1 // Copyright (c) 2011 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.
5 #include "base/string_number_conversions.h"
14 #include "base/logging.h"
15 #include "base/third_party/dmg_fp/dmg_fp.h"
16 #include "base/utf_string_conversions.h"
22 template <typename STR, typename INT, typename UINT, bool NEG>
24 // This is to avoid a compiler warning about unary minus on unsigned type.
25 // For example, say you had the following code:
26 // template <typename INT>
27 // INT abs(INT value) { return value < 0 ? -value : value; }
28 // Even though if INT is unsigned, it's impossible for value < 0, so the
29 // unary minus will never be taken, the compiler will still generate a
30 // warning. We do a little specialization dance...
31 template <typename INT2, typename UINT2, bool NEG2>
32 struct ToUnsignedT {};
34 template <typename INT2, typename UINT2>
35 struct ToUnsignedT<INT2, UINT2, false> {
36 static UINT2 ToUnsigned(INT2 value) {
37 return static_cast<UINT2>(value);
41 template <typename INT2, typename UINT2>
42 struct ToUnsignedT<INT2, UINT2, true> {
43 static UINT2 ToUnsigned(INT2 value) {
44 return static_cast<UINT2>(value < 0 ? -value : value);
48 // This set of templates is very similar to the above templates, but
49 // for testing whether an integer is negative.
50 template <typename INT2, bool NEG2>
52 template <typename INT2>
53 struct TestNegT<INT2, false> {
54 static bool TestNeg(INT2 value) {
55 // value is unsigned, and can never be negative.
59 template <typename INT2>
60 struct TestNegT<INT2, true> {
61 static bool TestNeg(INT2 value) {
66 static STR IntToString(INT value) {
67 // log10(2) ~= 0.3 bytes needed per bit or per byte log10(2**8) ~= 2.4.
68 // So round up to allocate 3 output characters per byte, plus 1 for '-'.
69 const int kOutputBufSize = 3 * sizeof(INT) + 1;
71 // Allocate the whole string right away, we will right back to front, and
72 // then return the substr of what we ended up using.
73 STR outbuf(kOutputBufSize, 0);
75 bool is_neg = TestNegT<INT, NEG>::TestNeg(value);
76 // Even though is_neg will never be true when INT is parameterized as
77 // unsigned, even the presence of the unary operation causes a warning.
78 UINT res = ToUnsignedT<INT, UINT, NEG>::ToUnsigned(value);
80 for (typename STR::iterator it = outbuf.end();;) {
82 DCHECK(it != outbuf.begin());
83 *it = static_cast<typename STR::value_type>((res % 10) + '0');
90 DCHECK(it != outbuf.begin());
91 *it = static_cast<typename STR::value_type>('-');
93 return STR(it, outbuf.end());
101 // Utility to convert a character to a digit in a given base
102 template<typename CHAR, int BASE, bool BASE_LTE_10> class BaseCharToDigit {
105 // Faster specialization for bases <= 10
106 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, true> {
108 static bool Convert(CHAR c, uint8* digit) {
109 if (c >= '0' && c < '0' + BASE) {
117 // Specialization for bases where 10 < base <= 36
118 template<typename CHAR, int BASE> class BaseCharToDigit<CHAR, BASE, false> {
120 static bool Convert(CHAR c, uint8* digit) {
121 if (c >= '0' && c <= '9') {
123 } else if (c >= 'a' && c < 'a' + BASE - 10) {
124 *digit = c - 'a' + 10;
125 } else if (c >= 'A' && c < 'A' + BASE - 10) {
126 *digit = c - 'A' + 10;
134 template<int BASE, typename CHAR> bool CharToDigit(CHAR c, uint8* digit) {
135 return BaseCharToDigit<CHAR, BASE, BASE <= 10>::Convert(c, digit);
138 // There is an IsWhitespace for wchars defined in string_util.h, but it is
139 // locale independent, whereas the functions we are replacing were
140 // locale-dependent. TBD what is desired, but for the moment let's not introduce
141 // a change in behaviour.
