1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 #include "base/strings/string_number_conversions.h" 6 7 #include <ctype.h> 8 #include <errno.h> 9 #include <stdlib.h> 10 #include <wctype.h> 11 12 #include <limits> 13 14 #include "base/logging.h" 15 #include "base/numerics/safe_conversions.h" 16 #include "base/numerics/safe_math.h" 17 #include "base/strings/utf_string_conversions.h" 18 19 namespace base { 20 21 namespace { 22 23 template <typename STR, typename INT> 24 struct IntToStringT { 25 static STR IntToString(INT value) { 26 // log10(2) ~= 0.3 bytes needed per bit or per byte log10(2**8) ~= 2.4. 27 // So round up to allocate 3 output characters per byte, plus 1 for '-'. 28 const size_t kOutputBufSize = 29 3 * sizeof(INT) + std::numeric_limits<INT>::is_signed; 30 31 // Create the string in a temporary buffer, write it back to front, and 32 // then return the substr of what we ended up using. 33 using CHR = typename STR::value_type; 34 CHR outbuf[kOutputBufSize]; 35 36 // The ValueOrDie call below can never fail, because UnsignedAbs is valid 37 // for all valid inputs. 38 auto res = CheckedNumeric<INT>(value).UnsignedAbs().ValueOrDie(); 39 40 CHR* end = outbuf + kOutputBufSize; 41 CHR* i = end; 42 do { 43 --i; 44 DCHECK(i != outbuf); 45 *i = static_cast<CHR>((res % 10) + '0'); 46 res /= 10; 47 } while (res != 0); 48 if (IsValueNegative(value)) { 49 --i; 50 DCHECK(i != outbuf); 51 *i = static_cast<CHR>('-'); 52 } 53 return STR(i, end); 54 } 55 }; 56 57 // Utility to convert a character to a digit in a given base 58 template<typename CHAR, int BASE, bool BASE_LTE_10> class BaseCharToDigit { 59 }; 60 61 // Faster specialization for bases <= 10 62 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, true> { 63 public: 64 static bool Convert(CHAR c, uint8_t* digit) { 65 if (c >= '0' && c < '0' + BASE) { 66 *digit = static_cast<uint8_t>(c - '0'); 67 return true; 68 } 69 return false; 70 } 71 }; 72 73 // Specialization for bases where 10 < base <= 36 74 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, false> { 75 public: 76 static bool Convert(CHAR c, uint8_t* digit) { 77 if (c >= '0' && c <= '9') { 78 *digit = c - '0'; 79 } else if (c >= 'a' && c < 'a' + BASE - 10) { 80 *digit = c - 'a' + 10; 81 } else if (c >= 'A' && c < 'A' + BASE - 10) { 82 *digit = c - 'A' + 10; 83 } else { 84 return false; 85 } 86 return true; 87 } 88 }; 89 90 template <int BASE, typename CHAR> 91 bool CharToDigit(CHAR c, uint8_t* digit) { 92 return BaseCharToDigit<CHAR, BASE, BASE <= 10>::Convert(c, digit); 93 } 94 95 // There is an IsUnicodeWhitespace for wchars defined in string_util.h, but it 96 // is locale independent, whereas the functions we are replacing were 97 // locale-dependent. TBD what is desired, but for the moment let's not 98 // introduce a change in behaviour. 99 template<typename CHAR> class WhitespaceHelper { 100 }; 101 102 template<> class WhitespaceHelper<char> { 103 public: 104 static bool Invoke(char c) { 105 return 0 != isspace(static_cast<unsigned char>(c)); 106 } 107 }; 108 109 template<> class WhitespaceHelper<char16> { 110 public: 111 static bool Invoke(char16 c) { 112 return 0 != iswspace(c); 113 } 114 }; 115 116 template<typename CHAR> bool LocalIsWhitespace(CHAR c) { 117 return WhitespaceHelper<CHAR>::Invoke(c); 118 } 119 120 // IteratorRangeToNumberTraits should provide: 121 // - a typedef for iterator_type, the iterator type used as input. 122 // - a typedef for value_type, the target numeric type. 123 // - static functions min, max (returning the minimum and maximum permitted 124 // values) 125 // - constant kBase, the base in which to interpret the input 126 template<typename IteratorRangeToNumberTraits> 127 class IteratorRangeToNumber { 128 public: 129 typedef IteratorRangeToNumberTraits traits; 130 typedef typename traits::iterator_type const_iterator; 131 typedef typename traits::value_type value_type; 132 133 // Generalized iterator-range-to-number conversion. 