring to return a set of io.Readers instead of the full contents themselves (allow for optional buffering).
 }
 
 // ImageCatalogers returns a slice of locally implemented catalogers that are fit for detecting installations of packages.

syft/cataloger/common/generic_cataloger.go

@@ -52,6 +52,7 @@ func (c *GenericCataloger) clear() {
 	c.parsers = make(map[file.Reference]ParserFn)
 }
 
+// Catalog is given an object to resolve file references and content, this function returns any discovered Packages after analyzing the catalog source.
 func (c *GenericCataloger) Catalog(resolver scope.Resolver) ([]pkg.Package, error) {
 	fileSelection := c.selectFiles(resolver)
 	contents, err := resolver.MultipleFileContentsByRef(fileSelection...)
@@ -88,7 +89,7 @@ func (c *GenericCataloger) selectFiles(resolver scope.FileResolver) []file.Refer
 	return c.selectedFiles
 }
 
-// Catalog takes a set of file contents and uses any configured parser functions to resolve and return discovered packages
+// catalog takes a set of file contents and uses any configured parser functions to resolve and return discovered packages
 func (c *GenericCataloger) catalog(contents map[file.Reference]string) ([]pkg.Package, error) {
 	defer c.clear()
 

syft/cataloger/common/generic_cataloger_test.go

@@ -20,6 +20,10 @@ func newTestResolver() *testResolver {
 	}
 }
 
+func (r *testResolver) FileContentsByRef(_ file.Reference) (string, error) {
+	return "", fmt.Errorf("not implemented")
+}
+
 func (r *testResolver) MultipleFileContentsByRef(_ ...file.Reference) (map[file.Reference]string, error) {
 	return r.contents, nil
 }

syft/scope/resolver.go

@@ -16,7 +16,9 @@ type Resolver interface {
 
 // ContentResolver knows how to get file content for given file.References
 type ContentResolver interface {
+	FileContentsByRef(ref file.Reference) (string, error)
 	MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error)
+	// TODO: we should consider refactoring to return a set of io.Readers or file.Openers instead of the full contents themselves (allow for optional buffering).
 }
 
 //  FileResolver knows how to get file.References for given string paths and globs

syft/scope/resolvers/all_layers_resolver.go

@@ -114,3 +114,9 @@ func (r *AllLayersResolver) FilesByGlob(patterns ...string) ([]file.Reference, e
 func (r *AllLayersResolver) MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error) {
 	return r.img.MultipleFileContentsByRef(f...)
 }
+
+// FileContentsByRef fetches file contents for a single file reference, irregardless of the source layer.
+// If the path does not exist an error is returned.
+func (r *AllLayersResolver) FileContentsByRef(ref file.Reference) (string, error) {
+	return r.img.FileContentsByRef(ref)
+}

syft/scope/resolvers/directory_resolver.go

@@ -82,9 +82,20 @@ func (s DirectoryResolver) MultipleFileContentsByRef(f ...file.Reference) (map[f
 		contents, err := fileContents(fileRef.Path)
 
 		if err != nil {
-			return refContents, fmt.Errorf("could not read contents of file: %s", fileRef.Path)
+			return nil, fmt.Errorf("could not read contents of file: %s", fileRef.Path)
 		}
 		refContents[fileRef] = string(contents)
 	}
 	return refContents, nil
 }
+
+// FileContentsByRef fetches file contents for a single file reference relative to a directory.
+// If the path does not exist an error is returned.
+func (s DirectoryResolver) FileContentsByRef(ref file.Reference) (string, error) {
+	contents, err := fileContents(ref.Path)
+	if err != nil {
+		return "", fmt.Errorf("could not read contents of file: %s", ref.Path)
+	}
+
+	return string(contents), nil
+}

syft/scope/resolvers/image_squash_resolver.go

@@ -78,3 +78,9 @@ func (r *ImageSquashResolver) FilesByGlob(patterns ...string) ([]file.Reference,
 func (r *ImageSquashResolver) MultipleFileContentsByRef(f ...file.Reference) (map[file.Reference]string, error) {
 	return r.img.MultipleFileContentsByRef(f...)
 }
+
+// FileContentsByRef fetches file contents for a single file reference, irregardless of the source layer.
+// If the path does not exist an error is returned.
+func (r *ImageSquashResolver) FileContentsByRef(ref file.Reference) (string, error) {
+	return r.img.FileContentsByRef(ref)
+}