license.bot | f003cfe | 2008-08-24 09:55:55 +0900 | [diff] [blame^] | 1 | // Copyright (c) 2006-2008 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. |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 4 | |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 5 | #include "base/string_util.h" |
| 6 | |
| 7 | #include <string.h> |
| 8 | #include <vector> |
| 9 | |
| 10 | #include "base/basictypes.h" |
| 11 | #include "base/logging.h" |
| 12 | #include "base/singleton.h" |
| 13 | #include "unicode/ucnv.h" |
| 14 | #include "unicode/numfmt.h" |
| 15 | #include "unicode/ustring.h" |
| 16 | |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 17 | namespace { |
| 18 | |
| 19 | // ReadUnicodeCharacter -------------------------------------------------------- |
| 20 | |
| 21 | // Reads a UTF-8 stream, placing the next code point into the given output |
| 22 | // |*code_point|. |src| represents the entire string to read, and |*char_index| |
| 23 | // is the character offset within the string to start reading at. |*char_index| |
| 24 | // will be updated to index the last character read, such that incrementing it |
| 25 | // (as in a for loop) will take the reader to the next character. |
| 26 | // |
| 27 | // Returns true on success. On false, |*code_point| will be invalid. |
| 28 | bool ReadUnicodeCharacter(const char* src, int32 src_len, |
evanm@google.com | e1581aa | 2008-08-19 09:31:24 +0900 | [diff] [blame] | 29 | int32* char_index, uint32* code_point_out) { |
| 30 | // U8_NEXT expects to be able to use -1 to signal an error, so we must |
| 31 | // use a signed type for code_point. But this function returns false |
| 32 | // on error anyway, so code_point_out is unsigned. |
| 33 | int32 code_point; |
| 34 | U8_NEXT(src, *char_index, src_len, code_point); |
| 35 | *code_point_out = static_cast<uint32>(code_point); |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 36 | |
| 37 | // The ICU macro above moves to the next char, we want to point to the last |
| 38 | // char consumed. |
| 39 | (*char_index)--; |
| 40 | |
| 41 | // Validate the decoded value. |
evanm@google.com | e1581aa | 2008-08-19 09:31:24 +0900 | [diff] [blame] | 42 | return U_IS_UNICODE_CHAR(code_point); |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 43 | } |
| 44 | |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 45 | // Reads a UTF-16 character. The usage is the same as the 8-bit version above. |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 46 | bool ReadUnicodeCharacter(const char16* src, int32 src_len, |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 47 | int32* char_index, uint32* code_point) { |
| 48 | if (U16_IS_SURROGATE(src[*char_index])) { |
| 49 | if (!U16_IS_SURROGATE_LEAD(src[*char_index]) || |
| 50 | *char_index + 1 >= src_len || |
| 51 | !U16_IS_TRAIL(src[*char_index + 1])) { |
| 52 | // Invalid surrogate pair. |
| 53 | return false; |
| 54 | } |
| 55 | |
| 56 | // Valid surrogate pair. |
| 57 | *code_point = U16_GET_SUPPLEMENTARY(src[*char_index], |
| 58 | src[*char_index + 1]); |
| 59 | (*char_index)++; |
| 60 | } else { |
| 61 | // Not a surrogate, just one 16-bit word. |
| 62 | *code_point = src[*char_index]; |
| 63 | } |
| 64 | |
| 65 | return U_IS_UNICODE_CHAR(*code_point); |
| 66 | } |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 67 | |
| 68 | #if defined(WCHAR_T_IS_UTF32) |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 69 | // Reads UTF-32 character. The usage is the same as the 8-bit version above. |
