1 // Copyright 2013 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 // This file defines utility functions for working with strings. 6 7 #ifndef BASE_STRINGS_STRING_UTIL_H_ 8 #define BASE_STRINGS_STRING_UTIL_H_ 9 10 #include <ctype.h> 11 #include <stdarg.h> // va_list 12 #include <stddef.h> 13 #include <stdint.h> 14 15 #include <string> 16 #include <vector> 17 18 #include "base/base_export.h" 19 #include "base/compiler_specific.h" 20 #include "base/strings/string16.h" 21 #include "base/strings/string_piece.h" // For implicit conversions. 22 #include "build/build_config.h" 23 24 #if defined(ANDROID) 25 // On Android, bionic's stdio.h defines an snprintf macro when being built with 26 // clang. Undefine it here so it won't collide with base::snprintf(). 27 #undef snprintf 28 #endif // defined(ANDROID) 29 30 namespace base { 31 32 // C standard-library functions that aren't cross-platform are provided as 33 // "base::...", and their prototypes are listed below. These functions are 34 // then implemented as inline calls to the platform-specific equivalents in the 35 // platform-specific headers. 36 37 // Wrapper for vsnprintf that always null-terminates and always returns the 38 // number of characters that would be in an untruncated formatted 39 // string, even when truncation occurs. 40 int vsnprintf(char* buffer, size_t size, const char* format, va_list arguments) 41 PRINTF_FORMAT(3, 0); 42 43 // Some of these implementations need to be inlined. 44 45 // We separate the declaration from the implementation of this inline 46 // function just so the PRINTF_FORMAT works. 47 inline int snprintf(char* buffer, 48 size_t size, 49 _Printf_format_string_ const char* format, 50 ...) PRINTF_FORMAT(3, 4); 51 inline int snprintf(char* buffer, 52 size_t size, 53 _Printf_format_string_ const char* format, 54 ...) { 55 va_list arguments; 56 va_start(arguments, format); 57 int result = vsnprintf(buffer, size, format, arguments); 58 va_end(arguments); 59 return result; 60 } 61 62 // BSD-style safe and consistent string copy functions. 63 // Copies |src| to |dst|, where |dst_size| is the total allocated size of |dst|. 64 // Copies at most |dst_size|-1 characters, and always NULL terminates |dst|, as 65 // long as |dst_size| is not 0. Returns the length of |src| in characters. 66 // If the return value is >= dst_size, then the output was truncated. 67 // NOTE: All sizes are in number of characters, NOT in bytes. 68 BASE_EXPORT size_t strlcpy(char* dst, const char* src, size_t dst_size); 69 BASE_EXPORT size_t wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size); 70 71 // Scan a wprintf format string to determine whether it's portable across a 72 // variety of systems. This function only checks that the conversion 73 // specifiers used by the format string are supported and have the same meaning 74 // on a variety of systems. It doesn't check for other errors that might occur 75 // within a format string. 76 // 77 // Nonportable conversion specifiers for wprintf are: 78 // - 's' and 'c' without an 'l' length modifier. %s and %c operate on char 79 // data on all systems except Windows, which treat them as wchar_t data. 80 // Use %ls and %lc for wchar_t data instead. 81 // - 'S' and 'C', which operate on wchar_t data on all systems except Windows, 82 // which treat them as char data. Use %ls and %lc for wchar_t data 83 // instead. 84 // - 'F', which is not identified by Windows wprintf documentation. 85 // - 'D', 'O', and 'U', which are deprecated and not available on all systems. 86 // Use %ld, %lo, and %lu instead. 87 // 88 // Note that there is no portable conversion specifier for char data when 89 // working with wprintf. 90 // 91 // This function is intended to be called from base::vswprintf. 92 BASE_EXPORT bool IsWprintfFormatPortable(const wchar_t* format); 93 94 // ASCII-specific tolower. The standard library's tolower is locale sensitive, 95 // so we don't want to use it here. 96 inline char ToLowerASCII(char c) { 97 return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c; 98 } 99 inline char16 ToLowerASCII(char16 c) { 100 return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c; 101 } 102 103 // ASCII-specific toupper. The standard library's toupper is locale sensitive, 104 // so we don't want to use it here. 105 inline char ToUpperASCII(char c) { 106 return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c; 107 } 108 inline char16 ToUpperASCII(char16 c) { 109 return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c; 110 } 111 112 // Converts the given string to it's ASCII-lowercase equivalent. 