Replaces embedded NUL bytes and invalid UTF-8 byte sequences with '?' to avoid Xerces treating NUL as a string terminator and to keep input reasonably UTF-8-like.
| 43 | // Replaces embedded NUL bytes and invalid UTF-8 byte sequences with '?' to avoid Xerces treating |
| 44 | // NUL as a string terminator and to keep input reasonably UTF-8-like. |
| 45 | void sanitizeUtf8Bytes(XMLByte* const toFill, const std::size_t len) |
| 46 | { |
| 47 | auto* data = reinterpret_cast<std::uint8_t*>( |
| 48 | toFill |
| 49 | ); // NOLINT(cppcoreguidelines-pro-type-reinterpret-cast) |
| 50 | |
| 51 | auto markInvalidByte = [&](std::size_t pos) { |
| 52 | data[pos] = static_cast<std::uint8_t>('?'); |
| 53 | }; |
| 54 | |
| 55 | auto isCont = [&](std::uint8_t b) { |
| 56 | return (b & 0xC0) == 0x80; |
| 57 | }; |
| 58 | |
| 59 | std::size_t i = 0; |
| 60 | while (i < len) { |
| 61 | const std::uint8_t b0 = data[i]; |
| 62 | |
| 63 | if (b0 == 0) { |
| 64 | markInvalidByte(i); |
| 65 | ++i; |
| 66 | continue; |
| 67 | } |
| 68 | |
| 69 | if (b0 < 0x80) { |
| 70 | ++i; |
| 71 | continue; |
| 72 | } |
| 73 | |
| 74 | // Reject stray continuation bytes |
| 75 | if (isCont(b0)) { |
| 76 | markInvalidByte(i); |
| 77 | ++i; |
| 78 | continue; |
| 79 | } |
| 80 | |
| 81 | // 2-byte sequence: C2..DF 80..BF |
| 82 | if (b0 >= 0xC2 && b0 <= 0xDF) { |
| 83 | if (i + 1 >= len || !isCont(data[i + 1])) { |
| 84 | markInvalidByte(i); |
| 85 | ++i; |
| 86 | continue; |
| 87 | } |
| 88 | i += 2; |
| 89 | continue; |
| 90 | } |
| 91 | |
| 92 | // 3-byte sequences |
| 93 | if (b0 >= 0xE0 && b0 <= 0xEF) { |
| 94 | if (i + 2 >= len) { |
| 95 | markInvalidByte(i); |
| 96 | ++i; |
| 97 | continue; |
| 98 | } |
| 99 | |
| 100 | const std::uint8_t b1 = data[i + 1]; |
| 101 | const std::uint8_t b2 = data[i + 2]; |
| 102 | if (!isCont(b1) || !isCont(b2)) { |