AK: Invalidate overlong UTF-8 code point encodings

For example, the code point U+002F could be encoded as UTF-8 with the
bytes 0x80 0xAF. This trick has historically been used to bypass
security checks.

AK/Utf8View.h

@@ -7,7 +7,6 @@
 
 #pragma once
 
-#include <AK/CharacterTypes.h>
 #include <AK/DeprecatedString.h>
 #include <AK/Format.h>
 #include <AK/StringView.h>
@@ -142,7 +141,7 @@ public:
                 code_point |= code_point_bits;
             }
 
-            if (!is_unicode(code_point))
+            if (!is_valid_code_point(code_point, byte_length))
                 return false;
 
             valid_bytes += byte_length;
@@ -162,13 +161,15 @@ private:
         size_t byte_length { 0 };
         u8 encoding_bits { 0 };
         u8 encoding_mask { 0 };
+        u32 first_code_point { 0 };
+        u32 last_code_point { 0 };
     };
 
     static constexpr Array<Utf8EncodedByteData, 4> utf8_encoded_byte_data { {
-        { 1, 0b0000'0000, 0b1000'0000 },
+        { 1, 0b0000'0000, 0b1000'0000, 0x0000, 0x007F },
-        { 2, 0b1100'0000, 0b1110'0000 },
+        { 2, 0b1100'0000, 0b1110'0000, 0x0080, 0x07FF },
-        { 3, 0b1110'0000, 0b1111'0000 },
+        { 3, 0b1110'0000, 0b1111'0000, 0x0800, 0xFFFF },
-        { 4, 0b1111'0000, 0b1111'1000 },
+        { 4, 0b1111'0000, 0b1111'1000, 0x10000, 0x10FFFF },
     } };
 
     struct LeadingByte {
@@ -208,6 +209,16 @@ private:
         return { .is_valid = false };
     }
 
+    static constexpr bool is_valid_code_point(u32 code_point, size_t byte_length)
+    {
+        for (auto const& data : utf8_encoded_byte_data) {
+            if (code_point >= data.first_code_point && code_point <= data.last_code_point)
+                return byte_length == data.byte_length;
+        }
+
+        return false;
+    }
+
     StringView m_string;
     mutable size_t m_length { 0 };
     mutable bool m_have_length { false };

Tests/AK/TestUtf8.cpp

@@ -82,6 +82,47 @@ TEST_CASE(validate_invalid_ut8)
     EXPECT(valid_bytes == 0);
 }
 
+TEST_CASE(validate_overlong_utf8)
+{
+    size_t valid_bytes = 0;
+
+    // Overlong 2-byte encoding of U+002F
+    char invalid_utf8_1[] = { 42, 35, static_cast<char>(0xc0), static_cast<char>(0xaf) };
+    Utf8View utf8_1 { StringView { invalid_utf8_1, sizeof(invalid_utf8_1) } };
+    EXPECT(!utf8_1.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+
+    // Overlong 3-byte encoding of U+002F
+    char invalid_utf8_2[] = { 42, 35, static_cast<char>(0xe0), static_cast<char>(0x80), static_cast<char>(0xaf) };
+    Utf8View utf8_2 { StringView { invalid_utf8_2, sizeof(invalid_utf8_2) } };
+    EXPECT(!utf8_2.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+
+    // Overlong 4-byte encoding of U+002F
+    char invalid_utf8_3[] = { 42, 35, static_cast<char>(0xf0), static_cast<char>(0x80), static_cast<char>(0x80), static_cast<char>(0xaf) };
+    Utf8View utf8_3 { StringView { invalid_utf8_3, sizeof(invalid_utf8_3) } };
+    EXPECT(!utf8_3.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+
+    // Overlong 3-byte encoding of U+00FF
+    char invalid_utf8_4[] = { 42, 35, static_cast<char>(0xe0), static_cast<char>(0x83), static_cast<char>(0xbf) };
+    Utf8View utf8_4 { StringView { invalid_utf8_4, sizeof(invalid_utf8_4) } };
+    EXPECT(!utf8_4.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+
+    // Overlong 4-byte encoding of U+00FF
+    char invalid_utf8_5[] = { 42, 35, static_cast<char>(0xf0), static_cast<char>(0x80), static_cast<char>(0x83), static_cast<char>(0xbf) };
+    Utf8View utf8_5 { StringView { invalid_utf8_5, sizeof(invalid_utf8_5) } };
+    EXPECT(!utf8_5.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+
+    // Overlong 4-byte encoding of U+0FFF
+    char invalid_utf8_6[] = { 42, 35, static_cast<char>(0xf0), static_cast<char>(0x8f), static_cast<char>(0xbf), static_cast<char>(0xbf) };
+    Utf8View utf8_6 { StringView { invalid_utf8_6, sizeof(invalid_utf8_6) } };
+    EXPECT(!utf8_6.validate(valid_bytes));
+    EXPECT(valid_bytes == 2);
+}
+
 TEST_CASE(iterate_utf8)
 {
     Utf8View view("Some weird characters \u00A9\u266A\uA755"sv);