Home | History | Annotate | Download | only in strings
      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 #include <type_traits>
     14 
     15 #include "base/logging.h"
     16 #include "base/numerics/safe_math.h"
     17 #include "base/scoped_clear_errno.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     typename std::make_unsigned<INT>::type res =
     39         CheckedNumeric<INT>(value).UnsignedAbs().ValueOrDie();
     40 
     41     CHR* end = outbuf + kOutputBufSize;
     42     CHR* i = end;
     43     do {
     44       --i;
     45       DCHECK(i != outbuf);
     46       *i = static_cast<CHR>((res % 10) + '0');
     47       res /= 10;
     48     } while (res != 0);
     49     if (IsValueNegative(value)) {
     50       --i;
     51       DCHECK(i != outbuf);
     52       *i = static_cast<CHR>('-');
     53     }
     54     return STR(i, end);
     55   }
     56 };
     57 
     58 // Utility to convert a character to a digit in a given base
     59 template<typename CHAR, int BASE, bool BASE_LTE_10> class BaseCharToDigit {
     60 };
     61 
     62 // Faster specialization for bases <= 10
     63 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, true> {
     64  public:
     65   static bool Convert(CHAR c, uint8_t* digit) {
     66     if (c >= '0' && c < '0' + BASE) {
     67       *digit = static_cast<uint8_t>(c - '0');
     68       return true;
     69     }
     70     return false;
     71   }
     72 };
     73 
     74 // Specialization for bases where 10 < base <= 36
     75 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, false> {
     76  public:
     77   static bool Convert(CHAR c, uint8_t* digit) {
     78     if (c >= '0' && c <= '9') {
     79       *digit = c - '0';
     80     } else if (c >= 'a' && c < 'a' + BASE - 10) {
     81       *digit = c - 'a' + 10;
     82     } else if (c >= 'A' && c < 'A' + BASE - 10) {
     83       *digit = c - 'A' + 10;
     84     } else {
     85       return false;
     86     }
     87     return true;
     88   }
     89 };
     90 
     91 template <int BASE, typename CHAR>
     92 bool CharToDigit(CHAR c, uint8_t* digit) {
     93   return BaseCharToDigit<CHAR, BASE, BASE <= 10>::Convert(c, digit);
     94 }
     95 
     96 // There is an IsUnicodeWhitespace for wchars defined in string_util.h, but it
     97 // is locale independent, whereas the functions we are replacing were
     98 // locale-dependent. TBD what is desired, but for the moment let's not
     99 // introduce a change in behaviour.
    100 template<typename CHAR> class WhitespaceHelper {
    101 };
    102 
    103 template<> class WhitespaceHelper<char> {
    104  public:
    105   static bool Invoke(char c) {
    106     return 0 != isspace(static_cast<unsigned char>(c));
    107   }
    108 };
    109 
    110 template<> class WhitespaceHelper<char16> {
    111  public:
    112   static bool Invoke(char16 c) {
    113     return 0 != iswspace(c);
    114   }
    115 };
    116 
    117 template<typename CHAR> bool LocalIsWhitespace(CHAR c) {
    118   return WhitespaceHelper<CHAR>::Invoke(c);
    119 }
    120 
    121 // IteratorRangeToNumberTraits should provide:
    122 //  - a typedef for iterator_type, the iterator type used as input.
    123 //  - a typedef for value_type, the target numeric type.
    124 //  - static functions min, max (returning the minimum and maximum permitted
    125 //    values)
    126 //  - constant kBase, the base in which to interpret the input
    127 template<typename IteratorRangeToNumberTraits>
    128 class IteratorRangeToNumber {
    129  public:
    130   typedef IteratorRangeToNumberTraits traits;
    131   typedef typename traits::iterator_type const_iterator;
    132   typedef typename traits::value_type value_type;
    133 
    134   // Generalized iterator-range-to-number conversion.
