Wed, 31 Dec 2014 06:09:35 +0100
Cloned upstream origin tor-browser at tor-browser-31.3.0esr-4.5-1-build1
revision ID fc1c9ff7c1b2defdbc039f12214767608f46423f for hacking purpose.
michael@0 | 1 | // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved. |
michael@0 | 2 | // Use of this source code is governed by a BSD-style license that can be |
michael@0 | 3 | // found in the LICENSE file. |
michael@0 | 4 | |
michael@0 | 5 | #include "base/string_util.h" |
michael@0 | 6 | |
michael@0 | 7 | #include "build/build_config.h" |
michael@0 | 8 | |
michael@0 | 9 | #include <ctype.h> |
michael@0 | 10 | #include <errno.h> |
michael@0 | 11 | #include <math.h> |
michael@0 | 12 | #include <stdarg.h> |
michael@0 | 13 | #include <stdio.h> |
michael@0 | 14 | #include <stdlib.h> |
michael@0 | 15 | #include <string.h> |
michael@0 | 16 | #include <time.h> |
michael@0 | 17 | #include <wchar.h> |
michael@0 | 18 | #include <wctype.h> |
michael@0 | 19 | |
michael@0 | 20 | #include <algorithm> |
michael@0 | 21 | #include <vector> |
michael@0 | 22 | |
michael@0 | 23 | #include "base/basictypes.h" |
michael@0 | 24 | #include "base/logging.h" |
michael@0 | 25 | #include "base/singleton.h" |
michael@0 | 26 | |
michael@0 | 27 | namespace { |
michael@0 | 28 | |
michael@0 | 29 | // Force the singleton used by Empty[W]String[16] to be a unique type. This |
michael@0 | 30 | // prevents other code that might accidentally use Singleton<string> from |
michael@0 | 31 | // getting our internal one. |
michael@0 | 32 | struct EmptyStrings { |
michael@0 | 33 | EmptyStrings() {} |
michael@0 | 34 | const std::string s; |
michael@0 | 35 | const std::wstring ws; |
michael@0 | 36 | const string16 s16; |
michael@0 | 37 | }; |
michael@0 | 38 | |
michael@0 | 39 | // Hack to convert any char-like type to its unsigned counterpart. |
michael@0 | 40 | // For example, it will convert char, signed char and unsigned char to unsigned |
michael@0 | 41 | // char. |
michael@0 | 42 | template<typename T> |
michael@0 | 43 | struct ToUnsigned { |
michael@0 | 44 | typedef T Unsigned; |
michael@0 | 45 | }; |
michael@0 | 46 | |
michael@0 | 47 | template<> |
michael@0 | 48 | struct ToUnsigned<char> { |
michael@0 | 49 | typedef unsigned char Unsigned; |
michael@0 | 50 | }; |
michael@0 | 51 | template<> |
michael@0 | 52 | struct ToUnsigned<signed char> { |
michael@0 | 53 | typedef unsigned char Unsigned; |
michael@0 | 54 | }; |
michael@0 | 55 | template<> |
michael@0 | 56 | struct ToUnsigned<wchar_t> { |
michael@0 | 57 | #if defined(WCHAR_T_IS_UTF16) |
michael@0 | 58 | typedef unsigned short Unsigned; |
michael@0 | 59 | #elif defined(WCHAR_T_IS_UTF32) |
michael@0 | 60 | typedef uint32_t Unsigned; |
michael@0 | 61 | #endif |
michael@0 | 62 | }; |
michael@0 | 63 | template<> |
michael@0 | 64 | struct ToUnsigned<short> { |
michael@0 | 65 | typedef unsigned short Unsigned; |
michael@0 | 66 | }; |
michael@0 | 67 | |
michael@0 | 68 | // Generalized string-to-number conversion. |
michael@0 | 69 | // |
michael@0 | 70 | // StringToNumberTraits should provide: |
michael@0 | 71 | // - a typedef for string_type, the STL string type used as input. |
michael@0 | 72 | // - a typedef for value_type, the target numeric type. |
michael@0 | 73 | // - a static function, convert_func, which dispatches to an appropriate |
michael@0 | 74 | // strtol-like function and returns type value_type. |
michael@0 | 75 | // - a static function, valid_func, which validates |input| and returns a bool |
michael@0 | 76 | // indicating whether it is in proper form. This is used to check for |
michael@0 | 77 | // conditions that convert_func tolerates but should result in |
michael@0 | 78 | // StringToNumber returning false. For strtol-like funtions, valid_func |
michael@0 | 79 | // should check for leading whitespace. |
michael@0 | 80 | template<typename StringToNumberTraits> |
michael@0 | 81 | bool StringToNumber(const typename StringToNumberTraits::string_type& input, |
michael@0 | 82 | typename StringToNumberTraits::value_type* output) { |
michael@0 | 83 | typedef StringToNumberTraits traits; |
michael@0 | 84 | |
michael@0 | 85 | errno = 0; // Thread-safe? It is on at least Mac, Linux, and Windows. |
michael@0 | 86 | typename traits::string_type::value_type* endptr = NULL; |
michael@0 | 87 | typename traits::value_type value = traits::convert_func(input.c_str(), |
michael@0 | 88 | &endptr); |
michael@0 | 89 | *output = value; |
michael@0 | 90 | |
michael@0 | 91 | // Cases to return false: |
michael@0 | 92 | // - If errno is ERANGE, there was an overflow or underflow. |
michael@0 | 93 | // - If the input string is empty, there was nothing to parse. |