134 // 135 static bool Invoke(const_iterator begin, 136 const_iterator end, 137 value_type* output) { 138 bool valid = true; 139 140 while (begin != end && LocalIsWhitespace(*begin)) { 141 valid = false; 142 ++begin; 143 } 144 145 if (begin != end && *begin == '-') { 146 if (!std::numeric_limits<value_type>::is_signed) { 147 valid = false; 148 } else if (!Negative::Invoke(begin + 1, end, output)) { 149 valid = false; 150 } 151 } else { 152 if (begin != end && *begin == '+') { 153 ++begin; 154 } 155 if (!Positive::Invoke(begin, end, output)) { 156 valid = false; 157 } 158 } 159 160 return valid; 161 } 162 163 private: 164 // Sign provides: 165 // - a static function, CheckBounds, that determines whether the next digit 166 // causes an overflow/underflow 167 // - a static function, Increment, that appends the next digit appropriately 168 // according to the sign of the number being parsed. 169 template<typename Sign> 170 class Base { 171 public: 172 static bool Invoke(const_iterator begin, const_iterator end, 173 typename traits::value_type* output) { 174 *output = 0; 175 176 if (begin == end) { 177 return false; 178 } 179 180 // Note: no performance difference was found when using template 181 // specialization to remove this check in bases other than 16 182 if (traits::kBase == 16 && end - begin > 2 && *begin == '0' && 183 (*(begin + 1) == 'x' || *(begin + 1) == 'X')) { 184 begin += 2; 185 } 186 187 for (const_iterator current = begin; current != end; ++current) { 188 uint8_t new_digit = 0; 189 190 if (!CharToDigit<traits::kBase>(*current, &new_digit)) { 191 return false; 192 } 193 194 if (current != begin) { 195 if (!Sign::CheckBounds(output, new_digit)) { 196 return false; 197 } 198 *output *= traits::kBase; 199 } 200 201 Sign::Increment(new_digit, output); 202 } 203 return true; 204 } 205 }; 206 207 class Positive : public Base<Positive> { 208 public: 209 static bool CheckBounds(value_type* output, uint8_t new_digit) { 210 if (*output > static_cast<value_type>(traits::max() / traits::kBase) || 211 (*output == static_cast<value_type>(traits::max() / traits::kBase) && 212 new_digit > traits::max() % traits::kBase)) { 213 *output = traits::max(); 214 return false; 215 } 216 return true; 217 } 218 static void Increment(uint8_t increment, value_type* output) { 219 *output += increment; 220 } 221 }; 222 223 class Negative : public Base<Negative> { 224 public: 225 static bool CheckBounds(value_type* output, uint8_t new_digit) { 226 if (*output < traits::min() / traits::kBase || 227 (*output == traits::min() / traits::kBase && 228 new_digit > 0 - traits::min() % traits::kBase)) { 229 *output = traits::min(); 230 return false; 231 } 232 return true; 233 } 234 static void Increment(uint8_t increment, value_type* output) { 235 *output -= increment; 236 } 237 }; 238 }; 239 240 template<typename ITERATOR, typename VALUE, int BASE> 241 class BaseIteratorRangeToNumberTraits { 242 public: 243 typedef ITERATOR iterator_type; 244 typedef VALUE value_type; 245 static value_type min() { 246 return std::numeric_limits<value_type>::min(); 247 } 248 static value_type max() { 249 return std::numeric_limits<value_type>::max(); 250 } 251 static const int kBase = BASE; 252 }; 253 254 template<typename ITERATOR> 255 class BaseHexIteratorRangeToIntTraits 256 : public BaseIteratorRangeToNumberTraits<ITERATOR, int, 16> { 257 }; 258 259 template <typename ITERATOR> 260 class BaseHexIteratorRangeToUIntTraits 261 : public BaseIteratorRangeToNumberTraits<ITERATOR, uint32_t, 16> {}; 262 263 template <typename ITERATOR> 264 class BaseHexIteratorRangeToInt64Traits 265 : public BaseIteratorRangeToNumberTraits<ITERATOR, int64_t, 16> {}; 266 267 template <typename ITERATOR> 268 class BaseHexIteratorRangeToUInt64Traits 269 : public BaseIteratorRangeToNumberTraits<ITERATOR, uint64_t, 16> {}; 270 271 typedef BaseHexIteratorRangeToIntTraits<StringPiece::const_iterator> 272 HexIteratorRangeToIntTraits; 273 274 typedef BaseHexIteratorRangeToUIntTraits<StringPiece::const_iterator> 275 HexIteratorRangeToUIntTraits; 276 277 typedef BaseHexIteratorRangeToInt64Traits<StringPiece::const_iterator> 278 HexIteratorRangeToInt64Traits; 279 280 typedef BaseHexIteratorRangeToUInt64Traits<StringPiece::const_iterator> 281 HexIteratorRangeToUInt64Traits; 282 283 template <typename STR> 284 bool HexStringToBytesT(const STR& input, std::vector<uint8_t>* output) { 285 DCHECK_EQ(output->size(), 0u); 286 size_t count = input.size(); 287 if (count == 0 || (count % 2) != 0) 288 return false; 289 for (uintptr_t i = 0; i < count / 2; ++i) { 290 uint8_t msb = 0; // most significant 4 bits 291 uint8_t lsb = 0; // least significant 4 bits 292 if (!CharToDigit<16>(input[i * 2], &msb) || 293 !CharToDigit<16>(input[i * 2 + 1], &lsb)) 294 return false; 295 output->push_back((msb << 4) | lsb); 296 } 297 return true; 298 } 299 300 template <typename VALUE, int BASE> 301 class StringPieceToNumberTraits 302 : public BaseIteratorRangeToNumberTraits<StringPiece::const_iterator, 303 VALUE, 304 BASE> { 305 }; 306 307 template <typename VALUE> 308 bool StringToIntImpl(const StringPiece& input, VALUE* output) { 309 return IteratorRangeToNumber<StringPieceToNumberTraits<VALUE, 10> >::Invoke( 310 input.begin(), input.end(), output); 311 } 312 313 template <typename VALUE, int BASE> 314 class StringPiece16ToNumberTraits 315 : public BaseIteratorRangeToNumberTraits<StringPiece16::const_iterator, 316 VALUE, 317 BASE> { 318 }; 319 320 template <typename VALUE> 321 bool String16ToIntImpl(const StringPiece16& input, VALUE* output) { 322 return IteratorRangeToNumber<StringPiece16ToNumberTraits<VALUE, 10> >::Invoke( 323 input.begin(), input.end(), output); 324 } 325 326 } // namespace 327 328 std::string IntToString(int value) { 329 return IntToStringT<std::string, int>::IntToString(value); 330 } 331 332 string16 IntToString16(int value) { 333 return IntToStringT<string16, int>::IntToString(value); 334 } 335 336 std::string UintToString(unsigned int value) { 337 return IntToStringT<std::string, unsigned int>::IntToString(value); 338 } 339 340 string16 UintToString16(unsigned int value) { 341 return IntToStringT<string16, unsigned int>::IntToString(value); 342 } 343 344 std::string Int64ToString(int64_t value) { 345 return IntToStringT<std::string, int64_t>::IntToString(value); 346 } 347 348 string16 Int64ToString16(int64_t value) { 349 return IntToStringT<string16, int64_t>::IntToString(value); 350 } 351 352 std::string Uint64ToString(uint64_t value) { 353 return IntToStringT<std::string, uint64_t>::IntToString(value); 354 } 355 356 string16 Uint64ToString16(uint64_t value) { 357 return IntToStringT<string16, uint64_t>::IntToString(value); 358 } 359 360 std::string SizeTToString(size_t value) { 361 return IntToStringT<std::string, size_t>::IntToString(value); 362 } 363 364 string16 SizeTToString16(size_t value) { 365 return IntToStringT<string16, size_t>::IntToString(value); 366 } 367 368 std::string DoubleToString(double value) { 369 auto ret = std::to_string(value); 370 // If this returned an integer, don't do anything. 371 if (ret.find('.') == std::string::npos) { 372 return ret; 373 } 374 // Otherwise, it has an annoying tendency to leave trailing zeros. 