add symlink content fetching support

imgbom/cataloger/common/generic_cataloger.go

@@ -4,13 +4,11 @@ import (
 	"strings"
 
 	"github.com/anchore/imgbom/imgbom/pkg"
+	"github.com/anchore/imgbom/imgbom/scope"
 	"github.com/anchore/imgbom/internal/log"
 	"github.com/anchore/stereoscope/pkg/file"
-	"github.com/anchore/stereoscope/pkg/tree"
 )
 
-// TODO: put under test...
-
 // GenericCataloger implements the Catalog interface and is responsible for dispatching the proper parser function for
 // a given path or glob pattern. This is intended to be reusable across many package cataloger types.
 type GenericCataloger struct {
@@ -45,25 +43,26 @@ func (a *GenericCataloger) clear() {
 }
 
 // SelectFiles takes a set of file trees and resolves and file references of interest for future cataloging
-func (a *GenericCataloger) SelectFiles(trees []tree.FileTreeReader) []file.Reference {
-	for _, t := range trees {
-		// select by exact path
-		for path, parser := range a.pathParsers {
-			f := t.File(file.Path(path))
-			if f != nil {
-				a.register([]file.Reference{*f}, parser)
-			}
+func (a *GenericCataloger) SelectFiles(resolver scope.FileResolver) []file.Reference {
+	// select by exact path
+	for path, parser := range a.pathParsers {
+		files, err := resolver.FilesByPath(file.Path(path))
+		if err != nil {
+			log.Errorf("cataloger failed to select files by path: %w", err)
 		}
+		if files != nil {
+			a.register(files, parser)
+		}
+	}
 
-		// select by pattern
-		for globPattern, parser := range a.globParsers {
-			fileMatches, err := t.FilesByGlob(globPattern)
-			if err != nil {
-				log.Errorf("failed to find files by glob: %s", globPattern)
-			}
-			if fileMatches != nil {
-				a.register(fileMatches, parser)
-			}
+	// select by glob pattern
+	for globPattern, parser := range a.globParsers {
+		fileMatches, err := resolver.FilesByGlob(globPattern)
+		if err != nil {
+			log.Errorf("failed to find files by glob: %s", globPattern)
+		}
+		if fileMatches != nil {
+			a.register(fileMatches, parser)
 		}
 	}
 

imgbom/cataloger/common/generic_cataloger_test.go

@@ -0,0 +1,117 @@
+package common
+
+import (
+	"fmt"
+	"io"
+	"io/ioutil"
+	"testing"
+
+	"github.com/anchore/imgbom/imgbom/pkg"
+	"github.com/anchore/imgbom/internal"
+	"github.com/anchore/stereoscope/pkg/file"
+)
+
+type testResolver struct {
+	contents map[file.Reference]string
+}
+
+func newTestResolver() *testResolver {
+	return &testResolver{
+		contents: make(map[file.Reference]string),
+	}
+}
+
+func (r *testResolver) FilesByPath(paths ...file.Path) ([]file.Reference, error) {
+	results := make([]file.Reference, len(paths))
+
+	for idx, p := range paths {
+		results[idx] = file.NewFileReference(p)
+		r.contents[results[idx]] = fmt.Sprintf("%s file contents!", p)
+	}
+
+	return results, nil
+}
+
+func (r *testResolver) FilesByGlob(patterns ...string) ([]file.Reference, error) {
+	path := "/a-path.txt"
+	ref := file.NewFileReference(file.Path(path))
+	r.contents[ref] = fmt.Sprintf("%s file contents!", path)
+	return []file.Reference{ref}, nil
+}
+
+func parser(reader io.Reader) ([]pkg.Package, error) {
+	contents, err := ioutil.ReadAll(reader)
+	if err != nil {
+		panic(err)
+	}
+	return []pkg.Package{
+		{
+			Name: string(contents),
+		},
+	}, nil
+}
+
+func TestGenericCataloger(t *testing.T) {
+
+	globParsers := map[string]ParserFn{
+		"**a-path.txt": parser,
+	}
+	pathParsers := map[string]ParserFn{
+		"/another-path.txt": parser,
+		"/last/path.txt":    parser,
+	}
+
+	resolver := newTestResolver()
+	cataloger := NewGenericCataloger(pathParsers, globParsers)
+
+	selected := cataloger.SelectFiles(resolver)
+
+	if len(selected) != 3 {
+		t.Fatalf("unexpected selection length: %d", len(selected))
+	}
+
+	expectedSelection := internal.NewStringSetFromSlice([]string{"/last/path.txt", "/another-path.txt", "/a-path.txt"})
+	selectionByPath := make(map[string]file.Reference)
+	for _, s := range selected {
+		if !expectedSelection.Contains(string(s.Path)) {
+			t.Errorf("unexpected selection path: %+v", s.Path)
+		}
+		selectionByPath[string(s.Path)] = s
+	}
+
+	upstream := "some-other-cataloger"
+	expectedPkgs := make(map[file.Reference]pkg.Package)
+	for path, ref := range selectionByPath {
+		expectedPkgs[ref] = pkg.Package{
+			FoundBy: upstream,
+			Source:  []file.Reference{ref},
+			Name:    fmt.Sprintf("%s file contents!", path),
+		}
+	}
+
+	actualPkgs, err := cataloger.Catalog(resolver.contents, upstream)
+	if err != nil {
+		t.Fatalf("cataloger catalog action failed: %+v", err)
+	}
+
+	if len(actualPkgs) != len(expectedPkgs) {
+		t.Fatalf("unexpected packages len: %d", len(actualPkgs))
+	}
+
+	for _, p := range actualPkgs {
+		ref := p.Source[0]
+		exP, ok := expectedPkgs[ref]
+		if !ok {
+			t.Errorf("missing expected pkg: ref=%+v", ref)
+			continue
+		}
+
+		if p.FoundBy != exP.FoundBy {
+			t.Errorf("bad upstream: %s", p.FoundBy)
+		}
+
+		if exP.Name != p.Name {
+			t.Errorf("bad contents mapping: %+v", p.Source)
+		}
+	}
+}