Read a vector of bytes. See "include/loader/filemgr.h".
| 268 | |
| 269 | // Read a vector of bytes. See "include/loader/filemgr.h". |
| 270 | Expect<std::string> FileMgr::readName() { |
| 271 | if (unlikely(Status != ErrCode::Value::Success)) { |
| 272 | return Unexpect(Status); |
| 273 | } |
| 274 | // If UTF-8 validation, readU32(), or readBytes() failed, the last successful |
| 275 | // reading offset will be at the start of `Name`. |
| 276 | LastPos = Pos; |
| 277 | |
| 278 | // Read the name size. |
| 279 | EXPECTED_TRY(uint32_t SizeToRead, readU32()); |
| 280 | |
| 281 | // Check whether the string length exceeds the data boundary. |
| 282 | if (auto Res = testRead(SizeToRead); unlikely(!Res)) { |
| 283 | return Unexpect(ErrCode::Value::LengthOutOfBounds); |
| 284 | } |
| 285 | |
| 286 | // Read the UTF-8 bytes. |
| 287 | std::string Str(SizeToRead, '\0'); |
| 288 | EXPECTED_TRY( |
| 289 | readBytes(Span<Byte>(reinterpret_cast<Byte *>(Str.data()), Str.size()))); |
| 290 | |
| 291 | // UTF-8 validation. |
| 292 | bool Valid = true; |
| 293 | for (uint32_t I = 0; I < Str.size() && Valid; ++I) { |
| 294 | char C = Str.data()[I]; |
| 295 | uint32_t N = 0; |
| 296 | if ((C & '\x80') == 0) { |
| 297 | // 0xxxxxxx, 7 bits UCS, ASCII |
| 298 | N = 0; |
| 299 | } else if ((C & '\xE0') == '\xC0') { |
| 300 | // 110xxxxx, 11 bits UCS, U+80 to U+7FF |
| 301 | N = 1; |
| 302 | } else if ((C & '\xF0') == '\xE0') { |
| 303 | // 1110xxxx, 16 bits UCS, U+800 to U+D7FF and U+E000 to U+FFFF |
| 304 | N = 2; |
| 305 | } else if ((C & '\xF8') == '\xF0') { |
| 306 | // 11110xxx, 21 bits UCS, U+10000 to U+10FFFF |
| 307 | N = 3; |
| 308 | } else { |
| 309 | Valid = false; |
| 310 | } |
| 311 | |
| 312 | // Need N more bytes. |
| 313 | if (I + N >= Str.size()) { |
| 314 | Valid = false; |
| 315 | } |
| 316 | // Invalid ranges |
| 317 | if (N == 1 && (C & '\xDE') == '\xC0') { |
| 318 | // 11 bits UCS, U+0 to U+80, FAIL |
| 319 | Valid = false; |
| 320 | } else if (N == 2 && |
| 321 | ((C == '\xE0' && (Str.data()[I + 1] & '\xA0') == '\x80') || |
| 322 | // 16 bits UCS, U+0 to U+7FF, FAIL |
| 323 | (C == '\xED' && (Str.data()[I + 1] & '\xA0') == '\xA0') |
| 324 | // 16 bits UCS, U+D800 to U+DFFF, FAIL |
| 325 | )) { |
| 326 | Valid = false; |
| 327 | } else if (N == 3 && |