    135   //
    136   static bool Invoke(const_iterator begin,
    137                      const_iterator end,
    138                      value_type* output) {
    139     bool valid = true;
    140 
    141     while (begin != end && LocalIsWhitespace(*begin)) {
    142       valid = false;
    143       ++begin;
    144     }
    145 
    146     if (begin != end && *begin == '-') {
    147       if (!std::numeric_limits<value_type>::is_signed) {
    148         *output = 0;
    149         valid = false;
    150       } else if (!Negative::Invoke(begin + 1, end, output)) {
    151         valid = false;
    152       }
    153     } else {
    154       if (begin != end && *begin == '+') {
    155         ++begin;
    156       }
    157       if (!Positive::Invoke(begin, end, output)) {
    158         valid = false;
    159       }
    160     }
    161 
    162     return valid;
    163   }
    164 
    165  private:
    166   // Sign provides:
    167   //  - a static function, CheckBounds, that determines whether the next digit
    168   //    causes an overflow/underflow
    169   //  - a static function, Increment, that appends the next digit appropriately
    170   //    according to the sign of the number being parsed.
    171   template<typename Sign>
    172   class Base {
    173    public:
    174     static bool Invoke(const_iterator begin, const_iterator end,
    175                        typename traits::value_type* output) {
    176       *output = 0;
    177 
    178       if (begin == end) {
    179         return false;
    180       }
    181 
    182       // Note: no performance difference was found when using template
    183       // specialization to remove this check in bases other than 16
    184       if (traits::kBase == 16 && end - begin > 2 && *begin == '0' &&
    185           (*(begin + 1) == 'x' || *(begin + 1) == 'X')) {
    186         begin += 2;
    187       }
    188 
    189       for (const_iterator current = begin; current != end; ++current) {
    190         uint8_t new_digit = 0;
    191 
    192         if (!CharToDigit<traits::kBase>(*current, &new_digit)) {
    193           return false;
    194         }
    195 
    196         if (current != begin) {
    197           if (!Sign::CheckBounds(output, new_digit)) {
    198             return false;
    199           }
    200           *output *= traits::kBase;
    201         }
    202 
    203         Sign::Increment(new_digit, output);
    204       }
    205       return true;
    206     }
    207   };
    208 
    209   class Positive : public Base<Positive> {
    210    public:
    211     static bool CheckBounds(value_type* output, uint8_t new_digit) {
    212       if (*output > static_cast<value_type>(traits::max() / traits::kBase) ||
    213           (*output == static_cast<value_type>(traits::max() / traits::kBase) &&
    214            new_digit > traits::max() % traits::kBase)) {
    215         *output = traits::max();
    216         return false;
    217       }
    218       return true;
    219     }
    220     static void Increment(uint8_t increment, value_type* output) {
    221       *output += increment;
    222     }
    223   };
    224 
    225   class Negative : public Base<Negative> {
    226    public:
    227     static bool CheckBounds(value_type* output, uint8_t new_digit) {
    228       if (*output < traits::min() / traits::kBase ||
    229           (*output == traits::min() / traits::kBase &&
    230            new_digit > 0 - traits::min() % traits::kBase)) {
    231         *output = traits::min();
    232         return false;
    233       }
    234       return true;
    235     }
    236     static void Increment(uint8_t increment, value_type* output) {
    237       *output -= increment;
    238     }
    239   };
    240 };
    241 
    242 template<typename ITERATOR, typename VALUE, int BASE>
    243 class BaseIteratorRangeToNumberTraits {
    244  public:
    245   typedef ITERATOR iterator_type;
    246   typedef VALUE value_type;
    247   static value_type min() {
    248     return std::numeric_limits<value_type>::min();
    249   }
    250   static value_type max() {
    251     return std::numeric_limits<value_type>::max();
    252   }
    253   static const int kBase = BASE;
    254 };
    255 
    256 template<typename ITERATOR>
    257 class BaseHexIteratorRangeToIntTraits
    258     : public BaseIteratorRangeToNumberTraits<ITERATOR, int, 16> {
    259 };
    260 
    261 template <typename ITERATOR>
    262 class BaseHexIteratorRangeToUIntTraits