pinkerton@google.com | c5c7a72 | 2008-08-08 07:36:01 +0900 | [diff] [blame] | 70 | bool ReadUnicodeCharacter(const wchar_t* src, int32 src_len, |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 71 | int32* char_index, uint32* code_point) { |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 72 | // Conversion is easy since the source is 32-bit. |
| 73 | *code_point = src[*char_index]; |
| 74 | |
| 75 | // Validate the value. |
| 76 | return U_IS_UNICODE_CHAR(*code_point); |
| 77 | } |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 78 | #endif // defined(WCHAR_T_IS_UTF32) |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 79 | |
| 80 | // WriteUnicodeCharacter ------------------------------------------------------- |
| 81 | |
| 82 | // Appends a UTF-8 character to the given 8-bit string. |
| 83 | void WriteUnicodeCharacter(uint32 code_point, std::basic_string<char>* output) { |
| 84 | if (code_point <= 0x7f) { |
| 85 | // Fast path the common case of one byte. |
| 86 | output->push_back(code_point); |
| 87 | return; |
| 88 | } |
| 89 | |
| 90 | // U8_APPEND_UNSAFE can append up to 4 bytes. |
| 91 | int32 char_offset = static_cast<int32>(output->length()); |
| 92 | output->resize(char_offset + U8_MAX_LENGTH); |
| 93 | |
| 94 | U8_APPEND_UNSAFE(&(*output)[0], char_offset, code_point); |
| 95 | |
| 96 | // U8_APPEND_UNSAFE will advance our pointer past the inserted character, so |
| 97 | // it will represent the new length of the string. |
| 98 | output->resize(char_offset); |
| 99 | } |
| 100 | |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 101 | // Appends the given code point as a UTF-16 character to the STL string. |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 102 | void WriteUnicodeCharacter(uint32 code_point, |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 103 | std::basic_string<char16>* output) { |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 104 | if (U16_LENGTH(code_point) == 1) { |
| 105 | // Thie code point is in the Basic Multilingual Plane (BMP). |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 106 | output->push_back(static_cast<char16>(code_point)); |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 107 | } else { |
| 108 | // Non-BMP characters use a double-character encoding. |
| 109 | int32 char_offset = static_cast<int32>(output->length()); |
| 110 | output->resize(char_offset + U16_MAX_LENGTH); |
| 111 | U16_APPEND_UNSAFE(&(*output)[0], char_offset, code_point); |
| 112 | } |
| 113 | } |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 114 | |
| 115 | #if defined(WCHAR_T_IS_UTF32) |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 116 | // Appends the given UTF-32 character to the given 32-bit string. |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 117 | inline void WriteUnicodeCharacter(uint32 code_point, |
| 118 | std::basic_string<wchar_t>* output) { |
| 119 | // This is the easy case, just append the character. |
| 120 | output->push_back(code_point); |
| 121 | } |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 122 | #endif // defined(WCHAR_T_IS_UTF32) |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 123 | |
| 124 | // Generalized Unicode converter ----------------------------------------------- |
| 125 | |
| 126 | // Converts the given source Unicode character type to the given destination |
| 127 | // Unicode character type as a STL string. The given input buffer and size |
| 128 | // determine the source, and the given output STL string will be replaced by |
| 129 | // the result. |
| 130 | template<typename SRC_CHAR, typename DEST_CHAR> |
| 131 | bool ConvertUnicode(const SRC_CHAR* src, size_t src_len, |
| 132 | std::basic_string<DEST_CHAR>* output) { |
| 133 | output->clear(); |
| 134 | |
