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 <initializer_list> 16 #include <string> 17 #include <vector> 18 19 #include "base/base_export.h" 20 #include "base/compiler_specific.h" 21 #include "base/strings/string16.h" 22 #include "base/strings/string_piece.h" // For implicit conversions. 23 #include "build/build_config.h" 24 25 namespace base { 26 27 // C standard-library functions that aren't cross-platform are provided as 28 // "base::...", and their prototypes are listed below. These functions are 29 // then implemented as inline calls to the platform-specific equivalents in the 30 // platform-specific headers. 31 32 // Wrapper for vsnprintf that always null-terminates and always returns the 33 // number of characters that would be in an untruncated formatted 34 // string, even when truncation occurs. 35 int vsnprintf(char* buffer, size_t size, const char* format, va_list arguments) 36 PRINTF_FORMAT(3, 0); 37 38 // Some of these implementations need to be inlined. 39 40 // We separate the declaration from the implementation of this inline 41 // function just so the PRINTF_FORMAT works. 42 inline int snprintf(char* buffer, 43 size_t size, 44 _Printf_format_string_ const char* format, 45 ...) PRINTF_FORMAT(3, 4); 46 inline int snprintf(char* buffer, 47 size_t size, 48 _Printf_format_string_ const char* format, 49 ...) { 50 va_list arguments; 51 va_start(arguments, format); 52 int result = vsnprintf(buffer, size, format, arguments); 53 va_end(arguments); 54 return result; 55 } 56 57 // BSD-style safe and consistent string copy functions. 58 // Copies |src| to |dst|, where |dst_size| is the total allocated size of |dst|. 59 // Copies at most |dst_size|-1 characters, and always NULL terminates |dst|, as 60 // long as |dst_size| is not 0. Returns the length of |src| in characters. 61 // If the return value is >= dst_size, then the output was truncated. 62 // NOTE: All sizes are in number of characters, NOT in bytes. 63 BASE_EXPORT size_t strlcpy(char* dst, const char* src, size_t dst_size); 64 BASE_EXPORT size_t wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size); 65 66 // Scan a wprintf format string to determine whether it's portable across a 67 // variety of systems. This function only checks that the conversion 68 // specifiers used by the format string are supported and have the same meaning 69 // on a variety of systems. It doesn't check for other errors that might occur 70 // within a format string. 71 // 72 // Nonportable conversion specifiers for wprintf are: 73 // - 's' and 'c' without an 'l' length modifier. %s and %c operate on char 74 // data on all systems except Windows, which treat them as wchar_t data. 75 // Use %ls and %lc for wchar_t data instead. 76 // - 'S' and 'C', which operate on wchar_t data on all systems except Windows, 77 // which treat them as char data. Use %ls and %lc for wchar_t data 78 // instead. 79 // - 'F', which is not identified by Windows wprintf documentation. 80 // - 'D', 'O', and 'U', which are deprecated and not available on all systems. 81 // Use %ld, %lo, and %lu instead. 82 // 83 // Note that there is no portable conversion specifier for char data when 84 // working with wprintf. 85 // 86 // This function is intended to be called from base::vswprintf. 87 BASE_EXPORT bool IsWprintfFormatPortable(const wchar_t* format); 88 89 // ASCII-specific tolower. The standard library's tolower is locale sensitive, 90 // so we don't want to use it here. 91 inline char ToLowerASCII(char c) { 92 return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c; 93 } 94 inline char16 ToLowerASCII(char16 c) { 95 return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c; 96 } 97 98 // ASCII-specific toupper. The standard library's toupper is locale sensitive, 99 // so we don't want to use it here. 100 inline char ToUpperASCII(char c) { 101 return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c; 102 } 103 inline char16 ToUpperASCII(char16 c) { 104 return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c; 105 } 106 107 // Converts the given string to it's ASCII-lowercase equivalent. 108 BASE_EXPORT std::string ToLowerASCII(StringPiece str); 109 BASE_EXPORT string16 ToLowerASCII(StringPiece16 str); 110 111 // Converts the given string to it's ASCII-uppercase equivalent. 