From 61a3d1784a1ede4e92a55384d50d5e1ba9952c25 Mon Sep 17 00:00:00 2001
From: Adam McClenaghan <adam@mcclenaghan.co.uk>
Date: Thu, 24 Apr 2025 17:22:36 +0100
Subject: [PATCH] (feat): support skipping archive extraction with file source
 (#3795)

Signed-off-by: Adam McClenaghan <adam@mcclenaghan.co.uk>
---
 syft/source/filesource/file_source.go      | 21 +++++++++-----
 syft/source/filesource/file_source_test.go | 32 +++++++++++++++-------
 2 files changed, 36 insertions(+), 17 deletions(-)

diff --git a/syft/source/filesource/file_source.go b/syft/source/filesource/file_source.go
index a87ace55d..fead9cce5 100644
--- a/syft/source/filesource/file_source.go
+++ b/syft/source/filesource/file_source.go
@@ -26,10 +26,11 @@ import (
 var _ source.Source = (*fileSource)(nil)
 
 type Config struct {
-	Path             string
-	Exclude          source.ExcludeConfig
-	DigestAlgorithms []crypto.Hash
-	Alias            source.Alias
+	Path               string
+	Exclude            source.ExcludeConfig
+	DigestAlgorithms   []crypto.Hash
+	Alias              source.Alias
+	SkipExtractArchive bool
 }
 
 type fileSource struct {
@@ -58,7 +59,7 @@ func New(cfg Config) (source.Source, error) {
 		return nil, fmt.Errorf("given path is a directory: %q", cfg.Path)
 	}
 
-	analysisPath, cleanupFn := fileAnalysisPath(cfg.Path)
+	analysisPath, cleanupFn := fileAnalysisPath(cfg.Path, cfg.SkipExtractArchive)
 
 	var digests []file.Digest
 	if len(cfg.DigestAlgorithms) > 0 {
@@ -206,9 +207,15 @@ func (s *fileSource) Close() error {
 
 // fileAnalysisPath returns the path given, or in the case the path is an archive, the location where the archive
 // contents have been made available. A cleanup function is provided for any temp files created (if any).
-func fileAnalysisPath(path string) (string, func() error) {
-	var analysisPath = path
+// Users can disable unpacking archives, allowing individual cataloguers to extract them instead (where
+// supported)
+func fileAnalysisPath(path string, skipExtractArchive bool) (string, func() error) {
 	var cleanupFn = func() error { return nil }
+	var analysisPath = path
+
+	if skipExtractArchive {
+		return analysisPath, cleanupFn
+	}
 
 	// if the given file is an archive (as indicated by the file extension and not MIME type) then unarchive it and
 	// use the contents as the source. Note: this does NOT recursively unarchive contents, only the given path is
diff --git a/syft/source/filesource/file_source_test.go b/syft/source/filesource/file_source_test.go
index 1f046d253..ff5bc824a 100644
--- a/syft/source/filesource/file_source_test.go
+++ b/syft/source/filesource/file_source_test.go
@@ -99,13 +99,14 @@ func TestNewFromFile_WithArchive(t *testing.T) {
 	testutil.Chdir(t, "..") // run with source/test-fixtures
 
 	testCases := []struct {
-		desc       string
-		input      string
-		expString  string
-		inputPaths []string
-		expRefs    int
-		layer2     bool
-		contents   string
+		desc               string
+		input              string
+		expString          string
+		inputPaths         []string
+		expRefs            int
+		layer2             bool
+		contents           string
+		skipExtractArchive bool
 	}{
 		{
 			desc:       "path detected",
@@ -121,14 +122,25 @@ func TestNewFromFile_WithArchive(t *testing.T) {
 			layer2:     true,
 			contents:   "Another .vimrc file",
 		},
+		{
+			desc:               "skip extract archive",
+			input:              "test-fixtures/path-detected",
+			inputPaths:         []string{"/.vimrc"},
+			expRefs:            0,
+			layer2:             false,
+			skipExtractArchive: true,
+		},
 	}
 	for _, test := range testCases {
 		t.Run(test.desc, func(t *testing.T) {
 			archivePath := setupArchiveTest(t, test.input, test.layer2)
 
-			src, err := New(Config{
-				Path: archivePath,
-			})
+			cfg := Config{
+				Path:               archivePath,
+				SkipExtractArchive: test.skipExtractArchive,
+			}
+
+			src, err := New(cfg)
 			require.NoError(t, err)
 			t.Cleanup(func() {
 				require.NoError(t, src.Close())