michael@0 | 94 | // - If endptr does not point to the end of the string, there are either |
michael@0 | 95 | // characters remaining in the string after a parsed number, or the string |
michael@0 | 96 | // does not begin with a parseable number. endptr is compared to the |
michael@0 | 97 | // expected end given the string's stated length to correctly catch cases |
michael@0 | 98 | // where the string contains embedded NUL characters. |
michael@0 | 99 | // - valid_func determines that the input is not in preferred form. |
michael@0 | 100 | return errno == 0 && |
michael@0 | 101 | !input.empty() && |
michael@0 | 102 | input.c_str() + input.length() == endptr && |
michael@0 | 103 | traits::valid_func(input); |
michael@0 | 104 | } |
michael@0 | 105 | |
michael@0 | 106 | class StringToLongTraits { |
michael@0 | 107 | public: |
michael@0 | 108 | typedef std::string string_type; |
michael@0 | 109 | typedef long value_type; |
michael@0 | 110 | static const int kBase = 10; |
michael@0 | 111 | static inline value_type convert_func(const string_type::value_type* str, |
michael@0 | 112 | string_type::value_type** endptr) { |
michael@0 | 113 | return strtol(str, endptr, kBase); |
michael@0 | 114 | } |
michael@0 | 115 | static inline bool valid_func(const string_type& str) { |
michael@0 | 116 | return !str.empty() && !isspace(str[0]); |
michael@0 | 117 | } |
michael@0 | 118 | }; |
michael@0 | 119 | |
michael@0 | 120 | class String16ToLongTraits { |
michael@0 | 121 | public: |
michael@0 | 122 | typedef string16 string_type; |
michael@0 | 123 | typedef long value_type; |
michael@0 | 124 | static const int kBase = 10; |
michael@0 | 125 | static inline value_type convert_func(const string_type::value_type* str, |
michael@0 | 126 | string_type::value_type** endptr) { |
michael@0 | 127 | #if defined(WCHAR_T_IS_UTF16) |
michael@0 | 128 | return wcstol(str, endptr, kBase); |
michael@0 | 129 | #elif defined(WCHAR_T_IS_UTF32) |
michael@0 | 130 | std::string ascii_string = UTF16ToASCII(string16(str)); |
michael@0 | 131 | char* ascii_end = NULL; |
michael@0 | 132 | value_type ret = strtol(ascii_string.c_str(), &ascii_end, kBase); |
michael@0 | 133 | if (ascii_string.c_str() + ascii_string.length() == ascii_end) { |
michael@0 | 134 | *endptr = |
michael@0 | 135 | const_cast<string_type::value_type*>(str) + ascii_string.length(); |
michael@0 | 136 | } |
michael@0 | 137 | return ret; |
michael@0 | 138 | #endif |
michael@0 | 139 | } |
michael@0 | 140 | static inline bool valid_func(const string_type& str) { |
michael@0 | 141 | return !str.empty() && !iswspace(str[0]); |
michael@0 | 142 | } |
michael@0 | 143 | }; |
michael@0 | 144 | |
michael@0 | 145 | class StringToInt64Traits { |
michael@0 | 146 | public: |
michael@0 | 147 | typedef std::string string_type; |
michael@0 | 148 | typedef int64_t value_type; |
michael@0 | 149 | static const int kBase = 10; |
michael@0 | 150 | static inline value_type convert_func(const string_type::value_type* str, |
michael@0 | 151 | string_type::value_type** endptr) { |
michael@0 | 152 | #ifdef OS_WIN |
michael@0 | 153 | return _strtoi64(str, endptr, kBase); |
michael@0 | 154 | #else // assume OS_POSIX |
michael@0 | 155 | return strtoll(str, endptr, kBase); |
michael@0 | 156 | #endif |
michael@0 | 157 | } |
michael@0 | 158 | static inline bool valid_func(const string_type& str) { |
michael@0 | 159 | return !str.empty() && !isspace(str[0]); |
michael@0 | 160 | } |
michael@0 | 161 | }; |
michael@0 | 162 | |
michael@0 | 163 | class String16ToInt64Traits { |
michael@0 | 164 | public: |
michael@0 | 165 | typedef string16 string_type; |
michael@0 | 166 | typedef int64_t value_type; |
michael@0 | 167 | static const int kBase = 10; |
michael@0 | 168 | static inline value_type convert_func(const string_type::value_type* str, |
michael@0 | 169 | string_type::value_type** endptr) { |
michael@0 | 170 | #ifdef OS_WIN |
michael@0 | 171 | return _wcstoi64(str, endptr, kBase); |
michael@0 | 172 | #else // assume OS_POSIX |
michael@0 | 173 | std::string ascii_string = UTF16ToASCII(string16(str)); |
michael@0 | 174 | char* ascii_end = NULL; |
michael@0 | 175 | value_type ret = strtoll(ascii_string.c_str(), &ascii_end, kBase); |
michael@0 | 176 | if (ascii_string.c_str() + ascii_string.length() == ascii_end) { |
michael@0 | 177 | *endptr = |
michael@0 | 178 | const_cast<string_type::value_type*>(str) + ascii_string.length(); |
michael@0 | 179 | } |
michael@0 | 180 | return ret; |
michael@0 | 181 | #endif |
michael@0 | 182 | } |
michael@0 | 183 | static inline bool valid_func(const string_type& str) { |