112 BASE_EXPORT std::string ToUpperASCII(StringPiece str); 113 BASE_EXPORT string16 ToUpperASCII(StringPiece16 str); 114 115 // Functor for case-insensitive ASCII comparisons for STL algorithms like 116 // std::search. 117 // 118 // Note that a full Unicode version of this functor is not possible to write 119 // because case mappings might change the number of characters, depend on 120 // context (combining accents), and require handling UTF-16. If you need 121 // proper Unicode support, use base::i18n::ToLower/FoldCase and then just 122 // use a normal operator== on the result. 123 template<typename Char> struct CaseInsensitiveCompareASCII { 124 public: 125 bool operator()(Char x, Char y) const { 126 return ToLowerASCII(x) == ToLowerASCII(y); 127 } 128 }; 129 130 // Like strcasecmp for case-insensitive ASCII characters only. Returns: 131 // -1 (a < b) 132 // 0 (a == b) 133 // 1 (a > b) 134 // (unlike strcasecmp which can return values greater or less than 1/-1). For 135 // full Unicode support, use base::i18n::ToLower or base::i18h::FoldCase 136 // and then just call the normal string operators on the result. 137 BASE_EXPORT int CompareCaseInsensitiveASCII(StringPiece a, StringPiece b); 138 BASE_EXPORT int CompareCaseInsensitiveASCII(StringPiece16 a, StringPiece16 b); 139 140 // Equality for ASCII case-insensitive comparisons. For full Unicode support, 141 // use base::i18n::ToLower or base::i18h::FoldCase and then compare with either 142 // == or !=. 143 BASE_EXPORT bool EqualsCaseInsensitiveASCII(StringPiece a, StringPiece b); 144 BASE_EXPORT bool EqualsCaseInsensitiveASCII(StringPiece16 a, StringPiece16 b); 145 146 // These threadsafe functions return references to globally unique empty 147 // strings. 148 // 149 // It is likely faster to construct a new empty string object (just a few 150 // instructions to set the length to 0) than to get the empty string singleton 151 // returned by these functions (which requires threadsafe singleton access). 152 // 153 // Therefore, DO NOT USE THESE AS A GENERAL-PURPOSE SUBSTITUTE FOR DEFAULT 154 // CONSTRUCTORS. There is only one case where you should use these: functions 155 // which need to return a string by reference (e.g. as a class member 156 // accessor), and don't have an empty string to use (e.g. in an error case). 157 // These should not be used as initializers, function arguments, or return 158 // values for functions which return by value or outparam. 159 BASE_EXPORT const std::string& EmptyString(); 160 BASE_EXPORT const string16& EmptyString16(); 161 162 // Contains the set of characters representing whitespace in the corresponding 163 // encoding. Null-terminated. The ASCII versions are the whitespaces as defined 164 // by HTML5, and don't include control characters. 165 BASE_EXPORT extern const wchar_t kWhitespaceWide[]; // Includes Unicode. 166 BASE_EXPORT extern const char16 kWhitespaceUTF16[]; // Includes Unicode. 167 BASE_EXPORT extern const char kWhitespaceASCII[]; 168 BASE_EXPORT extern const char16 kWhitespaceASCIIAs16[]; // No unicode. 169 170 // Null-terminated string representing the UTF-8 byte order mark. 171 BASE_EXPORT extern const char kUtf8ByteOrderMark[]; 172 173 // Removes characters in |remove_chars| from anywhere in |input|. Returns true 174 // if any characters were removed. |remove_chars| must be null-terminated. 175 // NOTE: Safe to use the same variable for both |input| and |output|. 176 BASE_EXPORT bool RemoveChars(const string16& input, 177 const StringPiece16& remove_chars, 178 string16* output); 179 BASE_EXPORT bool RemoveChars(const std::string& input, 180 const StringPiece& remove_chars, 181 std::string* output); 182 183 // Replaces characters in |replace_chars| from anywhere in |input| with 184 // |replace_with|. Each character in |replace_chars| will be replaced with 185 // the |replace_with| string. Returns true if any characters were replaced. 186 // |replace_chars| must be null-terminated. 187 // NOTE: Safe to use the same variable for both |input| and |output|. 