Use different approach to fix tests

Signed-off-by: egibs <20933572+egibs@users.noreply.github.com>
chainguard-dev · Dec 18, 2024 · d3c9c8c · d3c9c8c
1 parent 53dc757
commit d3c9c8c
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 20 deletions.
diff --git a/pkg/action/scan.go b/pkg/action/scan.go
@@ -498,7 +498,7 @@ func processArchive(ctx context.Context, c malcontent.Config, rfs []fs.FS, archi
 	if err != nil {
 		return nil, fmt.Errorf("failed to determine file type: %w", err)
 	}
-	if ft != nil && ft.MIME == "application/octet-stream" {
+	if ft != nil && ft.MIME == "application/x-python-joblib" {
 		logger.Debugf("skipping unsupported archive: %s", archivePath)
 		return nil, nil
 	}

diff --git a/pkg/archive/archive.go b/pkg/archive/archive.go
@@ -32,7 +32,7 @@ func extractNestedArchive(
 	if err != nil {
 		return fmt.Errorf("failed to determine file type: %w", err)
 	}
-	if ft != nil && ft.MIME == "application/octet-stream" {
+	if ft != nil && ft.MIME == "application/x-python-joblib" {
 		return nil
 	}
 	if ft != nil && ft.MIME == "application/zlib" {
@@ -55,7 +55,7 @@ func extractNestedArchive(
 		if err != nil {
 			return fmt.Errorf("failed to determine file type: %w", err)
 		}
-		if ft != nil && ft.MIME == "application/octet-stream" {
+		if ft != nil && ft.MIME == "application/x-python-joblib" {
 			return nil
 		}
 		if ft != nil && ft.MIME == "application/zlib" {
@@ -109,7 +109,7 @@ func ExtractArchiveToTempDir(ctx context.Context, path string) (string, error) {
 	if err != nil {
 		return "", fmt.Errorf("failed to determine file type: %w", err)
 	}
-	if ft != nil && ft.MIME == "application/octet-stream" {
+	if ft != nil && ft.MIME == "application/x-python-joblib" {
 		return "", nil
 	}
 	if ft != nil && ft.MIME == "application/zlib" {

diff --git a/pkg/programkind/programkind.go b/pkg/programkind/programkind.go
@@ -66,6 +66,7 @@ var supportedKind = map[string]string{
 	"hh":      "text/x-h",
 	"html":    "",
 	"java":    "text/x-java",
+	"joblib":  "application/x-python-joblib",
 	"js":      "application/javascript",
 	"lnk":     "application/x-ms-shortcut",
 	"lua":     "text/x-lua",
@@ -140,20 +141,6 @@ func makeFileType(path string, ext string, mime string) *FileType {
 		return &FileType{MIME: "application/json", Ext: ext}
 	}
 
-	// Capture joblib files that cannot be decompressed externally
-	// https://joblib.readthedocs.io/en/stable/generated/joblib.dump.html
-	joblibExts := []string{
-		"z",
-		"gz",
-		"bz2",
-		"xz",
-		"lzma",
-	}
-
-	if slices.Contains(joblibExts, ext) && mime == "application/octet-stream" {
-		return &FileType{MIME: "application/octet-stream", Ext: ext}
-	}
-
 	if supportedKind[ext] == "" {
 		return nil
 	}
@@ -193,8 +180,9 @@ func File(path string) (*FileType, error) {
 	// first strategy: mimetype
 	mtype, err := mimetype.DetectFile(path)
 	if err == nil {
-		ft := makeFileType(path, GetExt(path), mtype.String())
-		return ft, nil
+		if ft := makeFileType(path, mtype.Extension(), mtype.String()); ft != nil {
+			return ft, nil
+		}
 	}
 
 	// second strategy: path (extension, mostly)
@@ -249,6 +237,21 @@ func File(path string) (*FileType, error) {
 		return Path(".gzip"), nil
 	case hdr[0] == '\x78' && hdr[1] == '\x5E':
 		return Path(".Z"), nil
+	// Capture joblib files that cannot be decompressed externally
+	// https://joblib.readthedocs.io/en/stable/generated/joblib.dump.html
+	// Check the header, file extension, and MIME type to be as specific as possible
+	case hdr[0] == '\x5A' && hdr[1] == '\x46' && hdr[2] == '\x30' && hdr[3] == '\x78':
+		joblibExts := []string{
+			".z",
+			".gz",
+			".bz2",
+			".xz",
+			".lzma",
+		}
+
+		if slices.Contains(joblibExts, GetExt(path)) && mtype.String() == "application/octet-stream" {
+			return Path(".joblib"), nil
+		}
 	}
 	return nil, nil
 }