michael@0 | 184 | return !str.empty() && !iswspace(str[0]); |
michael@0 | 185 | } |
michael@0 | 186 | }; |
michael@0 | 187 | |
michael@0 | 188 | } // namespace |
michael@0 | 189 | |
michael@0 | 190 | |
michael@0 | 191 | namespace base { |
michael@0 | 192 | |
michael@0 | 193 | bool IsWprintfFormatPortable(const wchar_t* format) { |
michael@0 | 194 | for (const wchar_t* position = format; *position != '\0'; ++position) { |
michael@0 | 195 | |
michael@0 | 196 | if (*position == '%') { |
michael@0 | 197 | bool in_specification = true; |
michael@0 | 198 | bool modifier_l = false; |
michael@0 | 199 | while (in_specification) { |
michael@0 | 200 | // Eat up characters until reaching a known specifier. |
michael@0 | 201 | if (*++position == '\0') { |
michael@0 | 202 | // The format string ended in the middle of a specification. Call |
michael@0 | 203 | // it portable because no unportable specifications were found. The |
michael@0 | 204 | // string is equally broken on all platforms. |
michael@0 | 205 | return true; |
michael@0 | 206 | } |
michael@0 | 207 | |
michael@0 | 208 | if (*position == 'l') { |
michael@0 | 209 | // 'l' is the only thing that can save the 's' and 'c' specifiers. |
michael@0 | 210 | modifier_l = true; |
michael@0 | 211 | } else if (((*position == 's' || *position == 'c') && !modifier_l) || |
michael@0 | 212 | *position == 'S' || *position == 'C' || *position == 'F' || |
michael@0 | 213 | *position == 'D' || *position == 'O' || *position == 'U') { |
michael@0 | 214 | // Not portable. |
michael@0 | 215 | return false; |
michael@0 | 216 | } |
michael@0 | 217 | |
michael@0 | 218 | if (wcschr(L"diouxXeEfgGaAcspn%", *position)) { |
michael@0 | 219 | // Portable, keep scanning the rest of the format string. |
michael@0 | 220 | in_specification = false; |
michael@0 | 221 | } |
michael@0 | 222 | } |
michael@0 | 223 | } |
michael@0 | 224 | |
michael@0 | 225 | } |
michael@0 | 226 | |
michael@0 | 227 | return true; |
michael@0 | 228 | } |
michael@0 | 229 | |
michael@0 | 230 | |
michael@0 | 231 | } // namespace base |
michael@0 | 232 | |
michael@0 | 233 | static const wchar_t kWhitespaceWide[] = { |
michael@0 | 234 | 0x0009, // <control-0009> to <control-000D> |
michael@0 | 235 | 0x000A, |
michael@0 | 236 | 0x000B, |
michael@0 | 237 | 0x000C, |
michael@0 | 238 | 0x000D, |
michael@0 | 239 | 0x0020, // Space |
michael@0 | 240 | 0x0085, // <control-0085> |
michael@0 | 241 | 0x00A0, // No-Break Space |
michael@0 | 242 | 0x1680, // Ogham Space Mark |
michael@0 | 243 | 0x180E, // Mongolian Vowel Separator |
michael@0 | 244 | 0x2000, // En Quad to Hair Space |
michael@0 | 245 | 0x2001, |
michael@0 | 246 | 0x2002, |
michael@0 | 247 | 0x2003, |
michael@0 | 248 | 0x2004, |
michael@0 | 249 | 0x2005, |
michael@0 | 250 | 0x2006, |
michael@0 | 251 | 0x2007, |
michael@0 | 252 | 0x2008, |
michael@0 | 253 | 0x2009, |
michael@0 | 254 | 0x200A, |
michael@0 | 255 | 0x200C, // Zero Width Non-Joiner |
michael@0 | 256 | 0x2028, // Line Separator |
michael@0 | 257 | 0x2029, // Paragraph Separator |
michael@0 | 258 | 0x202F, // Narrow No-Break Space |
michael@0 | 259 | 0x205F, // Medium Mathematical Space |
michael@0 | 260 | 0x3000, // Ideographic Space |
michael@0 | 261 | 0 |
michael@0 | 262 | }; |
michael@0 | 263 | static const char kWhitespaceASCII[] = { |
michael@0 | 264 | 0x09, // <control-0009> to <control-000D> |
michael@0 | 265 | 0x0A, |
michael@0 | 266 | 0x0B, |
michael@0 | 267 | 0x0C, |
michael@0 | 268 | 0x0D, |
michael@0 | 269 | 0x20, // Space |
michael@0 | 270 | 0 |
michael@0 | 271 | }; |
michael@0 | 272 | |
michael@0 | 273 | template<typename STR> |
michael@0 | 274 | TrimPositions TrimStringT(const STR& input, |
michael@0 | 275 | const typename STR::value_type trim_chars[], |
michael@0 | 276 | TrimPositions positions, |
michael@0 | 277 | STR* output) { |
michael@0 | 278 | // Find the edges of leading/trailing whitespace as desired. |
michael@0 | 279 | const typename STR::size_type last_char = input.length() - 1; |
michael@0 | 280 | const typename STR::size_type first_good_char = (positions & TRIM_LEADING) ? |
michael@0 | 281 | input.find_first_not_of(trim_chars) : 0; |
michael@0 | 282 | const typename STR::size_type last_good_char = (positions & TRIM_TRAILING) ? |
michael@0 | 283 | input.find_last_not_of(trim_chars) : last_char; |
michael@0 | 284 | |
michael@0 | 285 | // When the string was all whitespace, report that we stripped off whitespace |
michael@0 | 286 | // from whichever position the caller was interested in. For empty input, we |
michael@0 | 287 | // stripped no whitespace, but we still need to clear |output|. |