188 BASE_EXPORT bool ReplaceChars(const string16& input, 189 const StringPiece16& replace_chars, 190 const string16& replace_with, 191 string16* output); 192 BASE_EXPORT bool ReplaceChars(const std::string& input, 193 const StringPiece& replace_chars, 194 const std::string& replace_with, 195 std::string* output); 196 197 enum TrimPositions { 198 TRIM_NONE = 0, 199 TRIM_LEADING = 1 << 0, 200 TRIM_TRAILING = 1 << 1, 201 TRIM_ALL = TRIM_LEADING | TRIM_TRAILING, 202 }; 203 204 // Removes characters in |trim_chars| from the beginning and end of |input|. 205 // The 8-bit version only works on 8-bit characters, not UTF-8. 206 // 207 // It is safe to use the same variable for both |input| and |output| (this is 208 // the normal usage to trim in-place). 209 BASE_EXPORT bool TrimString(const string16& input, 210 StringPiece16 trim_chars, 211 string16* output); 212 BASE_EXPORT bool TrimString(const std::string& input, 213 StringPiece trim_chars, 214 std::string* output); 215 216 // StringPiece versions of the above. The returned pieces refer to the original 217 // buffer. 218 BASE_EXPORT StringPiece16 TrimString(StringPiece16 input, 219 const StringPiece16& trim_chars, 220 TrimPositions positions); 221 BASE_EXPORT StringPiece TrimString(StringPiece input, 222 const StringPiece& trim_chars, 223 TrimPositions positions); 224 225 // Truncates a string to the nearest UTF-8 character that will leave 226 // the string less than or equal to the specified byte size. 227 BASE_EXPORT void TruncateUTF8ToByteSize(const std::string& input, 228 const size_t byte_size, 229 std::string* output); 230 231 // Trims any whitespace from either end of the input string. 232 // 233 // The StringPiece versions return a substring referencing the input buffer. 234 // The ASCII versions look only for ASCII whitespace. 235 // 236 // The std::string versions return where whitespace was found. 237 // NOTE: Safe to use the same variable for both input and output. 238 BASE_EXPORT TrimPositions TrimWhitespace(const string16& input, 239 TrimPositions positions, 240 string16* output); 241 BASE_EXPORT StringPiece16 TrimWhitespace(StringPiece16 input, 242 TrimPositions positions); 243 BASE_EXPORT TrimPositions TrimWhitespaceASCII(const std::string& input, 244 TrimPositions positions, 245 std::string* output); 246 BASE_EXPORT StringPiece TrimWhitespaceASCII(StringPiece input, 247 TrimPositions positions); 248 249 // Searches for CR or LF characters. Removes all contiguous whitespace 250 // strings that contain them. This is useful when trying to deal with text 251 // copied from terminals. 252 // Returns |text|, with the following three transformations: 253 // (1) Leading and trailing whitespace is trimmed. 254 // (2) If |trim_sequences_with_line_breaks| is true, any other whitespace 255 // sequences containing a CR or LF are trimmed. 256 // (3) All other whitespace sequences are converted to single spaces. 257 BASE_EXPORT string16 CollapseWhitespace( 258 const string16& text, 259 bool trim_sequences_with_line_breaks); 260 BASE_EXPORT std::string CollapseWhitespaceASCII( 261 const std::string& text, 262 bool trim_sequences_with_line_breaks); 263 264 // Returns true if |input| is empty or contains only characters found in 265 // |characters|. 266 BASE_EXPORT bool ContainsOnlyChars(const StringPiece& input, 267 const StringPiece& characters); 268 BASE_EXPORT bool ContainsOnlyChars(const StringPiece16& input, 269 const StringPiece16& characters); 270 271 // Returns true if the specified string matches the criteria. How can a wide 272 // string be 8-bit or UTF8? It contains only characters that are < 256 (in the 273 // first case) or characters that use only 8-bits and whose 8-bit 274 // representation looks like a UTF-8 string (the second case). 275 // 276 // Note that IsStringUTF8 checks not only if the input is structurally 277 // valid but also if it doesn't contain any non-character codepoint 278 // (e.g. U+FFFE). It's done on purpose because all the existing callers want 279 // to have the maximum 'discriminating' power from other encodings. If 280 // there's a use case for just checking the structural validity, we have to 281 // add a new function for that. 282 // 283 // IsStringASCII assumes the input is likely all ASCII, and does not leave early 284 // if it is not the case. 285 BASE_EXPORT bool IsStringUTF8(const StringPiece& str); 286 BASE_EXPORT bool IsStringASCII(const StringPiece& str); 287 BASE_EXPORT bool IsStringASCII(const StringPiece16& str); 288 BASE_EXPORT bool IsStringASCII(const string16& str); 289 #if defined(WCHAR_T_IS_UTF32) 290 BASE_EXPORT bool IsStringASCII(const std::wstring& str); 291 #endif 292 293 // Compare the lower-case form of the given string against the given 294 // previously-lower-cased ASCII string (typically a constant). 