142 template<typename CHAR> class WhitespaceHelper {
145 template<> class WhitespaceHelper<char> {
147 static bool Invoke(char c) {
148 return 0 != isspace(static_cast<unsigned char>(c));
152 template<> class WhitespaceHelper<char16> {
154 static bool Invoke(char16 c) {
155 return 0 != iswspace(c);
159 template<typename CHAR> bool LocalIsWhitespace(CHAR c) {
160 return WhitespaceHelper<CHAR>::Invoke(c);
163 // IteratorRangeToNumberTraits should provide:
164 // - a typedef for iterator_type, the iterator type used as input.
165 // - a typedef for value_type, the target numeric type.
166 // - static functions min, max (returning the minimum and maximum permitted
168 // - constant kBase, the base in which to interpret the input
169 template<typename IteratorRangeToNumberTraits>
170 class IteratorRangeToNumber {
172 typedef IteratorRangeToNumberTraits traits;
173 typedef typename traits::iterator_type const_iterator;
174 typedef typename traits::value_type value_type;
176 // Generalized iterator-range-to-number conversion.
178 static bool Invoke(const_iterator begin,
180 value_type* output) {
183 while (begin != end && LocalIsWhitespace(*begin)) {
188 if (begin != end && *begin == '-') {
189 if (!Negative::Invoke(begin + 1, end, output)) {
193 if (begin != end && *begin == '+') {
196 if (!Positive::Invoke(begin, end, output)) {
206 // - a static function, CheckBounds, that determines whether the next digit
207 // causes an overflow/underflow
208 // - a static function, Increment, that appends the next digit appropriately
209 // according to the sign of the number being parsed.
210 template<typename Sign>
213 static bool Invoke(const_iterator begin, const_iterator end,
214 typename traits::value_type* output) {
221 // Note: no performance difference was found when using template
222 // specialization to remove this check in bases other than 16
223 if (traits::kBase == 16 && end - begin > 2 && *begin == '0' &&
224 (*(begin + 1) == 'x' || *(begin + 1) == 'X')) {
228 for (const_iterator current = begin; current != end; ++current) {
231 if (!CharToDigit<traits::kBase>(*current, &new_digit)) {
235 if (current != begin) {
236 if (!Sign::CheckBounds(output, new_digit)) {
239 *output *= traits::kBase;
242 Sign::Increment(new_digit, output);
248 class Positive : public Base<Positive> {
250 static bool CheckBounds(value_type* output, uint8 new_digit) {
251 if (*output > static_cast<value_type>(traits::max() / traits::kBase) ||
252 (*output == static_cast<value_type>(traits::max() / traits::kBase) &&
253 new_digit > traits::max() % traits::kBase)) {
254 *output = traits::max();
259 static void Increment(uint8 increment, value_type* output) {
260 *output += increment;
264 class Negative : public Base<Negative> {
266 static bool CheckBounds(value_type* output, uint8 new_digit) {
267 if (*output < traits::min() / traits::kBase ||
268 (*output == traits::min() / traits::kBase &&
269 new_digit > 0 - traits::min() % traits::kBase)) {
270 *output = traits::min();
275 static void Increment(uint8 increment, value_type* output) {
276 *output -= increment;
281 template<typename ITERATOR, typename VALUE, int BASE>