michael@0 | 288 | if (input.empty() || |
michael@0 | 289 | (first_good_char == STR::npos) || (last_good_char == STR::npos)) { |
michael@0 | 290 | bool input_was_empty = input.empty(); // in case output == &input |
michael@0 | 291 | output->clear(); |
michael@0 | 292 | return input_was_empty ? TRIM_NONE : positions; |
michael@0 | 293 | } |
michael@0 | 294 | |
michael@0 | 295 | // Trim the whitespace. |
michael@0 | 296 | *output = |
michael@0 | 297 | input.substr(first_good_char, last_good_char - first_good_char + 1); |
michael@0 | 298 | |
michael@0 | 299 | // Return where we trimmed from. |
michael@0 | 300 | return static_cast<TrimPositions>( |
michael@0 | 301 | ((first_good_char == 0) ? TRIM_NONE : TRIM_LEADING) | |
michael@0 | 302 | ((last_good_char == last_char) ? TRIM_NONE : TRIM_TRAILING)); |
michael@0 | 303 | } |
michael@0 | 304 | |
michael@0 | 305 | TrimPositions TrimWhitespace(const std::wstring& input, |
michael@0 | 306 | TrimPositions positions, |
michael@0 | 307 | std::wstring* output) { |
michael@0 | 308 | return TrimStringT(input, kWhitespaceWide, positions, output); |
michael@0 | 309 | } |
michael@0 | 310 | |
michael@0 | 311 | TrimPositions TrimWhitespaceASCII(const std::string& input, |
michael@0 | 312 | TrimPositions positions, |
michael@0 | 313 | std::string* output) { |
michael@0 | 314 | return TrimStringT(input, kWhitespaceASCII, positions, output); |
michael@0 | 315 | } |
michael@0 | 316 | |
michael@0 | 317 | // This function is only for backward-compatibility. |
michael@0 | 318 | // To be removed when all callers are updated. |
michael@0 | 319 | TrimPositions TrimWhitespace(const std::string& input, |
michael@0 | 320 | TrimPositions positions, |
michael@0 | 321 | std::string* output) { |
michael@0 | 322 | return TrimWhitespaceASCII(input, positions, output); |
michael@0 | 323 | } |
michael@0 | 324 | |
michael@0 | 325 | std::string WideToASCII(const std::wstring& wide) { |
michael@0 | 326 | DCHECK(IsStringASCII(wide)); |
michael@0 | 327 | return std::string(wide.begin(), wide.end()); |
michael@0 | 328 | } |
michael@0 | 329 | |
michael@0 | 330 | std::wstring ASCIIToWide(const std::string& ascii) { |
michael@0 | 331 | DCHECK(IsStringASCII(ascii)); |
michael@0 | 332 | return std::wstring(ascii.begin(), ascii.end()); |
michael@0 | 333 | } |
michael@0 | 334 | |
michael@0 | 335 | std::string UTF16ToASCII(const string16& utf16) { |
michael@0 | 336 | DCHECK(IsStringASCII(utf16)); |
michael@0 | 337 | return std::string(utf16.begin(), utf16.end()); |
michael@0 | 338 | } |
michael@0 | 339 | |
michael@0 | 340 | string16 ASCIIToUTF16(const std::string& ascii) { |
michael@0 | 341 | DCHECK(IsStringASCII(ascii)); |
michael@0 | 342 | return string16(ascii.begin(), ascii.end()); |
michael@0 | 343 | } |
michael@0 | 344 | |
michael@0 | 345 | template<class STR> |
michael@0 | 346 | static bool DoIsStringASCII(const STR& str) { |
michael@0 | 347 | for (size_t i = 0; i < str.length(); i++) { |
michael@0 | 348 | typename ToUnsigned<typename STR::value_type>::Unsigned c = str[i]; |
michael@0 | 349 | if (c > 0x7F) |
michael@0 | 350 | return false; |
michael@0 | 351 | } |
michael@0 | 352 | return true; |
michael@0 | 353 | } |
michael@0 | 354 | |
michael@0 | 355 | bool IsStringASCII(const std::wstring& str) { |
michael@0 | 356 | return DoIsStringASCII(str); |
michael@0 | 357 | } |
michael@0 | 358 | |
michael@0 | 359 | #if !defined(WCHAR_T_IS_UTF16) |
michael@0 | 360 | bool IsStringASCII(const string16& str) { |
michael@0 | 361 | return DoIsStringASCII(str); |
michael@0 | 362 | } |
michael@0 | 363 | #endif |
michael@0 | 364 | |
michael@0 | 365 | bool IsStringASCII(const std::string& str) { |
michael@0 | 366 | return DoIsStringASCII(str); |
michael@0 | 367 | } |
michael@0 | 368 | |
michael@0 | 369 | // Overloaded wrappers around vsnprintf and vswprintf. The buf_size parameter |
michael@0 | 370 | // is the size of the buffer. These return the number of characters in the |
michael@0 | 371 | // formatted string excluding the NUL terminator. If the buffer is not |
michael@0 | 372 | // large enough to accommodate the formatted string without truncation, they |
michael@0 | 373 | // return the number of characters that would be in the fully-formatted string |
michael@0 | 374 | // (vsnprintf, and vswprintf on Windows), or -1 (vswprintf on POSIX platforms). |
michael@0 | 375 | inline int vsnprintfT(char* buffer, |
michael@0 | 376 | size_t buf_size, |
michael@0 | 377 | const char* format, |
michael@0 | 378 | va_list argptr) { |
michael@0 | 379 | return base::vsnprintf(buffer, buf_size, format, argptr); |
michael@0 | 380 | } |
michael@0 | 381 | |