113 BASE_EXPORT std::string ToLowerASCII(StringPiece str); 114 BASE_EXPORT string16 ToLowerASCII(StringPiece16 str); 115 116 // Converts the given string to it's ASCII-uppercase equivalent. 117 BASE_EXPORT std::string ToUpperASCII(StringPiece str); 118 BASE_EXPORT string16 ToUpperASCII(StringPiece16 str); 119 120 // Functor for case-insensitive ASCII comparisons for STL algorithms like 121 // std::search. 122 // 123 // Note that a full Unicode version of this functor is not possible to write 124 // because case mappings might change the number of characters, depend on 125 // context (combining accents), and require handling UTF-16. If you need 126 // proper Unicode support, use base::i18n::ToLower/FoldCase and then just 127 // use a normal operator== on the result. 128 template<typename Char> struct CaseInsensitiveCompareASCII { 129 public: 130 bool operator()(Char x, Char y) const { 131 return ToLowerASCII(x) == ToLowerASCII(y); 132 } 133 }; 134 135 // Like strcasecmp for case-insensitive ASCII characters only. Returns: 136 // -1 (a < b) 137 // 0 (a == b) 138 // 1 (a > b) 139 // (unlike strcasecmp which can return values greater or less than 1/-1). For 140 // full Unicode support, use base::i18n::ToLower or base::i18h::FoldCase 141 // and then just call the normal string operators on the result. 142 BASE_EXPORT int CompareCaseInsensitiveASCII(StringPiece a, StringPiece b); 143 BASE_EXPORT int CompareCaseInsensitiveASCII(StringPiece16 a, StringPiece16 b); 144 145 // Equality for ASCII case-insensitive comparisons. For full Unicode support, 146 // use base::i18n::ToLower or base::i18h::FoldCase and then compare with either 147 // == or !=. 148 BASE_EXPORT bool EqualsCaseInsensitiveASCII(StringPiece a, StringPiece b); 149 BASE_EXPORT bool EqualsCaseInsensitiveASCII(StringPiece16 a, StringPiece16 b); 150 151 // These threadsafe functions return references to globally unique empty 152 // strings. 153 // 154 // It is likely faster to construct a new empty string object (just a few 155 // instructions to set the length to 0) than to get the empty string singleton 156 // returned by these functions (which requires threadsafe singleton access). 157 // 158 // Therefore, DO NOT USE THESE AS A GENERAL-PURPOSE SUBSTITUTE FOR DEFAULT 159 // CONSTRUCTORS. There is only one case where you should use these: functions 160 // which need to return a string by reference (e.g. as a class member 161 // accessor), and don't have an empty string to use (e.g. in an error case). 162 // These should not be used as initializers, function arguments, or return 163 // values for functions which return by value or outparam. 164 BASE_EXPORT const std::string& EmptyString(); 165 BASE_EXPORT const string16& EmptyString16(); 166 167 // Contains the set of characters representing whitespace in the corresponding 168 // encoding. Null-terminated. The ASCII versions are the whitespaces as defined 169 // by HTML5, and don't include control characters. 170 BASE_EXPORT extern const wchar_t kWhitespaceWide[]; // Includes Unicode. 171 BASE_EXPORT extern const char16 kWhitespaceUTF16[]; // Includes Unicode. 172 BASE_EXPORT extern const char kWhitespaceASCII[]; 173 BASE_EXPORT extern const char16 kWhitespaceASCIIAs16[]; // No unicode. 174 175 // Null-terminated string representing the UTF-8 byte order mark. 176 BASE_EXPORT extern const char kUtf8ByteOrderMark[]; 177 178 // Removes characters in |remove_chars| from anywhere in |input|. Returns true 179 // if any characters were removed. |remove_chars| must be null-terminated. 180 // NOTE: Safe to use the same variable for both |input| and |output|. 