| 135 | // ICU requires 32-bit numbers. |
| 136 | bool success = true; |
| 137 | int32 src_len32 = static_cast<int32>(src_len); |
| 138 | for (int32 i = 0; i < src_len32; i++) { |
| 139 | uint32 code_point; |
| 140 | if (ReadUnicodeCharacter(src, src_len32, &i, &code_point)) |
| 141 | WriteUnicodeCharacter(code_point, output); |
| 142 | else |
| 143 | success = false; |
| 144 | } |
| 145 | return success; |
| 146 | } |
| 147 | |
| 148 | } // namespace |
| 149 | |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 150 | // UTF-8 <-> Wide -------------------------------------------------------------- |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 151 | |
| 152 | std::string WideToUTF8(const std::wstring& wide) { |
| 153 | std::string ret; |
| 154 | if (wide.empty()) |
| 155 | return ret; |
| 156 | |
| 157 | // Ignore the success flag of this call, it will do the best it can for |
| 158 | // invalid input, which is what we want here. |
| 159 | WideToUTF8(wide.data(), wide.length(), &ret); |
| 160 | return ret; |
| 161 | } |
| 162 | |
| 163 | bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output) { |
| 164 | if (src_len == 0) { |
| 165 | output->clear(); |
| 166 | return true; |
| 167 | } |
| 168 | |
| 169 | // Intelligently guess the size of the output string. When it's an ASCII |
| 170 | // character, assume the rest will be ASCII and use a buffer size the same as |
| 171 | // the input. When it's not ASCII, assume 3-bytes per character as the |
| 172 | // starting point. This will be resized internally later if it's too small. |
mmentovai@google.com | 38cabad | 2008-08-13 10:17:18 +0900 | [diff] [blame] | 173 | if (static_cast<uint32>(src[0]) < 0x80) |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 174 | output->reserve(src_len); |
| 175 | else |
| 176 | output->reserve(src_len * 3); |
| 177 | return ConvertUnicode<wchar_t, char>(src, src_len, output); |
| 178 | } |
| 179 | |
| 180 | std::wstring UTF8ToWide(const std::string& utf8) { |
| 181 | std::wstring ret; |
| 182 | if (utf8.empty()) |
| 183 | return ret; |
| 184 | |
| 185 | UTF8ToWide(utf8.data(), utf8.length(), &ret); |
| 186 | return ret; |
| 187 | } |
| 188 | |
| 189 | bool UTF8ToWide(const char* src, size_t src_len, std::wstring* output) { |
| 190 | if (src_len == 0) { |
| 191 | output->clear(); |
| 192 | return true; |
| 193 | } |
| 194 | |
| 195 | // Intelligently guess the size of the output string. When it's an ASCII |
| 196 | // character, assume the rest will be ASCII and use a buffer size the same as |
| 197 | // the input. When it's not ASCII, assume the UTF-8 takes 2 bytes per |
| 198 | // character (this is more conservative than 3 which we use above when |
| 199 | // converting the other way). |
mmentovai@google.com | 38cabad | 2008-08-13 10:17:18 +0900 | [diff] [blame] | 200 | if (static_cast<unsigned char>(src[0]) < 0x80) |
brettw@google.com | fed55ab | 2008-08-08 00:29:49 +0900 | [diff] [blame] | 201 | output->reserve(src_len); |
| 202 | else |
| 203 | output->reserve(src_len / 2); |
| 204 | return ConvertUnicode<char, wchar_t>(src, src_len, output); |
| 205 | } |
| 206 | |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 207 | // UTF-16 <-> Wide ------------------------------------------------------------- |
| 208 | |
| 209 | #if defined(WCHAR_T_IS_UTF16) |
| 210 | |
| 211 | // When wide == UTF-16, then conversions are a NOP. |
| 212 | std::string16 WideToUTF16(const std::wstring& wide) { |
| 213 | return wide; |
| 214 | } |
| 215 | |
| 216 | bool WideToUTF16(const wchar_t* src, size_t src_len, std::string16* output) { |
| 217 | output->assign(src, src_len); |
| 218 | return true; |