michael@0 | 382 | inline int vsnprintfT(wchar_t* buffer, |
michael@0 | 383 | size_t buf_size, |
michael@0 | 384 | const wchar_t* format, |
michael@0 | 385 | va_list argptr) { |
michael@0 | 386 | return base::vswprintf(buffer, buf_size, format, argptr); |
michael@0 | 387 | } |
michael@0 | 388 | |
michael@0 | 389 | // Templatized backend for StringPrintF/StringAppendF. This does not finalize |
michael@0 | 390 | // the va_list, the caller is expected to do that. |
michael@0 | 391 | template <class StringType> |
michael@0 | 392 | static void StringAppendVT(StringType* dst, |
michael@0 | 393 | const typename StringType::value_type* format, |
michael@0 | 394 | va_list ap) { |
michael@0 | 395 | // First try with a small fixed size buffer. |
michael@0 | 396 | // This buffer size should be kept in sync with StringUtilTest.GrowBoundary |
michael@0 | 397 | // and StringUtilTest.StringPrintfBounds. |
michael@0 | 398 | typename StringType::value_type stack_buf[1024]; |
michael@0 | 399 | |
michael@0 | 400 | va_list backup_ap; |
michael@0 | 401 | base_va_copy(backup_ap, ap); |
michael@0 | 402 | |
michael@0 | 403 | #if !defined(OS_WIN) |
michael@0 | 404 | errno = 0; |
michael@0 | 405 | #endif |
michael@0 | 406 | int result = vsnprintfT(stack_buf, arraysize(stack_buf), format, backup_ap); |
michael@0 | 407 | va_end(backup_ap); |
michael@0 | 408 | |
michael@0 | 409 | if (result >= 0 && result < static_cast<int>(arraysize(stack_buf))) { |
michael@0 | 410 | // It fit. |
michael@0 | 411 | dst->append(stack_buf, result); |
michael@0 | 412 | return; |
michael@0 | 413 | } |
michael@0 | 414 | |
michael@0 | 415 | // Repeatedly increase buffer size until it fits. |
michael@0 | 416 | int mem_length = arraysize(stack_buf); |
michael@0 | 417 | while (true) { |
michael@0 | 418 | if (result < 0) { |
michael@0 | 419 | #if !defined(OS_WIN) |
michael@0 | 420 | // On Windows, vsnprintfT always returns the number of characters in a |
michael@0 | 421 | // fully-formatted string, so if we reach this point, something else is |
michael@0 | 422 | // wrong and no amount of buffer-doubling is going to fix it. |
michael@0 | 423 | if (errno != 0 && errno != EOVERFLOW) |
michael@0 | 424 | #endif |
michael@0 | 425 | { |
michael@0 | 426 | // If an error other than overflow occurred, it's never going to work. |
michael@0 | 427 | DLOG(WARNING) << "Unable to printf the requested string due to error."; |
michael@0 | 428 | return; |
michael@0 | 429 | } |
michael@0 | 430 | // Try doubling the buffer size. |
michael@0 | 431 | mem_length *= 2; |
michael@0 | 432 | } else { |
michael@0 | 433 | // We need exactly "result + 1" characters. |
michael@0 | 434 | mem_length = result + 1; |
michael@0 | 435 | } |
michael@0 | 436 | |
michael@0 | 437 | if (mem_length > 32 * 1024 * 1024) { |
michael@0 | 438 | // That should be plenty, don't try anything larger. This protects |
michael@0 | 439 | // against huge allocations when using vsnprintfT implementations that |
michael@0 | 440 | // return -1 for reasons other than overflow without setting errno. |
michael@0 | 441 | DLOG(WARNING) << "Unable to printf the requested string due to size."; |
michael@0 | 442 | return; |
michael@0 | 443 | } |
michael@0 | 444 | |
michael@0 | 445 | std::vector<typename StringType::value_type> mem_buf(mem_length); |
michael@0 | 446 | |
michael@0 | 447 | // Restore the va_list before we use it again. |
michael@0 | 448 | base_va_copy(backup_ap, ap); |
michael@0 | 449 | |
michael@0 | 450 | result = vsnprintfT(&mem_buf[0], mem_length, format, ap); |
michael@0 | 451 | va_end(backup_ap); |
michael@0 | 452 | |
michael@0 | 453 | if ((result >= 0) && (result < mem_length)) { |
michael@0 | 454 | // It fit. |
michael@0 | 455 | dst->append(&mem_buf[0], result); |
michael@0 | 456 | return; |
michael@0 | 457 | } |
michael@0 | 458 | } |
michael@0 | 459 | } |
michael@0 | 460 | |
michael@0 | 461 | namespace { |
michael@0 | 462 | |
michael@0 | 463 | template <typename STR, typename INT, typename UINT, bool NEG> |
michael@0 | 464 | struct IntToStringT { |
michael@0 | 465 | |
michael@0 | 466 | // This is to avoid a compiler warning about unary minus on unsigned type. |
michael@0 | 467 | // For example, say you had the following code: |
michael@0 | 468 | // template <typename INT> |
michael@0 | 469 | // INT abs(INT value) { return value < 0 ? -value : value; } |
michael@0 | 470 | // Even though if INT is unsigned, it's impossible for value < 0, so the |
michael@0 | 471 | // unary minus will never be taken, the compiler will still generate a |
michael@0 | 472 | // warning. We do a little specialization dance... |
michael@0 | 473 | template <typename INT2, typename UINT2, bool NEG2> |