    263     : public BaseIteratorRangeToNumberTraits<ITERATOR, uint32_t, 16> {};
    264 
    265 template <typename ITERATOR>
    266 class BaseHexIteratorRangeToInt64Traits
    267     : public BaseIteratorRangeToNumberTraits<ITERATOR, int64_t, 16> {};
    268 
    269 template <typename ITERATOR>
    270 class BaseHexIteratorRangeToUInt64Traits
    271     : public BaseIteratorRangeToNumberTraits<ITERATOR, uint64_t, 16> {};
    272 
    273 typedef BaseHexIteratorRangeToIntTraits<StringPiece::const_iterator>
    274     HexIteratorRangeToIntTraits;
    275 
    276 typedef BaseHexIteratorRangeToUIntTraits<StringPiece::const_iterator>
    277     HexIteratorRangeToUIntTraits;
    278 
    279 typedef BaseHexIteratorRangeToInt64Traits<StringPiece::const_iterator>
    280     HexIteratorRangeToInt64Traits;
    281 
    282 typedef BaseHexIteratorRangeToUInt64Traits<StringPiece::const_iterator>
    283     HexIteratorRangeToUInt64Traits;
    284 
    285 template <typename STR>
    286 bool HexStringToBytesT(const STR& input, std::vector<uint8_t>* output) {
    287   DCHECK_EQ(output->size(), 0u);
    288   size_t count = input.size();
    289   if (count == 0 || (count % 2) != 0)
    290     return false;
    291   for (uintptr_t i = 0; i < count / 2; ++i) {
    292     uint8_t msb = 0;  // most significant 4 bits
    293     uint8_t lsb = 0;  // least significant 4 bits
    294     if (!CharToDigit<16>(input[i * 2], &msb) ||
    295         !CharToDigit<16>(input[i * 2 + 1], &lsb))
    296       return false;
    297     output->push_back((msb << 4) | lsb);
    298   }
    299   return true;
    300 }
    301 
    302 template <typename VALUE, int BASE>
    303 class StringPieceToNumberTraits
    304     : public BaseIteratorRangeToNumberTraits<StringPiece::const_iterator,
    305                                              VALUE,
    306                                              BASE> {
    307 };
    308 
    309 template <typename VALUE>
    310 bool StringToIntImpl(const StringPiece& input, VALUE* output) {
    311   return IteratorRangeToNumber<StringPieceToNumberTraits<VALUE, 10> >::Invoke(
    312       input.begin(), input.end(), output);
    313 }
    314 
    315 template <typename VALUE, int BASE>
    316 class StringPiece16ToNumberTraits
    317     : public BaseIteratorRangeToNumberTraits<StringPiece16::const_iterator,
    318                                              VALUE,
    319                                              BASE> {
    320 };
    321 
    322 template <typename VALUE>
    323 bool String16ToIntImpl(const StringPiece16& input, VALUE* output) {
    324   return IteratorRangeToNumber<StringPiece16ToNumberTraits<VALUE, 10> >::Invoke(
    325       input.begin(), input.end(), output);
    326 }
    327 
    328 }  // namespace
    329 
    330 std::string IntToString(int value) {
    331   return IntToStringT<std::string, int>::IntToString(value);
    332 }
    333 
    334 string16 IntToString16(int value) {
    335   return IntToStringT<string16, int>::IntToString(value);
    336 }
    337 
    338 std::string UintToString(unsigned int value) {
    339   return IntToStringT<std::string, unsigned int>::IntToString(value);
    340 }
    341 
    342 string16 UintToString16(unsigned int value) {
    343   return IntToStringT<string16, unsigned int>::IntToString(value);
    344 }
    345 
    346 std::string Int64ToString(int64_t value) {
    347   return IntToStringT<std::string, int64_t>::IntToString(value);
    348 }
    349 
    350 string16 Int64ToString16(int64_t value) {
    351   return IntToStringT<string16, int64_t>::IntToString(value);
    352 }
    353 
    354 std::string Uint64ToString(uint64_t value) {
    355   return IntToStringT<std::string, uint64_t>::IntToString(value);
    356 }
    357 
    358 string16 Uint64ToString16(uint64_t value) {
    359   return IntToStringT<string16, uint64_t>::IntToString(value);
    360 }
    361 
    362 std::string SizeTToString(size_t value) {
    363   return IntToStringT<std::string, size_t>::IntToString(value);
    364 }
    365 
    366 string16 SizeTToString16(size_t value) {
    367   return IntToStringT<string16, size_t>::IntToString(value);
    368 }
    369 
    370 std::string DoubleToString(double value) {
    371   auto ret = std::to_string(value);
    372   // If this returned an integer, don't do anything.
    373   if (ret.find('.') == std::string::npos) {
    374     return ret;
    375   }
    376   // Otherwise, it has an annoying tendency to leave trailing zeros.