282 class BaseIteratorRangeToNumberTraits {
284 typedef ITERATOR iterator_type;
285 typedef VALUE value_type;
286 static value_type min() {
287 return std::numeric_limits<value_type>::min();
289 static value_type max() {
290 return std::numeric_limits<value_type>::max();
292 static const int kBase = BASE;
295 typedef BaseIteratorRangeToNumberTraits<std::string::const_iterator, int, 10>
296 IteratorRangeToIntTraits;
297 typedef BaseIteratorRangeToNumberTraits<string16::const_iterator, int, 10>
298 WideIteratorRangeToIntTraits;
299 typedef BaseIteratorRangeToNumberTraits<std::string::const_iterator, int64, 10>
300 IteratorRangeToInt64Traits;
301 typedef BaseIteratorRangeToNumberTraits<string16::const_iterator, int64, 10>
302 WideIteratorRangeToInt64Traits;
304 typedef BaseIteratorRangeToNumberTraits<const char*, int, 10>
305 CharBufferToIntTraits;
306 typedef BaseIteratorRangeToNumberTraits<const char16*, int, 10>
307 WideCharBufferToIntTraits;
308 typedef BaseIteratorRangeToNumberTraits<const char*, int64, 10>
309 CharBufferToInt64Traits;
310 typedef BaseIteratorRangeToNumberTraits<const char16*, int64, 10>
311 WideCharBufferToInt64Traits;
313 template<typename ITERATOR>
314 class BaseHexIteratorRangeToIntTraits
315 : public BaseIteratorRangeToNumberTraits<ITERATOR, int, 16> {
317 // Allow parsing of 0xFFFFFFFF, which is technically an overflow
318 static unsigned int max() {
319 return std::numeric_limits<unsigned int>::max();
323 typedef BaseHexIteratorRangeToIntTraits<std::string::const_iterator>
324 HexIteratorRangeToIntTraits;
325 typedef BaseHexIteratorRangeToIntTraits<const char*>
326 HexCharBufferToIntTraits;
328 template<typename STR>
329 bool HexStringToBytesT(const STR& input, std::vector<uint8>* output) {
330 DCHECK_EQ(output->size(), 0u);
331 size_t count = input.size();
332 if (count == 0 || (count % 2) != 0)
334 for (uintptr_t i = 0; i < count / 2; ++i) {
335 uint8 msb = 0; // most significant 4 bits
336 uint8 lsb = 0; // least significant 4 bits
337 if (!CharToDigit<16>(input[i * 2], &msb) ||
338 !CharToDigit<16>(input[i * 2 + 1], &lsb))
340 output->push_back((msb << 4) | lsb);
347 std::string IntToString(int value) {
348 return IntToStringT<std::string, int, unsigned int, true>::
352 string16 IntToString16(int value) {
353 return IntToStringT<string16, int, unsigned int, true>::
357 std::string UintToString(unsigned int value) {
358 return IntToStringT<std::string, unsigned int, unsigned int, false>::
362 string16 UintToString16(unsigned int value) {
363 return IntToStringT<string16, unsigned int, unsigned int, false>::
367 std::string Int64ToString(int64 value) {
368 return IntToStringT<std::string, int64, uint64, true>::
372 string16 Int64ToString16(int64 value) {
373 return IntToStringT<string16, int64, uint64, true>::IntToString(value);
376 std::string Uint64ToString(uint64 value) {
377 return IntToStringT<std::string, uint64, uint64, false>::
381 string16 Uint64ToString16(uint64 value) {
382 return IntToStringT<string16, uint64, uint64, false>::
386 std::string DoubleToString(double value) {
387 // According to g_fmt.cc, it is sufficient to declare a buffer of size 32.