michael@0 | 474 | struct ToUnsignedT { }; |
michael@0 | 475 | |
michael@0 | 476 | template <typename INT2, typename UINT2> |
michael@0 | 477 | struct ToUnsignedT<INT2, UINT2, false> { |
michael@0 | 478 | static UINT2 ToUnsigned(INT2 value) { |
michael@0 | 479 | return static_cast<UINT2>(value); |
michael@0 | 480 | } |
michael@0 | 481 | }; |
michael@0 | 482 | |
michael@0 | 483 | template <typename INT2, typename UINT2> |
michael@0 | 484 | struct ToUnsignedT<INT2, UINT2, true> { |
michael@0 | 485 | static UINT2 ToUnsigned(INT2 value) { |
michael@0 | 486 | return static_cast<UINT2>(value < 0 ? -value : value); |
michael@0 | 487 | } |
michael@0 | 488 | }; |
michael@0 | 489 | |
michael@0 | 490 | // This set of templates is very similar to the above templates, but |
michael@0 | 491 | // for testing whether an integer is negative. |
michael@0 | 492 | template <typename INT2, bool NEG2> |
michael@0 | 493 | struct TestNegT {}; |
michael@0 | 494 | template <typename INT2> |
michael@0 | 495 | struct TestNegT<INT2, false> { |
michael@0 | 496 | static bool TestNeg(INT2 value) { |
michael@0 | 497 | // value is unsigned, and can never be negative. |
michael@0 | 498 | return false; |
michael@0 | 499 | } |
michael@0 | 500 | }; |
michael@0 | 501 | template <typename INT2> |
michael@0 | 502 | struct TestNegT<INT2, true> { |
michael@0 | 503 | static bool TestNeg(INT2 value) { |
michael@0 | 504 | return value < 0; |
michael@0 | 505 | } |
michael@0 | 506 | }; |
michael@0 | 507 | |
michael@0 | 508 | static STR IntToString(INT value) { |
michael@0 | 509 | // log10(2) ~= 0.3 bytes needed per bit or per byte log10(2**8) ~= 2.4. |
michael@0 | 510 | // So round up to allocate 3 output characters per byte, plus 1 for '-'. |
michael@0 | 511 | const int kOutputBufSize = 3 * sizeof(INT) + 1; |
michael@0 | 512 | |
michael@0 | 513 | // Allocate the whole string right away, we will right back to front, and |
michael@0 | 514 | // then return the substr of what we ended up using. |
michael@0 | 515 | STR outbuf(kOutputBufSize, 0); |
michael@0 | 516 | |
michael@0 | 517 | bool is_neg = TestNegT<INT, NEG>::TestNeg(value); |
michael@0 | 518 | // Even though is_neg will never be true when INT is parameterized as |
michael@0 | 519 | // unsigned, even the presence of the unary operation causes a warning. |
michael@0 | 520 | UINT res = ToUnsignedT<INT, UINT, NEG>::ToUnsigned(value); |
michael@0 | 521 | |
michael@0 | 522 | for (typename STR::iterator it = outbuf.end();;) { |
michael@0 | 523 | --it; |
michael@0 | 524 | DCHECK(it != outbuf.begin()); |
michael@0 | 525 | *it = static_cast<typename STR::value_type>((res % 10) + '0'); |
michael@0 | 526 | res /= 10; |
michael@0 | 527 | |
michael@0 | 528 | // We're done.. |
michael@0 | 529 | if (res == 0) { |
michael@0 | 530 | if (is_neg) { |
michael@0 | 531 | --it; |
michael@0 | 532 | DCHECK(it != outbuf.begin()); |
michael@0 | 533 | *it = static_cast<typename STR::value_type>('-'); |
michael@0 | 534 | } |
michael@0 | 535 | return STR(it, outbuf.end()); |
michael@0 | 536 | } |
michael@0 | 537 | } |
michael@0 | 538 | NOTREACHED(); |
michael@0 | 539 | return STR(); |
michael@0 | 540 | } |
michael@0 | 541 | }; |
michael@0 | 542 | |
michael@0 | 543 | } |
michael@0 | 544 | |
michael@0 | 545 | std::string IntToString(int value) { |
michael@0 | 546 | return IntToStringT<std::string, int, unsigned int, true>:: |
michael@0 | 547 | IntToString(value); |
michael@0 | 548 | } |
michael@0 | 549 | std::wstring IntToWString(int value) { |
michael@0 | 550 | return IntToStringT<std::wstring, int, unsigned int, true>:: |
michael@0 | 551 | IntToString(value); |
michael@0 | 552 | } |
michael@0 | 553 | std::string UintToString(unsigned int value) { |
michael@0 | 554 | return IntToStringT<std::string, unsigned int, unsigned int, false>:: |
michael@0 | 555 | IntToString(value); |
michael@0 | 556 | } |
michael@0 | 557 | std::wstring UintToWString(unsigned int value) { |
michael@0 | 558 | return IntToStringT<std::wstring, unsigned int, unsigned int, false>:: |
michael@0 | 559 | IntToString(value); |
michael@0 | 560 | } |
michael@0 | 561 | std::string Int64ToString(int64_t value) { |
michael@0 | 562 | return IntToStringT<std::string, int64_t, uint64_t, true>:: |
michael@0 | 563 | IntToString(value); |
michael@0 | 564 | } |
michael@0 | 565 | std::wstring Int64ToWString(int64_t value) { |
michael@0 | 566 | return IntToStringT<std::wstring, int64_t, uint64_t, true>:: |
michael@0 | 567 | IntToString(value); |
michael@0 | 568 | } |
michael@0 | 569 | std::string Uint64ToString(uint64_t value) { |