    377   size_t len = ret.size();
    378   while (len >= 2 && ret[len - 1] == '0' && ret[len - 2] != '.') {
    379     --len;
    380   }
    381   ret.erase(len);
    382   return ret;
    383 }
    384 
    385 bool StringToInt(const StringPiece& input, int* output) {
    386   return StringToIntImpl(input, output);
    387 }
    388 
    389 bool StringToInt(const StringPiece16& input, int* output) {
    390   return String16ToIntImpl(input, output);
    391 }
    392 
    393 bool StringToUint(const StringPiece& input, unsigned* output) {
    394   return StringToIntImpl(input, output);
    395 }
    396 
    397 bool StringToUint(const StringPiece16& input, unsigned* output) {
    398   return String16ToIntImpl(input, output);
    399 }
    400 
    401 bool StringToInt64(const StringPiece& input, int64_t* output) {
    402   return StringToIntImpl(input, output);
    403 }
    404 
    405 bool StringToInt64(const StringPiece16& input, int64_t* output) {
    406   return String16ToIntImpl(input, output);
    407 }
    408 
    409 bool StringToUint64(const StringPiece& input, uint64_t* output) {
    410   return StringToIntImpl(input, output);
    411 }
    412 
    413 bool StringToUint64(const StringPiece16& input, uint64_t* output) {
    414   return String16ToIntImpl(input, output);
    415 }
    416 
    417 bool StringToSizeT(const StringPiece& input, size_t* output) {
    418   return StringToIntImpl(input, output);
    419 }
    420 
    421 bool StringToSizeT(const StringPiece16& input, size_t* output) {
    422   return String16ToIntImpl(input, output);
    423 }
    424 
    425 bool StringToDouble(const std::string& input, double* output) {
    426   char* endptr = nullptr;
    427   *output = strtod(input.c_str(), &endptr);
    428 
    429   // Cases to return false:
    430   //  - If the input string is empty, there was nothing to parse.
    431   //  - If endptr does not point to the end of the string, there are either
    432   //    characters remaining in the string after a parsed number, or the string
    433   //    does not begin with a parseable number.  endptr is compared to the
    434   //    expected end given the string's stated length to correctly catch cases
    435   //    where the string contains embedded NUL characters.
    436   //  - If the first character is a space, there was leading whitespace
    437   return !input.empty() &&
    438          input.c_str() + input.length() == endptr &&
    439          !isspace(input[0]) &&
    440          *output != std::numeric_limits<double>::infinity() &&
    441          *output != -std::numeric_limits<double>::infinity();
    442 }
    443 
    444 // Note: if you need to add String16ToDouble, first ask yourself if it's
    445 // really necessary. If it is, probably the best implementation here is to
    446 // convert to 8-bit and then use the 8-bit version.
    447 
    448 // Note: if you need to add an iterator range version of StringToDouble, first
    449 // ask yourself if it's really necessary. If it is, probably the best
    450 // implementation here is to instantiate a string and use the string version.
    451 
    452 std::string HexEncode(const void* bytes, size_t size) {
    453   static const char kHexChars[] = "0123456789ABCDEF";
    454 
    455   // Each input byte creates two output hex characters.
    456   std::string ret(size * 2, '\0');
    457 
    458   for (size_t i = 0; i < size; ++i) {
    459     char b = reinterpret_cast<const char*>(bytes)[i];
    460     ret[(i * 2)] = kHexChars[(b >> 4) & 0xf];
    461     ret[(i * 2) + 1] = kHexChars[b & 0xf];
    462   }
    463   return ret;
    464 }
    465 
    466 bool HexStringToInt(const StringPiece& input, int* output) {
    467   return IteratorRangeToNumber<HexIteratorRangeToIntTraits>::Invoke(
    468     input.begin(), input.end(), output);
    469 }
    470 
    471 bool HexStringToUInt(const StringPiece& input, uint32_t* output) {
    472   return IteratorRangeToNumber<HexIteratorRangeToUIntTraits>::Invoke(
    473       input.begin(), input.end(), output);
    474 }
    475 
    476 bool HexStringToInt64(const StringPiece& input, int64_t* output) {
    477   return IteratorRangeToNumber<HexIteratorRangeToInt64Traits>::Invoke(
    478     input.begin(), input.end(), output);
    479 }
    480 
    481 bool HexStringToUInt64(const StringPiece& input, uint64_t* output) {
    482   return IteratorRangeToNumber<HexIteratorRangeToUInt64Traits>::Invoke(
    483       input.begin(), input.end(), output);
    484 }
    485 
    486 bool HexStringToBytes(const std::string& input, std::vector<uint8_t>* output) {
    487   return HexStringToBytesT(input, output);
    488 }
    489 
    490 }  // namespace base
    491