181 BASE_EXPORT bool RemoveChars(const string16& input, 182 const StringPiece16& remove_chars, 183 string16* output); 184 BASE_EXPORT bool RemoveChars(const std::string& input, 185 const StringPiece& remove_chars, 186 std::string* output); 187 188 // Replaces characters in |replace_chars| from anywhere in |input| with 189 // |replace_with|. Each character in |replace_chars| will be replaced with 190 // the |replace_with| string. Returns true if any characters were replaced. 191 // |replace_chars| must be null-terminated. 192 // NOTE: Safe to use the same variable for both |input| and |output|. 193 BASE_EXPORT bool ReplaceChars(const string16& input, 194 const StringPiece16& replace_chars, 195 const string16& replace_with, 196 string16* output); 197 BASE_EXPORT bool ReplaceChars(const std::string& input, 198 const StringPiece& replace_chars, 199 const std::string& replace_with, 200 std::string* output); 201 202 enum TrimPositions { 203 TRIM_NONE = 0, 204 TRIM_LEADING = 1 << 0, 205 TRIM_TRAILING = 1 << 1, 206 TRIM_ALL = TRIM_LEADING | TRIM_TRAILING, 207 }; 208 209 // Removes characters in |trim_chars| from the beginning and end of |input|. 210 // The 8-bit version only works on 8-bit characters, not UTF-8. 211 // 212 // It is safe to use the same variable for both |input| and |output| (this is 213 // the normal usage to trim in-place). 214 BASE_EXPORT bool TrimString(const string16& input, 215 StringPiece16 trim_chars, 216 string16* output); 217 BASE_EXPORT bool TrimString(const std::string& input, 218 StringPiece trim_chars, 219 std::string* output); 220 221 // StringPiece versions of the above. The returned pieces refer to the original 222 // buffer. 223 BASE_EXPORT StringPiece16 TrimString(StringPiece16 input, 224 const StringPiece16& trim_chars, 225 TrimPositions positions); 226 BASE_EXPORT StringPiece TrimString(StringPiece input, 227 const StringPiece& trim_chars, 228 TrimPositions positions); 229 230 // Truncates a string to the nearest UTF-8 character that will leave 231 // the string less than or equal to the specified byte size. 232 BASE_EXPORT void TruncateUTF8ToByteSize(const std::string& input, 233 const size_t byte_size, 234 std::string* output); 235 236 // Trims any whitespace from either end of the input string. 237 // 238 // The StringPiece versions return a substring referencing the input buffer. 239 // The ASCII versions look only for ASCII whitespace. 240 // 241 // The std::string versions return where whitespace was found. 242 // NOTE: Safe to use the same variable for both input and output. 243 BASE_EXPORT TrimPositions TrimWhitespace(const string16& input, 244 TrimPositions positions, 245 string16* output); 246 BASE_EXPORT StringPiece16 TrimWhitespace(StringPiece16 input, 247 TrimPositions positions); 248 BASE_EXPORT TrimPositions TrimWhitespaceASCII(const std::string& input, 249 TrimPositions positions, 250 std::string* output); 251 BASE_EXPORT StringPiece TrimWhitespaceASCII(StringPiece input, 252 TrimPositions positions); 253 254 // Searches for CR or LF characters. Removes all contiguous whitespace 255 // strings that contain them. This is useful when trying to deal with text 256 // copied from terminals. 257 // Returns |text|, with the following three transformations: 258 // (1) Leading and trailing whitespace is trimmed. 259 // (2) If |trim_sequences_with_line_breaks| is true, any other whitespace 260 // sequences containing a CR or LF are trimmed. 261 // (3) All other whitespace sequences are converted to single spaces. 262 BASE_EXPORT string16 CollapseWhitespace( 263 const string16& text, 264 bool trim_sequences_with_line_breaks); 265 BASE_EXPORT std::string CollapseWhitespaceASCII( 266 const std::string& text, 267 bool trim_sequences_with_line_breaks); 268 269 // Returns true if |input| is empty or contains only characters found in 270 // |characters|. 271 BASE_EXPORT bool ContainsOnlyChars(const StringPiece& input, 272 const StringPiece& characters); 273 BASE_EXPORT bool ContainsOnlyChars(const StringPiece16& input, 274 const StringPiece16& characters); 275 276 // Returns true if the specified string matches the criteria. How can a wide 277 // string be 8-bit or UTF8? It contains only characters that are < 256 (in the 278 // first case) or characters that use only 8-bits and whose 8-bit 279 // representation looks like a UTF-8 string (the second case). 