295 BASE_EXPORT bool LowerCaseEqualsASCII(StringPiece str, 296 StringPiece lowecase_ascii); 297 BASE_EXPORT bool LowerCaseEqualsASCII(StringPiece16 str, 298 StringPiece lowecase_ascii); 299 300 // Performs a case-sensitive string compare of the given 16-bit string against 301 // the given 8-bit ASCII string (typically a constant). The behavior is 302 // undefined if the |ascii| string is not ASCII. 303 BASE_EXPORT bool EqualsASCII(StringPiece16 str, StringPiece ascii); 304 305 // Indicates case sensitivity of comparisons. Only ASCII case insensitivity 306 // is supported. Full Unicode case-insensitive conversions would need to go in 307 // base/i18n so it can use ICU. 308 // 309 // If you need to do Unicode-aware case-insensitive StartsWith/EndsWith, it's 310 // best to call base::i18n::ToLower() or base::i18n::FoldCase() (see 311 // base/i18n/case_conversion.h for usage advice) on the arguments, and then use 312 // the results to a case-sensitive comparison. 313 enum class CompareCase { 314 SENSITIVE, 315 INSENSITIVE_ASCII, 316 }; 317 318 BASE_EXPORT bool StartsWith(StringPiece str, 319 StringPiece search_for, 320 CompareCase case_sensitivity); 321 BASE_EXPORT bool StartsWith(StringPiece16 str, 322 StringPiece16 search_for, 323 CompareCase case_sensitivity); 324 BASE_EXPORT bool EndsWith(StringPiece str, 325 StringPiece search_for, 326 CompareCase case_sensitivity); 327 BASE_EXPORT bool EndsWith(StringPiece16 str, 328 StringPiece16 search_for, 329 CompareCase case_sensitivity); 330 331 // Determines the type of ASCII character, independent of locale (the C 332 // library versions will change based on locale). 333 template <typename Char> 334 inline bool IsAsciiWhitespace(Char c) { 335 return c == ' ' || c == '\r' || c == '\n' || c == '\t'; 336 } 337 template <typename Char> 338 inline bool IsAsciiAlpha(Char c) { 339 return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z'); 340 } 341 template <typename Char> 342 inline bool IsAsciiUpper(Char c) { 343 return c >= 'A' && c <= 'Z'; 344 } 345 template <typename Char> 346 inline bool IsAsciiLower(Char c) { 347 return c >= 'a' && c <= 'z'; 348 } 349 template <typename Char> 350 inline bool IsAsciiDigit(Char c) { 351 return c >= '0' && c <= '9'; 352 } 353 354 template <typename Char> 355 inline bool IsHexDigit(Char c) { 356 return (c >= '0' && c <= '9') || 357 (c >= 'A' && c <= 'F') || 358 (c >= 'a' && c <= 'f'); 359 } 360 361 // Returns the integer corresponding to the given hex character. For example: 362 // '4' -> 4 363 // 'a' -> 10 364 // 'B' -> 11 365 // Assumes the input is a valid hex character. DCHECKs in debug builds if not. 366 BASE_EXPORT char HexDigitToInt(wchar_t c); 367 368 // Returns true if it's a Unicode whitespace character. 369 BASE_EXPORT bool IsUnicodeWhitespace(wchar_t c); 370 371 // Return a byte string in human-readable format with a unit suffix. Not 372 // appropriate for use in any UI; use of FormatBytes and friends in ui/base is 373 // highly recommended instead. TODO(avi): Figure out how to get callers to use 374 // FormatBytes instead; remove this. 375 BASE_EXPORT string16 FormatBytesUnlocalized(int64_t bytes); 376 377 // Starting at |start_offset| (usually 0), replace the first instance of 378 // |find_this| with |replace_with|. 379 BASE_EXPORT void ReplaceFirstSubstringAfterOffset( 380 base::string16* str, 381 size_t start_offset, 382 StringPiece16 find_this, 383 StringPiece16 replace_with); 384 BASE_EXPORT void ReplaceFirstSubstringAfterOffset( 385 std::string* str, 386 size_t start_offset, 387 StringPiece find_this, 388 StringPiece replace_with); 389 390 // Starting at |start_offset| (usually 0), look through |str| and replace all 391 // instances of |find_this| with |replace_with|. 