| 219 | } |
| 220 | |
| 221 | std::wstring UTF16ToWide(const std::string16& utf16) { |
| 222 | return utf16; |
| 223 | } |
| 224 | |
| 225 | bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) { |
| 226 | output->assign(src, src_len); |
| 227 | return true; |
| 228 | } |
| 229 | |
| 230 | #elif defined(WCHAR_T_IS_UTF32) |
| 231 | |
| 232 | std::string16 WideToUTF16(const std::wstring& wide) { |
| 233 | std::string16 ret; |
| 234 | if (wide.empty()) |
| 235 | return ret; |
| 236 | |
pinkerton@google.com | c5c7a72 | 2008-08-08 07:36:01 +0900 | [diff] [blame] | 237 | WideToUTF16(wide.data(), wide.length(), &ret); |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 238 | return ret; |
| 239 | } |
| 240 | |
| 241 | bool WideToUTF16(const wchar_t* src, size_t src_len, std::string16* output) { |
| 242 | if (src_len == 0) { |
| 243 | output->clear(); |
| 244 | return true; |
| 245 | } |
| 246 | |
| 247 | // Assume that normally we won't have any non-BMP characters so the counts |
| 248 | // will be the same. |
| 249 | output->reserve(src_len); |
| 250 | return ConvertUnicode<wchar_t, char16>(src, src_len, output); |
| 251 | } |
| 252 | |
| 253 | std::wstring UTF16ToWide(const std::string16& utf16) { |
| 254 | std::wstring ret; |
| 255 | if (utf16.empty()) |
| 256 | return ret; |
| 257 | |
pinkerton@google.com | c5c7a72 | 2008-08-08 07:36:01 +0900 | [diff] [blame] | 258 | UTF16ToWide(utf16.data(), utf16.length(), &ret); |
brettw@google.com | fa49905 | 2008-08-08 05:27:57 +0900 | [diff] [blame] | 259 | return ret; |
| 260 | } |
| 261 | |
| 262 | bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) { |
| 263 | if (src_len == 0) { |
| 264 | output->clear(); |
| 265 | return true; |
| 266 | } |
| 267 | |
| 268 | // Assume that normally we won't have any non-BMP characters so the counts |
| 269 | // will be the same. |
| 270 | output->reserve(src_len); |
| 271 | return ConvertUnicode<char16, wchar_t>(src, src_len, output); |
| 272 | } |
| 273 | |
| 274 | #endif // defined(WCHAR_T_IS_UTF32) |
| 275 | |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 276 | // Codepage <-> Wide ----------------------------------------------------------- |
| 277 | |
| 278 | // Convert a unicode string into the specified codepage_name. If the codepage |
| 279 | // isn't found, return false. |
| 280 | bool WideToCodepage(const std::wstring& wide, |
| 281 | const char* codepage_name, |
| 282 | OnStringUtilConversionError::Type on_error, |
| 283 | std::string* encoded) { |
| 284 | encoded->clear(); |
| 285 | |
| 286 | UErrorCode status = U_ZERO_ERROR; |
| 287 | UConverter* converter = ucnv_open(codepage_name, &status); |
| 288 | if (!U_SUCCESS(status)) |
| 289 | return false; |
| 290 | |
| 291 | const UChar* uchar_src; |
| 292 | int uchar_len; |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 293 | #if defined(WCHAR_T_IS_UTF16) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 294 | uchar_src = wide.c_str(); |
| 295 | uchar_len = static_cast<int>(wide.length()); |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 296 | #elif defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 297 | // When wchar_t is wider than UChar (16 bits), transform |wide| into a |
| 298 | // UChar* string. Size the UChar* buffer to be large enough to hold twice |
mmentovai@google.com | 38cabad | 2008-08-13 10:17:18 +0900 | [diff] [blame] | 299 | // as many UTF-16 code points as there are UTF-16 characters, in case each |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 300 | // character translates to a UTF-16 surrogate pair, and leave room for a NUL |
| 301 | // terminator. |