280 // 281 // Note that IsStringUTF8 checks not only if the input is structurally 282 // valid but also if it doesn't contain any non-character codepoint 283 // (e.g. U+FFFE). It's done on purpose because all the existing callers want 284 // to have the maximum 'discriminating' power from other encodings. If 285 // there's a use case for just checking the structural validity, we have to 286 // add a new function for that. 287 // 288 // IsStringASCII assumes the input is likely all ASCII, and does not leave early 289 // if it is not the case. 290 BASE_EXPORT bool IsStringUTF8(const StringPiece& str); 291 BASE_EXPORT bool IsStringASCII(const StringPiece& str); 292 BASE_EXPORT bool IsStringASCII(const StringPiece16& str); 293 // A convenience adaptor for WebStrings, as they don't convert into 294 // StringPieces directly. 295 BASE_EXPORT bool IsStringASCII(const string16& str); 296 #if defined(WCHAR_T_IS_UTF32) 297 BASE_EXPORT bool IsStringASCII(const std::wstring& str); 298 #endif 299 300 // Compare the lower-case form of the given string against the given 301 // previously-lower-cased ASCII string (typically a constant). 302 BASE_EXPORT bool LowerCaseEqualsASCII(StringPiece str, 303 StringPiece lowecase_ascii); 304 BASE_EXPORT bool LowerCaseEqualsASCII(StringPiece16 str, 305 StringPiece lowecase_ascii); 306 307 // Performs a case-sensitive string compare of the given 16-bit string against 308 // the given 8-bit ASCII string (typically a constant). The behavior is 309 // undefined if the |ascii| string is not ASCII. 310 BASE_EXPORT bool EqualsASCII(StringPiece16 str, StringPiece ascii); 311 312 // Indicates case sensitivity of comparisons. Only ASCII case insensitivity 313 // is supported. Full Unicode case-insensitive conversions would need to go in 314 // base/i18n so it can use ICU. 315 // 316 // If you need to do Unicode-aware case-insensitive StartsWith/EndsWith, it's 317 // best to call base::i18n::ToLower() or base::i18n::FoldCase() (see 318 // base/i18n/case_conversion.h for usage advice) on the arguments, and then use 319 // the results to a case-sensitive comparison. 320 enum class CompareCase { 321 SENSITIVE, 322 INSENSITIVE_ASCII, 323 }; 324 325 BASE_EXPORT bool StartsWith(StringPiece str, 326 StringPiece search_for, 327 CompareCase case_sensitivity); 328 BASE_EXPORT bool StartsWith(StringPiece16 str, 329 StringPiece16 search_for, 330 CompareCase case_sensitivity); 331 BASE_EXPORT bool EndsWith(StringPiece str, 332 StringPiece search_for, 333 CompareCase case_sensitivity); 334 BASE_EXPORT bool EndsWith(StringPiece16 str, 335 StringPiece16 search_for, 336 CompareCase case_sensitivity); 337 338 // Determines the type of ASCII character, independent of locale (the C 339 // library versions will change based on locale). 340 template <typename Char> 341 inline bool IsAsciiWhitespace(Char c) { 342 return c == ' ' || c == '\r' || c == '\n' || c == '\t'; 343 } 344 template <typename Char> 345 inline bool IsAsciiAlpha(Char c) { 346 return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z'); 347 } 348 template <typename Char> 349 inline bool IsAsciiUpper(Char c) { 350 return c >= 'A' && c <= 'Z'; 351 } 352 template <typename Char> 353 inline bool IsAsciiLower(Char c) { 354 return c >= 'a' && c <= 'z'; 355 } 356 template <typename Char> 357 inline bool IsAsciiDigit(Char c) { 358 return c >= '0' && c <= '9'; 359 } 360 361 template <typename Char> 362 inline bool IsHexDigit(Char c) { 363 return (c >= '0' && c <= '9') || 364 (c >= 'A' && c <= 'F') || 365 (c >= 'a' && c <= 'f'); 366 } 367 368 // Returns the integer corresponding to the given hex character. For example: 369 // '4' -> 4 370 // 'a' -> 10 371 // 'B' -> 11 372 // Assumes the input is a valid hex character. DCHECKs in debug builds if not. 373 BASE_EXPORT char HexDigitToInt(wchar_t c); 374 375 // Returns true if it's a Unicode whitespace character. 376 BASE_EXPORT bool IsUnicodeWhitespace(wchar_t c); 377 378 // Return a byte string in human-readable format with a unit suffix. Not 379 // appropriate for use in any UI; use of FormatBytes and friends in ui/base is 380 // highly recommended instead. TODO(avi): Figure out how to get callers to use 381 // FormatBytes instead; remove this. 382 BASE_EXPORT string16 FormatBytesUnlocalized(int64_t bytes); 383 384 // Starting at |start_offset| (usually 0), replace the first instance of 385 // |find_this| with |replace_with|. 