Update GitParse to handle quoted binary filenames (#2391)

* fix(gitparse): quoted binary files

* fix(gitparse): use bytes.Cut instead of regexp

* fix lint warning

---------

Co-authored-by: Zachary Rice <zachary.rice@trufflesec.com>

pkg/gitparse/gitparse.go

@@ -458,10 +458,17 @@ func (c *Parser) FromReader(ctx context.Context, stdOut io.Reader, diffChan chan
 		case isBinaryLine(latestState, line):
 			latestState = BinaryFileLine
 
-			currentDiff.PathB = pathFromBinaryLine(line)
+			path, ok := pathFromBinaryLine(line)
+			if !ok {
+				err = fmt.Errorf(`expected line to match 'Binary files a/fileA and b/fileB differ', got "%s"`, line)
+				ctx.Logger().Error(err, "Failed to parse binary file line")
+				latestState = ParseFailure
+				continue
+			}
 
 			// Don't do anything if the file is deleted. (pathA has file path, pathB is /dev/null)
-			if currentDiff.PathB != "" {
+			if path != "" {
+				currentDiff.PathB = path
 				currentDiff.IsBinary = true
 			}
 		case isFromFileLine(latestState, line):
@@ -708,15 +715,27 @@ func isBinaryLine(latestState ParseState, line []byte) bool {
 }
 
 // Get the b/ file path. Ignoring the edge case of files having `and /b` in the name for simplicity.
-func pathFromBinaryLine(line []byte) string {
-	logger := context.Background().Logger()
-	sbytes := bytes.Split(line, []byte(" and b/"))
-	if len(sbytes) != 2 {
-		logger.V(2).Info("Expected binary line to be in 'Binary files a/fileA and b/fileB differ' format.", "got", line)
-		return ""
-	}
-	bRaw := sbytes[1]
-	return string(bRaw[:len(bRaw)-8]) // drop the "b/" and " differ\n"
+func pathFromBinaryLine(line []byte) (string, bool) {
+	if bytes.Contains(line, []byte("and /dev/null")) {
+		return "", true
+	}
+
+	_, after, ok := bytes.Cut(line, []byte(" and b/"))
+	if ok {
+		// drop the " differ\n"
+		return string(after[:len(after)-8]), true
+	}
+
+	// Edge case where the path is quoted.
+	// https://github.com/trufflesecurity/trufflehog/issues/2384
+	_, after, ok = bytes.Cut(line, []byte(` and "b/`))
+	if ok {
+		// drop the `" differ\n`
+		return string(after[:len(after)-9]), true
+	}
+
+	// Unknown format.
+	return "", false
 }
 
 // --- a/internal/addrs/move_endpoint_module.go

pkg/gitparse/gitparse_test.go

@@ -591,12 +591,17 @@ func TestLineChecksNoStaged(t *testing.T) {
 
 func TestBinaryPathParse(t *testing.T) {
 	cases := map[string]string{
-		"Binary files /dev/null and b/plugin.sig differ\n":                    "plugin.sig",
-		"Binary files /dev/null and b/ Lunch and Learn - HCDiag.pdf differ\n": " Lunch and Learn - HCDiag.pdf",
+		"Binary files a/trufflehog_3.42.0_linux_arm64.tar.gz and /dev/null differ\n":                                   "",
+		"Binary files /dev/null and b/plugin.sig differ\n":                                                             "plugin.sig",
+		"Binary files /dev/null and b/ Lunch and Learn - HCDiag.pdf differ\n":                                          " Lunch and Learn - HCDiag.pdf",
+		"Binary files /dev/null and \"b/assets/retailers/ON-ikony-Platforma-ecom \\342\\200\\224 kopia.png\" differ\n": "assets/retailers/ON-ikony-Platforma-ecom \\342\\200\\224 kopia.png",
 	}
 
 	for name, expected := range cases {
-		filename := pathFromBinaryLine([]byte(name))
+		filename, ok := pathFromBinaryLine([]byte(name))
+		if !ok {
+			t.Errorf("Failed to get path: %s", name)
+		}
 		if filename != expected {
 			t.Errorf("Expected: %s, Got: %s", expected, filename)
 		}