fix(domain): reject line separators and bidi marks in merge request text

commit: c7d4ad1e4076f28d00dfbdf42521925ac6474477
author: brice · 11d ago · 2026-10-10 17:37 UTC
parent: 59b81ca
signature: GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email

2 files · +20 -6

M internal/domain/text.go +9 -5


@@ -59,7 +59,8 @@ func (t MRTitle) IsZero() bool { return t.s == "" }
59	59
60	60		// PlainText is a validated multi-line plain text, such as a description or a
61	61		// comment: valid UTF-8, at most MaxTextLen characters, line breaks allowed,
62		-	// no other control character, no bidirectional override or isolate character.
	62	+	// no other control character, no U+2028 or U+2029, no bidirectional mark,
	63	+	// override or isolate character.
63	64		// The zero value is the empty text.
64	65		type PlainText struct{ s string }
65	66

@@ -103,6 +104,8 @@ func textProblem(s string, multiline bool, maxLen int) string {
103	104		}
104	105		case unicode.IsControl(r):
105	106		return fmt.Sprintf("contains the control character U+%04X", r)
	107	+	case r == '\u2028' \|\| r == '\u2029':
	108	+	return fmt.Sprintf("contains the line or paragraph separator U+%04X", r)
106	109		case isBidiControl(r):
107	110		return fmt.Sprintf("contains the bidirectional override or isolate character U+%04X", r)
108	111		}

@@ -110,11 +113,12 @@ func textProblem(s string, multiline bool, maxLen int) string {
110	113		return ""
111	114		}
112	115
113		-	// isBidiControl reports whether r is a bidirectional embedding, override or
114		-	// isolate character (U+202A to U+202E, U+2066 to U+2069), which can make text
115		-	// read differently from how it is stored.
	116	+	// isBidiControl reports whether r is a bidirectional mark, embedding, override
	117	+	// or isolate character (U+200E, U+200F, U+061C, U+202A to U+202E, U+2066 to
	118	+	// U+2069), which can make text read differently from how it is stored.
116	119		func isBidiControl(r rune) bool {
117		-	return (r >= 0x202a && r <= 0x202e) \|\| (r >= 0x2066 && r <= 0x2069)
	120	+	return r == 0x200e \|\| r == 0x200f \|\| r == 0x061c \|\|
	121	+	(r >= 0x202a && r <= 0x202e) \|\| (r >= 0x2066 && r <= 0x2069)
118	122		}
119	123
120	124		// FieldViolation names a request field that breaks a rule.

 // PlainText is a validated multi-line plain text, such as a description or a
 // comment: valid UTF-8, at most MaxTextLen characters, line breaks allowed,
-// no other control character, no bidirectional override or isolate character.
+// no other control character, no U+2028 or U+2029, no bidirectional mark,
+// override or isolate character.
 // The zero value is the empty text.
 type PlainText struct{ s string }
 			}
 		case unicode.IsControl(r):
 			return fmt.Sprintf("contains the control character U+%04X", r)
+		case r == '\u2028' || r == '\u2029':
+			return fmt.Sprintf("contains the line or paragraph separator U+%04X", r)
 		case isBidiControl(r):
 			return fmt.Sprintf("contains the bidirectional override or isolate character U+%04X", r)
 		}
 	return ""
 }
-// isBidiControl reports whether r is a bidirectional embedding, override or
+// isBidiControl reports whether r is a bidirectional mark, embedding, override
-// isolate character (U+202A to U+202E, U+2066 to U+2069), which can make text
+// or isolate character (U+200E, U+200F, U+061C, U+202A to U+202E, U+2066 to
-// read differently from how it is stored.
+// U+2069), which can make text read differently from how it is stored.
 func isBidiControl(r rune) bool {
-	return (r >= 0x202a && r <= 0x202e) || (r >= 0x2066 && r <= 0x2069)
+	return r == 0x200e || r == 0x200f || r == 0x061c ||
+		(r >= 0x202a && r <= 0x202e) || (r >= 0x2066 && r <= 0x2069)
 }
 // FieldViolation names a request field that breaks a rule.

M internal/domain/text_test.go +11 -1


@@ -39,6 +39,11 @@ func TestParseMRTitle(t *testing.T) {
39	39		{"bidi embedding", "a\u202ab", "U+202A"},
40	40		{"bidi isolate", "a\u2066b", "U+2066"},
41	41		{"bidi pop isolate", "a\u2069b", "U+2069"},
	42	+	{"line separator", "a\u2028b", "U+2028"},
	43	+	{"paragraph separator", "a\u2029b", "U+2029"},
	44	+	{"left-to-right mark", "a\u200eb", "U+200E"},
	45	+	{"right-to-left mark", "a\u200fb", "U+200F"},
	46	+	{"arabic letter mark", "a\u061cb", "U+061C"},
42	47		{"zero width joiner is plain text", "a\u200db", ""},
43	48		}
44	49		for _, tc := range tests {

@@ -78,7 +83,12 @@ func TestParsePlainText(t *testing.T) {
78	83		{"bidi override", "x\u202ey", "U+202E"},
79	84		{"bidi left-to-right isolate", "x\u2066y", "U+2066"},
80	85		{"bidi first strong isolate", "x\u2068y", "U+2068"},
81		-	{"left-to-right mark is plain text", "x\u200ey", ""},
	86	+	{"line separator", "x\u2028y", "U+2028"},
	87	+	{"paragraph separator", "x\u2029y", "U+2029"},
	88	+	{"left-to-right mark", "x\u200ey", "U+200E"},
	89	+	{"right-to-left mark", "x\u200fy", "U+200F"},
	90	+	{"arabic letter mark", "x\u061cy", "U+061C"},
	91	+	{"zero width joiner is plain text", "x\u200dy", ""},
82	92		}
83	93		for _, tc := range tests {
84	94		t.Run(tc.name, func(t *testing.T) {

 		{"bidi embedding", "a\u202ab", "U+202A"},
 		{"bidi isolate", "a\u2066b", "U+2066"},
 		{"bidi pop isolate", "a\u2069b", "U+2069"},
+		{"line separator", "a\u2028b", "U+2028"},
+		{"paragraph separator", "a\u2029b", "U+2029"},
+		{"left-to-right mark", "a\u200eb", "U+200E"},
+		{"right-to-left mark", "a\u200fb", "U+200F"},
+		{"arabic letter mark", "a\u061cb", "U+061C"},
 		{"zero width joiner is plain text", "a\u200db", ""},
 	}
 	for _, tc := range tests {
 		{"bidi override", "x\u202ey", "U+202E"},
 		{"bidi left-to-right isolate", "x\u2066y", "U+2066"},
 		{"bidi first strong isolate", "x\u2068y", "U+2068"},
-		{"left-to-right mark is plain text", "x\u200ey", ""},
+		{"line separator", "x\u2028y", "U+2028"},
+		{"paragraph separator", "x\u2029y", "U+2029"},
+		{"left-to-right mark", "x\u200ey", "U+200E"},
+		{"right-to-left mark", "x\u200fy", "U+200F"},
+		{"arabic letter mark", "x\u061cy", "U+061C"},
+		{"zero width joiner is plain text", "x\u200dy", ""},
 	}
 	for _, tc := range tests {
 		t.Run(tc.name, func(t *testing.T) {