| 302 | std::vector<UChar> wide_uchar(wide.length() * 2 + 1); |
| 303 | u_strFromWCS(&wide_uchar[0], wide_uchar.size(), &uchar_len, |
| 304 | wide.c_str(), wide.length(), &status); |
| 305 | uchar_src = &wide_uchar[0]; |
| 306 | DCHECK(U_SUCCESS(status)) << "failed to convert wstring to UChar*"; |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 307 | #endif // defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 308 | |
| 309 | int encoded_max_length = UCNV_GET_MAX_BYTES_FOR_STRING(uchar_len, |
| 310 | ucnv_getMaxCharSize(converter)); |
| 311 | encoded->resize(encoded_max_length); |
| 312 | |
| 313 | // Setup our error handler. |
| 314 | switch (on_error) { |
| 315 | case OnStringUtilConversionError::FAIL: |
| 316 | ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_STOP, 0, |
| 317 | NULL, NULL, &status); |
| 318 | break; |
| 319 | case OnStringUtilConversionError::SKIP: |
| 320 | ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SKIP, 0, |
| 321 | NULL, NULL, &status); |
| 322 | break; |
| 323 | default: |
| 324 | NOTREACHED(); |
| 325 | } |
| 326 | |
| 327 | // ucnv_fromUChars returns size not including terminating null |
| 328 | int actual_size = ucnv_fromUChars(converter, &(*encoded)[0], |
| 329 | encoded_max_length, uchar_src, uchar_len, &status); |
| 330 | encoded->resize(actual_size); |
| 331 | ucnv_close(converter); |
| 332 | if (U_SUCCESS(status)) |
| 333 | return true; |
| 334 | encoded->clear(); // Make sure the output is empty on error. |
| 335 | return false; |
| 336 | } |
| 337 | |
| 338 | // Converts a string of the given codepage into unicode. |
| 339 | // If the codepage isn't found, return false. |
| 340 | bool CodepageToWide(const std::string& encoded, |
| 341 | const char* codepage_name, |
| 342 | OnStringUtilConversionError::Type on_error, |
| 343 | std::wstring* wide) { |
| 344 | wide->clear(); |
| 345 | |
| 346 | UErrorCode status = U_ZERO_ERROR; |
| 347 | UConverter* converter = ucnv_open(codepage_name, &status); |
| 348 | if (!U_SUCCESS(status)) |
| 349 | return false; |
| 350 | |
| 351 | // The worst case is all the input characters are non-BMP (32-bit) ones. |
| 352 | size_t uchar_max_length = encoded.length() * 2 + 1; |
| 353 | |
| 354 | UChar* uchar_dst; |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 355 | #if defined(WCHAR_T_IS_UTF16) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 356 | uchar_dst = WriteInto(wide, uchar_max_length); |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 357 | #elif defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 358 | // When wchar_t is wider than UChar (16 bits), convert into a temporary |
| 359 | // UChar* buffer. |
| 360 | std::vector<UChar> wide_uchar(uchar_max_length); |
| 361 | uchar_dst = &wide_uchar[0]; |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 362 | #endif // defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 363 | |
| 364 | // Setup our error handler. |
| 365 | switch (on_error) { |
| 366 | case OnStringUtilConversionError::FAIL: |
| 367 | ucnv_setToUCallBack(converter, UCNV_TO_U_CALLBACK_STOP, 0, |
| 368 | NULL, NULL, &status); |
| 369 | break; |
| 370 | case OnStringUtilConversionError::SKIP: |
| 371 | ucnv_setToUCallBack(converter, UCNV_TO_U_CALLBACK_SKIP, 0, |
| 372 | NULL, NULL, &status); |
| 373 | break; |
| 374 | default: |
| 375 | NOTREACHED(); |
| 376 | } |
| 377 | |
| 378 | int actual_size = ucnv_toUChars(converter, |
| 379 | uchar_dst, |
| 380 | static_cast<int>(uchar_max_length), |
| 381 | encoded.data(), |
| 382 | static_cast<int>(encoded.length()), |
| 383 | &status); |
| 384 | ucnv_close(converter); |
| 385 | if (!U_SUCCESS(status)) { |