michael@0 | 570 | return IntToStringT<std::string, uint64_t, uint64_t, false>:: |
michael@0 | 571 | IntToString(value); |
michael@0 | 572 | } |
michael@0 | 573 | std::wstring Uint64ToWString(uint64_t value) { |
michael@0 | 574 | return IntToStringT<std::wstring, uint64_t, uint64_t, false>:: |
michael@0 | 575 | IntToString(value); |
michael@0 | 576 | } |
michael@0 | 577 | |
michael@0 | 578 | // Lower-level routine that takes a va_list and appends to a specified |
michael@0 | 579 | // string. All other routines are just convenience wrappers around it. |
michael@0 | 580 | static void StringAppendV(std::string* dst, const char* format, va_list ap) { |
michael@0 | 581 | StringAppendVT(dst, format, ap); |
michael@0 | 582 | } |
michael@0 | 583 | |
michael@0 | 584 | static void StringAppendV(std::wstring* dst, const wchar_t* format, va_list ap) { |
michael@0 | 585 | StringAppendVT(dst, format, ap); |
michael@0 | 586 | } |
michael@0 | 587 | |
michael@0 | 588 | std::string StringPrintf(const char* format, ...) { |
michael@0 | 589 | va_list ap; |
michael@0 | 590 | va_start(ap, format); |
michael@0 | 591 | std::string result; |
michael@0 | 592 | StringAppendV(&result, format, ap); |
michael@0 | 593 | va_end(ap); |
michael@0 | 594 | return result; |
michael@0 | 595 | } |
michael@0 | 596 | |
michael@0 | 597 | std::wstring StringPrintf(const wchar_t* format, ...) { |
michael@0 | 598 | va_list ap; |
michael@0 | 599 | va_start(ap, format); |
michael@0 | 600 | std::wstring result; |
michael@0 | 601 | StringAppendV(&result, format, ap); |
michael@0 | 602 | va_end(ap); |
michael@0 | 603 | return result; |
michael@0 | 604 | } |
michael@0 | 605 | |
michael@0 | 606 | const std::string& SStringPrintf(std::string* dst, const char* format, ...) { |
michael@0 | 607 | va_list ap; |
michael@0 | 608 | va_start(ap, format); |
michael@0 | 609 | dst->clear(); |
michael@0 | 610 | StringAppendV(dst, format, ap); |
michael@0 | 611 | va_end(ap); |
michael@0 | 612 | return *dst; |
michael@0 | 613 | } |
michael@0 | 614 | |
michael@0 | 615 | const std::wstring& SStringPrintf(std::wstring* dst, |
michael@0 | 616 | const wchar_t* format, ...) { |
michael@0 | 617 | va_list ap; |
michael@0 | 618 | va_start(ap, format); |
michael@0 | 619 | dst->clear(); |
michael@0 | 620 | StringAppendV(dst, format, ap); |
michael@0 | 621 | va_end(ap); |
michael@0 | 622 | return *dst; |
michael@0 | 623 | } |
michael@0 | 624 | |
michael@0 | 625 | void StringAppendF(std::string* dst, const char* format, ...) { |
michael@0 | 626 | va_list ap; |
michael@0 | 627 | va_start(ap, format); |
michael@0 | 628 | StringAppendV(dst, format, ap); |
michael@0 | 629 | va_end(ap); |
michael@0 | 630 | } |
michael@0 | 631 | |
michael@0 | 632 | void StringAppendF(std::wstring* dst, const wchar_t* format, ...) { |
michael@0 | 633 | va_list ap; |
michael@0 | 634 | va_start(ap, format); |
michael@0 | 635 | StringAppendV(dst, format, ap); |
michael@0 | 636 | va_end(ap); |
michael@0 | 637 | } |
michael@0 | 638 | |
michael@0 | 639 | template<typename STR> |
michael@0 | 640 | static void SplitStringT(const STR& str, |
michael@0 | 641 | const typename STR::value_type s, |
michael@0 | 642 | bool trim_whitespace, |
michael@0 | 643 | std::vector<STR>* r) { |
michael@0 | 644 | size_t last = 0; |
michael@0 | 645 | size_t i; |
michael@0 | 646 | size_t c = str.size(); |
michael@0 | 647 | for (i = 0; i <= c; ++i) { |
michael@0 | 648 | if (i == c || str[i] == s) { |
michael@0 | 649 | size_t len = i - last; |
michael@0 | 650 | STR tmp = str.substr(last, len); |
michael@0 | 651 | if (trim_whitespace) { |
michael@0 | 652 | STR t_tmp; |
michael@0 | 653 | TrimWhitespace(tmp, TRIM_ALL, &t_tmp); |
michael@0 | 654 | r->push_back(t_tmp); |
michael@0 | 655 | } else { |
michael@0 | 656 | r->push_back(tmp); |
michael@0 | 657 | } |
michael@0 | 658 | last = i + 1; |
michael@0 | 659 | } |
michael@0 | 660 | } |
michael@0 | 661 | } |
michael@0 | 662 | |
michael@0 | 663 | void SplitString(const std::wstring& str, |
michael@0 | 664 | wchar_t s, |
michael@0 | 665 | std::vector<std::wstring>* r) { |
michael@0 | 666 | SplitStringT(str, s, true, r); |
michael@0 | 667 | } |
michael@0 | 668 | |
michael@0 | 669 | void SplitString(const std::string& str, |
michael@0 | 670 | char s, |
michael@0 | 671 | std::vector<std::string>* r) { |
michael@0 | 672 | SplitStringT(str, s, true, r); |
michael@0 | 673 | } |
michael@0 | 674 | |
michael@0 | 675 | // For the various *ToInt conversions, there are no *ToIntTraits classes to use |
michael@0 | 676 | // because there's no such thing as strtoi. Use *ToLongTraits through a cast |
michael@0 | 677 | // instead, requiring that long and int are compatible and equal-width. They |