389 dmg_fp::g_fmt(buffer, value);
390 return std::string(buffer);
393 bool StringToInt(const std::string& input, int* output) {
394 return IteratorRangeToNumber<IteratorRangeToIntTraits>::Invoke(input.begin(),
399 bool StringToInt(std::string::const_iterator begin,
400 std::string::const_iterator end,
402 return IteratorRangeToNumber<IteratorRangeToIntTraits>::Invoke(begin,
407 #if !defined(STD_STRING_ITERATOR_IS_CHAR_POINTER)
408 bool StringToInt(const char* begin, const char* end, int* output) {
409 return IteratorRangeToNumber<CharBufferToIntTraits>::Invoke(begin,
415 bool StringToInt(const string16& input, int* output) {
416 return IteratorRangeToNumber<WideIteratorRangeToIntTraits>::Invoke(
417 input.begin(), input.end(), output);
420 bool StringToInt(string16::const_iterator begin,
421 string16::const_iterator end,
423 return IteratorRangeToNumber<WideIteratorRangeToIntTraits>::Invoke(begin,
428 #if !defined(BASE_STRING16_ITERATOR_IS_CHAR16_POINTER)
429 bool StringToInt(const char16* begin, const char16* end, int* output) {
430 return IteratorRangeToNumber<WideCharBufferToIntTraits>::Invoke(begin,
436 bool StringToInt64(const std::string& input, int64* output) {
437 return IteratorRangeToNumber<IteratorRangeToInt64Traits>::Invoke(
438 input.begin(), input.end(), output);
441 bool StringToInt64(std::string::const_iterator begin,
442 std::string::const_iterator end,
444 return IteratorRangeToNumber<IteratorRangeToInt64Traits>::Invoke(begin,
449 #if !defined(STD_STRING_ITERATOR_IS_CHAR_POINTER)
450 bool StringToInt64(const char* begin, const char* end, int64* output) {
451 return IteratorRangeToNumber<CharBufferToInt64Traits>::Invoke(begin,
457 bool StringToInt64(const string16& input, int64* output) {
458 return IteratorRangeToNumber<WideIteratorRangeToInt64Traits>::Invoke(
459 input.begin(), input.end(), output);
462 bool StringToInt64(string16::const_iterator begin,
463 string16::const_iterator end,
465 return IteratorRangeToNumber<WideIteratorRangeToInt64Traits>::Invoke(begin,
470 #if !defined(BASE_STRING16_ITERATOR_IS_CHAR16_POINTER)
471 bool StringToInt64(const char16* begin, const char16* end, int64* output) {
472 return IteratorRangeToNumber<WideCharBufferToInt64Traits>::Invoke(begin,
478 bool StringToDouble(const std::string& input, double* output) {
479 errno = 0; // Thread-safe? It is on at least Mac, Linux, and Windows.
481 *output = dmg_fp::strtod(input.c_str(), &endptr);
483 // Cases to return false:
484 // - If errno is ERANGE, there was an overflow or underflow.
485 // - If the input string is empty, there was nothing to parse.
486 // - If endptr does not point to the end of the string, there are either
487 // characters remaining in the string after a parsed number, or the string
488 // does not begin with a parseable number. endptr is compared to the
489 // expected end given the string's stated length to correctly catch cases
490 // where the string contains embedded NUL characters.
491 // - If the first character is a space, there was leading whitespace
494 input.c_str() + input.length() == endptr &&
498 // Note: if you need to add String16ToDouble, first ask yourself if it's
499 // really necessary. If it is, probably the best implementation here is to
500 // convert to 8-bit and then use the 8-bit version.
502 // Note: if you need to add an iterator range version of StringToDouble, first
503 // ask yourself if it's really necessary. If it is, probably the best
504 // implementation here is to instantiate a string and use the string version.
506 std::string HexEncode(const void* bytes, size_t size) {
507 static const char kHexChars[] = "0123456789ABCDEF";
509 // Each input byte creates two output hex characters.
510 std::string ret(size * 2, '\0');
512 for (size_t i = 0; i < size; ++i) {
513 char b = reinterpret_cast<const char*>(bytes)[i];
514 ret[(i * 2)] = kHexChars[(b >> 4) & 0xf];
515 ret[(i * 2) + 1] = kHexChars[b & 0xf];
520 bool HexStringToInt(const std::string& input, int* output) {
521 return IteratorRangeToNumber<HexIteratorRangeToIntTraits>::Invoke(
522 input.begin(), input.end(), output);
525 bool HexStringToInt(std::string::const_iterator begin,
526 std::string::const_iterator end,
528 return IteratorRangeToNumber<HexIteratorRangeToIntTraits>::Invoke(begin,
533 #if !defined(STD_STRING_ITERATOR_IS_CHAR_POINTER)
534 bool HexStringToInt(const char* begin, const char* end, int* output) {
535 return IteratorRangeToNumber<HexCharBufferToIntTraits>::Invoke(begin,
541 bool HexStringToBytes(const std::string& input, std::vector<uint8>* output) {
542 return HexStringToBytesT(input, output);