375 size_t len = ret.size(); 376 while (len >= 2 && ret[len - 1] == '0' && ret[len - 2] != '.') { 377 --len; 378 } 379 ret.erase(len); 380 return ret; 381 } 382 383 bool StringToInt(const StringPiece& input, int* output) { 384 return StringToIntImpl(input, output); 385 } 386 387 bool StringToInt(const StringPiece16& input, int* output) { 388 return String16ToIntImpl(input, output); 389 } 390 391 bool StringToUint(const StringPiece& input, unsigned* output) { 392 return StringToIntImpl(input, output); 393 } 394 395 bool StringToUint(const StringPiece16& input, unsigned* output) { 396 return String16ToIntImpl(input, output); 397 } 398 399 bool StringToInt64(const StringPiece& input, int64_t* output) { 400 return StringToIntImpl(input, output); 401 } 402 403 bool StringToInt64(const StringPiece16& input, int64_t* output) { 404 return String16ToIntImpl(input, output); 405 } 406 407 bool StringToUint64(const StringPiece& input, uint64_t* output) { 408 return StringToIntImpl(input, output); 409 } 410 411 bool StringToUint64(const StringPiece16& input, uint64_t* output) { 412 return String16ToIntImpl(input, output); 413 } 414 415 bool StringToSizeT(const StringPiece& input, size_t* output) { 416 return StringToIntImpl(input, output); 417 } 418 419 bool StringToSizeT(const StringPiece16& input, size_t* output) { 420 return String16ToIntImpl(input, output); 421 } 422 423 bool StringToDouble(const std::string& input, double* output) { 424 char* endptr = nullptr; 425 *output = strtod(input.c_str(), &endptr); 426 427 // Cases to return false: 428 // - If the input string is empty, there was nothing to parse. 429 // - If endptr does not point to the end of the string, there are either 430 // characters remaining in the string after a parsed number, or the string 431 // does not begin with a parseable number. endptr is compared to the 432 // expected end given the string's stated length to correctly catch cases 433 // where the string contains embedded NUL characters. 434 // - If the first character is a space, there was leading whitespace 435 return !input.empty() && 436 input.c_str() + input.length() == endptr && 437 !isspace(input[0]) && 438 *output != std::numeric_limits<double>::infinity() && 439 *output != -std::numeric_limits<double>::infinity(); 440 } 441 442 // Note: if you need to add String16ToDouble, first ask yourself if it's 443 // really necessary. If it is, probably the best implementation here is to 444 // convert to 8-bit and then use the 8-bit version. 445 446 // Note: if you need to add an iterator range version of StringToDouble, first 447 // ask yourself if it's really necessary. If it is, probably the best 448 // implementation here is to instantiate a string and use the string version. 449 450 std::string HexEncode(const void* bytes, size_t size) { 451 static const char kHexChars[] = "0123456789ABCDEF"; 452 453 // Each input byte creates two output hex characters. 454 std::string ret(size * 2, '\0'); 455 456 for (size_t i = 0; i < size; ++i) { 457 char b = reinterpret_cast<const char*>(bytes)[i]; 458 ret[(i * 2)] = kHexChars[(b >> 4) & 0xf]; 459 ret[(i * 2) + 1] = kHexChars[b & 0xf]; 460 } 461 return ret; 462 } 463 464 bool HexStringToInt(const StringPiece& input, int* output) { 465 return IteratorRangeToNumber<HexIteratorRangeToIntTraits>::Invoke( 466 input.begin(), input.end(), output); 467 } 468 469 bool HexStringToUInt(const StringPiece& input, uint32_t* output) { 470 return IteratorRangeToNumber<HexIteratorRangeToUIntTraits>::Invoke( 471 input.begin(), input.end(), output); 472 } 473 474 bool HexStringToInt64(const StringPiece& input, int64_t* output) { 475 return IteratorRangeToNumber<HexIteratorRangeToInt64Traits>::Invoke( 476 input.begin(), input.end(), output); 477 } 478 479 bool HexStringToUInt64(const StringPiece& input, uint64_t* output) { 480 return IteratorRangeToNumber<HexIteratorRangeToUInt64Traits>::Invoke( 481 input.begin(), input.end(), output); 482 } 483 484 bool HexStringToBytes(const std::string& input, std::vector<uint8_t>* output) { 485 return HexStringToBytesT(input, output); 486 } 487 488 } // namespace base 489