386 BASE_EXPORT void ReplaceFirstSubstringAfterOffset( 387 base::string16* str, 388 size_t start_offset, 389 StringPiece16 find_this, 390 StringPiece16 replace_with); 391 BASE_EXPORT void ReplaceFirstSubstringAfterOffset( 392 std::string* str, 393 size_t start_offset, 394 StringPiece find_this, 395 StringPiece replace_with); 396 397 // Starting at |start_offset| (usually 0), look through |str| and replace all 398 // instances of |find_this| with |replace_with|. 399 // 400 // This does entire substrings; use std::replace in <algorithm> for single 401 // characters, for example: 402 // std::replace(str.begin(), str.end(), 'a', 'b'); 403 BASE_EXPORT void ReplaceSubstringsAfterOffset( 404 string16* str, 405 size_t start_offset, 406 StringPiece16 find_this, 407 StringPiece16 replace_with); 408 BASE_EXPORT void ReplaceSubstringsAfterOffset( 409 std::string* str, 410 size_t start_offset, 411 StringPiece find_this, 412 StringPiece replace_with); 413 414 // Reserves enough memory in |str| to accommodate |length_with_null| characters, 415 // sets the size of |str| to |length_with_null - 1| characters, and returns a 416 // pointer to the underlying contiguous array of characters. This is typically 417 // used when calling a function that writes results into a character array, but 418 // the caller wants the data to be managed by a string-like object. It is 419 // convenient in that is can be used inline in the call, and fast in that it 420 // avoids copying the results of the call from a char* into a string. 421 // 422 // |length_with_null| must be at least 2, since otherwise the underlying string 423 // would have size 0, and trying to access &((*str)[0]) in that case can result 424 // in a number of problems. 425 // 426 // Internally, this takes linear time because the resize() call 0-fills the 427 // underlying array for potentially all 428 // (|length_with_null - 1| * sizeof(string_type::value_type)) bytes. Ideally we 429 // could avoid this aspect of the resize() call, as we expect the caller to 430 // immediately write over this memory, but there is no other way to set the size 431 // of the string, and not doing that will mean people who access |str| rather 432 // than str.c_str() will get back a string of whatever size |str| had on entry 433 // to this function (probably 0). 434 BASE_EXPORT char* WriteInto(std::string* str, size_t length_with_null); 435 BASE_EXPORT char16* WriteInto(string16* str, size_t length_with_null); 436 #ifndef OS_WIN 437 BASE_EXPORT wchar_t* WriteInto(std::wstring* str, size_t length_with_null); 438 #endif 439 440 // Does the opposite of SplitString(). 441 BASE_EXPORT std::string JoinString(const std::vector<std::string>& parts, 442 StringPiece separator); 443 BASE_EXPORT string16 JoinString(const std::vector<string16>& parts, 444 StringPiece16 separator); 445 446 // Replace $1-$2-$3..$9 in the format string with values from |subst|. 447 // Additionally, any number of consecutive '$' characters is replaced by that 448 // number less one. Eg $$->$, $$$->$$, etc. The offsets parameter here can be 449 // NULL. This only allows you to use up to nine replacements. 450 BASE_EXPORT string16 ReplaceStringPlaceholders( 451 const string16& format_string, 452 const std::vector<string16>& subst, 453 std::vector<size_t>* offsets); 454 455 BASE_EXPORT std::string ReplaceStringPlaceholders( 456 const StringPiece& format_string, 457 const std::vector<std::string>& subst, 458 std::vector<size_t>* offsets); 459 460 // Single-string shortcut for ReplaceStringHolders. |offset| may be NULL. 461 BASE_EXPORT string16 ReplaceStringPlaceholders(const string16& format_string, 462 const string16& a, 463 size_t* offset); 464 465 } // namespace base 466 467 #if defined(OS_WIN) 468 #include "base/strings/string_util_win.h" 469 #elif defined(OS_POSIX) 470 #include "base/strings/string_util_posix.h" 471 #else 472 #error Define string operations appropriately for your platform 473 #endif 474 475 #endif // BASE_STRINGS_STRING_UTIL_H_ 476