michael@0 | 678 | // are on our target platforms. |
michael@0 | 679 | |
michael@0 | 680 | // XXX Sigh. |
michael@0 | 681 | |
michael@0 | 682 | #if !defined(ARCH_CPU_64_BITS) |
michael@0 | 683 | bool StringToInt(const std::string& input, int* output) { |
michael@0 | 684 | COMPILE_ASSERT(sizeof(int) == sizeof(long), cannot_strtol_to_int); |
michael@0 | 685 | return StringToNumber<StringToLongTraits>(input, |
michael@0 | 686 | reinterpret_cast<long*>(output)); |
michael@0 | 687 | } |
michael@0 | 688 | |
michael@0 | 689 | bool StringToInt(const string16& input, int* output) { |
michael@0 | 690 | COMPILE_ASSERT(sizeof(int) == sizeof(long), cannot_wcstol_to_int); |
michael@0 | 691 | return StringToNumber<String16ToLongTraits>(input, |
michael@0 | 692 | reinterpret_cast<long*>(output)); |
michael@0 | 693 | } |
michael@0 | 694 | |
michael@0 | 695 | #else |
michael@0 | 696 | bool StringToInt(const std::string& input, int* output) { |
michael@0 | 697 | long tmp; |
michael@0 | 698 | bool ok = StringToNumber<StringToLongTraits>(input, &tmp); |
michael@0 | 699 | if (!ok || tmp > kint32max) { |
michael@0 | 700 | return false; |
michael@0 | 701 | } |
michael@0 | 702 | *output = static_cast<int>(tmp); |
michael@0 | 703 | return true; |
michael@0 | 704 | } |
michael@0 | 705 | |
michael@0 | 706 | bool StringToInt(const string16& input, int* output) { |
michael@0 | 707 | long tmp; |
michael@0 | 708 | bool ok = StringToNumber<String16ToLongTraits>(input, &tmp); |
michael@0 | 709 | if (!ok || tmp > kint32max) { |
michael@0 | 710 | return false; |
michael@0 | 711 | } |
michael@0 | 712 | *output = static_cast<int>(tmp); |
michael@0 | 713 | return true; |
michael@0 | 714 | } |
michael@0 | 715 | #endif // !defined(ARCH_CPU_64_BITS) |
michael@0 | 716 | |
michael@0 | 717 | bool StringToInt64(const std::string& input, int64_t* output) { |
michael@0 | 718 | return StringToNumber<StringToInt64Traits>(input, output); |
michael@0 | 719 | } |
michael@0 | 720 | |
michael@0 | 721 | bool StringToInt64(const string16& input, int64_t* output) { |
michael@0 | 722 | return StringToNumber<String16ToInt64Traits>(input, output); |
michael@0 | 723 | } |
michael@0 | 724 | |
michael@0 | 725 | int StringToInt(const std::string& value) { |
michael@0 | 726 | int result; |
michael@0 | 727 | StringToInt(value, &result); |
michael@0 | 728 | return result; |
michael@0 | 729 | } |
michael@0 | 730 | |
michael@0 | 731 | int StringToInt(const string16& value) { |
michael@0 | 732 | int result; |
michael@0 | 733 | StringToInt(value, &result); |
michael@0 | 734 | return result; |
michael@0 | 735 | } |
michael@0 | 736 | |
michael@0 | 737 | int64_t StringToInt64(const std::string& value) { |
michael@0 | 738 | int64_t result; |
michael@0 | 739 | StringToInt64(value, &result); |
michael@0 | 740 | return result; |
michael@0 | 741 | } |
michael@0 | 742 | |
michael@0 | 743 | int64_t StringToInt64(const string16& value) { |
michael@0 | 744 | int64_t result; |
michael@0 | 745 | StringToInt64(value, &result); |
michael@0 | 746 | return result; |
michael@0 | 747 | } |
michael@0 | 748 | |
michael@0 | 749 | // The following code is compatible with the OpenBSD lcpy interface. See: |
michael@0 | 750 | // http://www.gratisoft.us/todd/papers/strlcpy.html |
michael@0 | 751 | // ftp://ftp.openbsd.org/pub/OpenBSD/src/lib/libc/string/{wcs,str}lcpy.c |
michael@0 | 752 | |
michael@0 | 753 | namespace { |
michael@0 | 754 | |
michael@0 | 755 | template <typename CHAR> |
michael@0 | 756 | size_t lcpyT(CHAR* dst, const CHAR* src, size_t dst_size) { |
michael@0 | 757 | for (size_t i = 0; i < dst_size; ++i) { |
michael@0 | 758 | if ((dst[i] = src[i]) == 0) // We hit and copied the terminating NULL. |
michael@0 | 759 | return i; |
michael@0 | 760 | } |
michael@0 | 761 | |
michael@0 | 762 | // We were left off at dst_size. We over copied 1 byte. Null terminate. |
michael@0 | 763 | if (dst_size != 0) |
michael@0 | 764 | dst[dst_size - 1] = 0; |
michael@0 | 765 | |
michael@0 | 766 | // Count the rest of the |src|, and return it's length in characters. |
michael@0 | 767 | while (src[dst_size]) ++dst_size; |
michael@0 | 768 | return dst_size; |
michael@0 | 769 | } |
michael@0 | 770 | |
michael@0 | 771 | } // namespace |
michael@0 | 772 | |
michael@0 | 773 | size_t base::strlcpy(char* dst, const char* src, size_t dst_size) { |
michael@0 | 774 | return lcpyT<char>(dst, src, dst_size); |
michael@0 | 775 | } |
michael@0 | 776 | size_t base::wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size) { |
michael@0 | 777 | return lcpyT<wchar_t>(dst, src, dst_size); |
michael@0 | 778 | } |