add file contents cataloger

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

syft/file/contents_cataloger.go

@@ -0,0 +1,68 @@
+package file
+
+import (
+	"bytes"
+	"encoding/base64"
+	"fmt"
+	"io"
+
+	"github.com/anchore/syft/internal/log"
+	"github.com/anchore/syft/syft/source"
+)
+
+type ContentsCataloger struct {
+	globs              []string
+	skipFilesAboveSize int64
+}
+
+func NewContentsCataloger(globs []string, skipFilesAboveSize int64) (*ContentsCataloger, error) {
+	return &ContentsCataloger{
+		globs:              globs,
+		skipFilesAboveSize: skipFilesAboveSize,
+	}, nil
+}
+
+func (i *ContentsCataloger) Catalog(resolver source.FileResolver) (map[source.Location]string, error) {
+	results := make(map[source.Location]string)
+	var locations []source.Location
+
+	locations, err := resolver.FilesByGlob(i.globs...)
+	if err != nil {
+		return nil, err
+	}
+
+	for _, location := range locations {
+		metadata, err := resolver.FileMetadataByLocation(location)
+		if err != nil {
+			return nil, err
+		}
+
+		if i.skipFilesAboveSize > 0 && metadata.Size > i.skipFilesAboveSize {
+			continue
+		}
+
+		result, err := i.catalogLocation(resolver, location)
+		if err != nil {
+			return nil, err
+		}
+		results[location] = result
+	}
+	log.Debugf("file contents cataloger processed %d files", len(results))
+
+	return results, nil
+}
+
+func (i *ContentsCataloger) catalogLocation(resolver source.FileResolver, location source.Location) (string, error) {
+	contentReader, err := resolver.FileContentsByLocation(location)
+	if err != nil {
+		return "", err
+	}
+	defer contentReader.Close()
+
+	buf := &bytes.Buffer{}
+	if _, err = io.Copy(base64.NewEncoder(base64.StdEncoding, buf), contentReader); err != nil {
+		return "", fmt.Errorf("unable to observe contents of %+v: %+v", location.RealPath, err)
+	}
+
+	return buf.String(), nil
+}

syft/file/contents_cataloger_test.go

@@ -0,0 +1,89 @@
+package file
+
+import (
+	"testing"
+
+	"github.com/anchore/syft/syft/source"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestContentsCataloger(t *testing.T) {
+	allFiles := []string{"test-fixtures/last/path.txt", "test-fixtures/another-path.txt", "test-fixtures/a-path.txt"}
+
+	tests := []struct {
+		name       string
+		globs      []string
+		maxSize    int64
+		files      []string
+		expected   map[source.Location]string
+		catalogErr bool
+	}{
+		{
+			name:  "multi-pattern",
+			globs: []string{"test-fixtures/last/*.txt", "test-fixtures/*.txt"},
+			files: allFiles,
+			expected: map[source.Location]string{
+				source.NewLocation("test-fixtures/last/path.txt"):    "dGVzdC1maXh0dXJlcy9sYXN0L3BhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/another-path.txt"): "dGVzdC1maXh0dXJlcy9hbm90aGVyLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/a-path.txt"):       "dGVzdC1maXh0dXJlcy9hLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+			},
+		},
+		{
+			name:     "no-patterns",
+			globs:    []string{},
+			files:    []string{"test-fixtures/last/path.txt", "test-fixtures/another-path.txt", "test-fixtures/a-path.txt"},
+			expected: map[source.Location]string{},
+		},
+		{
+			name:  "all-txt",
+			globs: []string{"**/*.txt"},
+			files: allFiles,
+			expected: map[source.Location]string{
+				source.NewLocation("test-fixtures/last/path.txt"):    "dGVzdC1maXh0dXJlcy9sYXN0L3BhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/another-path.txt"): "dGVzdC1maXh0dXJlcy9hbm90aGVyLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/a-path.txt"):       "dGVzdC1maXh0dXJlcy9hLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+			},
+		},
+		{
+			name:  "subpath",
+			globs: []string{"test-fixtures/*.txt"},
+			files: allFiles,
+			expected: map[source.Location]string{
+				source.NewLocation("test-fixtures/another-path.txt"): "dGVzdC1maXh0dXJlcy9hbm90aGVyLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/a-path.txt"):       "dGVzdC1maXh0dXJlcy9hLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+			},
+		},
+		{
+			name:    "size-filter",
+			maxSize: 42,
+			globs:   []string{"**/*.txt"},
+			files:   allFiles,
+			expected: map[source.Location]string{
+				source.NewLocation("test-fixtures/last/path.txt"): "dGVzdC1maXh0dXJlcy9sYXN0L3BhdGgudHh0IGZpbGUgY29udGVudHMh",
+				source.NewLocation("test-fixtures/a-path.txt"):    "dGVzdC1maXh0dXJlcy9hLXBhdGgudHh0IGZpbGUgY29udGVudHMh",
+			},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			c, err := NewContentsCataloger(test.globs, test.maxSize)
+			if err != nil {
+				t.Fatalf("could not create cataloger: %+v", err)
+			}
+
+			resolver := source.NewMockResolverForPaths(test.files...)
+			actual, err := c.Catalog(resolver)
+			if err != nil && !test.catalogErr {
+				t.Fatalf("could not catalog (but should have been able to): %+v", err)
+			} else if err == nil && test.catalogErr {
+				t.Fatalf("expected catalog error but did not get one")
+			} else if test.catalogErr && err != nil {
+				return
+			}
+
+			assert.Equal(t, test.expected, actual, "mismatched contents")
+
+		})
+	}
+}

syft/source/mock_resolver.go

@@ -5,7 +5,7 @@ import (
 	"io"
 	"os"
 
-	"github.com/anchore/syft/internal/file"
+	"github.com/bmatcuk/doublestar/v2"
 )
 
 var _ FileResolver = (*MockResolver)(nil)
@@ -84,7 +84,11 @@ func (r MockResolver) FilesByGlob(patterns ...string) ([]Location, error) {
 	var results []Location
 	for _, pattern := range patterns {
 		for _, location := range r.Locations {
-			if file.GlobMatch(pattern, location.RealPath) {
+			matches, err := doublestar.Match(pattern, location.RealPath)
+			if err != nil {
+				return nil, err
+			}
+			if matches {
 				results = append(results, location)
 			}
 		}