| 386 | wide->clear(); // Make sure the output is empty on error. |
| 387 | return false; |
| 388 | } |
| 389 | |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 390 | #ifdef WCHAR_T_IS_UTF32 |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 391 | // When wchar_t is wider than UChar (16 bits), it's not possible to wind up |
| 392 | // with any more wchar_t elements than UChar elements. ucnv_toUChars |
| 393 | // returns the number of UChar elements not including the NUL terminator, so |
| 394 | // leave extra room for that. |
| 395 | u_strToWCS(WriteInto(wide, actual_size + 1), actual_size + 1, &actual_size, |
| 396 | uchar_dst, actual_size, &status); |
| 397 | DCHECK(U_SUCCESS(status)) << "failed to convert UChar* to wstring"; |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 398 | #endif // WCHAR_T_IS_UTF32 |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 399 | |
| 400 | wide->resize(actual_size); |
| 401 | return true; |
| 402 | } |
| 403 | |
| 404 | // Number formatting ----------------------------------------------------------- |
| 405 | |
mmentovai@google.com | 50a0414 | 2008-08-14 00:32:27 +0900 | [diff] [blame] | 406 | namespace { |
| 407 | |
| 408 | struct NumberFormatSingletonTraits |
| 409 | : public DefaultSingletonTraits<NumberFormat> { |
| 410 | static NumberFormat* New() { |
| 411 | UErrorCode status = U_ZERO_ERROR; |
| 412 | NumberFormat* formatter = NumberFormat::createInstance(status); |
| 413 | DCHECK(U_SUCCESS(status)); |
| 414 | return formatter; |
| 415 | } |
| 416 | // There's no ICU call to destroy a NumberFormat object other than |
| 417 | // operator delete, so use the default Delete, which calls operator delete. |
| 418 | // This can cause problems if a different allocator is used by this file than |
| 419 | // by ICU. |
| 420 | }; |
| 421 | |
| 422 | } // namespace |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 423 | |
| 424 | std::wstring FormatNumber(int64 number) { |
mmentovai@google.com | 50a0414 | 2008-08-14 00:32:27 +0900 | [diff] [blame] | 425 | NumberFormat* number_format = |
| 426 | Singleton<NumberFormat, NumberFormatSingletonTraits>::get(); |
| 427 | |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 428 | if (!number_format) { |
| 429 | // As a fallback, just return the raw number in a string. |
| 430 | return StringPrintf(L"%lld", number); |
| 431 | } |
| 432 | UnicodeString ustr; |
| 433 | number_format->format(number, ustr); |
| 434 | |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 435 | #if defined(WCHAR_T_IS_UTF16) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 436 | return std::wstring(ustr.getBuffer(), |
| 437 | static_cast<std::wstring::size_type>(ustr.length())); |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 438 | #elif defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 439 | wchar_t buffer[64]; // A int64 is less than 20 chars long, so 64 chars |
| 440 | // leaves plenty of room for formating stuff. |
| 441 | int length = 0; |
| 442 | UErrorCode error = U_ZERO_ERROR; |
| 443 | u_strToWCS(buffer, 64, &length, ustr.getBuffer(), ustr.length() , &error); |
| 444 | if (U_FAILURE(error)) { |
| 445 | NOTREACHED(); |
| 446 | // As a fallback, just return the raw number in a string. |
| 447 | return StringPrintf(L"%lld", number); |
| 448 | } |
| 449 | return std::wstring(buffer, static_cast<std::wstring::size_type>(length)); |
brettw@google.com | e3c034a | 2008-08-08 03:31:40 +0900 | [diff] [blame] | 450 | #endif // defined(WCHAR_T_IS_UTF32) |
initial.commit | 3f4a732 | 2008-07-27 06:49:38 +0900 | [diff] [blame] | 451 | } |
license.bot | f003cfe | 2008-08-24 09:55:55 +0900 | [diff] [blame^] | 452 | |