392 // 393 // This does entire substrings; use std::replace in <algorithm> for single 394 // characters, for example: 395 // std::replace(str.begin(), str.end(), 'a', 'b'); 396 BASE_EXPORT void ReplaceSubstringsAfterOffset( 397 string16* str, 398 size_t start_offset, 399 StringPiece16 find_this, 400 StringPiece16 replace_with); 401 BASE_EXPORT void ReplaceSubstringsAfterOffset( 402 std::string* str, 403 size_t start_offset, 404 StringPiece find_this, 405 StringPiece replace_with); 406 407 // Reserves enough memory in |str| to accommodate |length_with_null| characters, 408 // sets the size of |str| to |length_with_null - 1| characters, and returns a 409 // pointer to the underlying contiguous array of characters. This is typically 410 // used when calling a function that writes results into a character array, but 411 // the caller wants the data to be managed by a string-like object. It is 412 // convenient in that is can be used inline in the call, and fast in that it 413 // avoids copying the results of the call from a char* into a string. 414 // 415 // |length_with_null| must be at least 2, since otherwise the underlying string 416 // would have size 0, and trying to access &((*str)[0]) in that case can result 417 // in a number of problems. 418 // 419 // Internally, this takes linear time because the resize() call 0-fills the 420 // underlying array for potentially all 421 // (|length_with_null - 1| * sizeof(string_type::value_type)) bytes. Ideally we 422 // could avoid this aspect of the resize() call, as we expect the caller to 423 // immediately write over this memory, but there is no other way to set the size 424 // of the string, and not doing that will mean people who access |str| rather 425 // than str.c_str() will get back a string of whatever size |str| had on entry 426 // to this function (probably 0). 427 BASE_EXPORT char* WriteInto(std::string* str, size_t length_with_null); 428 BASE_EXPORT char16* WriteInto(string16* str, size_t length_with_null); 429 #ifndef OS_WIN 430 BASE_EXPORT wchar_t* WriteInto(std::wstring* str, size_t length_with_null); 431 #endif 432 433 // Does the opposite of SplitString()/SplitStringPiece(). Joins a vector or list 434 // of strings into a single string, inserting |separator| (which may be empty) 435 // in between all elements. 436 // 437 // If possible, callers should build a vector of StringPieces and use the 438 // StringPiece variant, so that they do not create unnecessary copies of 439 // strings. For example, instead of using SplitString, modifying the vector, 440 // then using JoinString, use SplitStringPiece followed by JoinString so that no 441 // copies of those strings are created until the final join operation. 442 BASE_EXPORT std::string JoinString(const std::vector<std::string>& parts, 443 StringPiece separator); 444 BASE_EXPORT string16 JoinString(const std::vector<string16>& parts, 445 StringPiece16 separator); 446 BASE_EXPORT std::string JoinString(const std::vector<StringPiece>& parts, 447 StringPiece separator); 448 BASE_EXPORT string16 JoinString(const std::vector<StringPiece16>& parts, 449 StringPiece16 separator); 450 // Explicit initializer_list overloads are required to break ambiguity when used 451 // with a literal initializer list (otherwise the compiler would not be able to 452 // decide between the string and StringPiece overloads). 453 BASE_EXPORT std::string JoinString(std::initializer_list<StringPiece> parts, 454 StringPiece separator); 455 BASE_EXPORT string16 JoinString(std::initializer_list<StringPiece16> parts, 456 StringPiece16 separator); 457 458 // Replace $1-$2-$3..$9 in the format string with values from |subst|. 459 // Additionally, any number of consecutive '$' characters is replaced by that 460 // number less one. Eg $$->$, $$$->$$, etc. The offsets parameter here can be 461 // NULL. This only allows you to use up to nine replacements. 462 BASE_EXPORT string16 ReplaceStringPlaceholders( 463 const string16& format_string, 464 const std::vector<string16>& subst, 465 std::vector<size_t>* offsets); 466 467 BASE_EXPORT std::string ReplaceStringPlaceholders( 468 const StringPiece& format_string, 469 const std::vector<std::string>& subst, 470 std::vector<size_t>* offsets); 471 472 // Single-string shortcut for ReplaceStringHolders. |offset| may be NULL. 473 BASE_EXPORT string16 ReplaceStringPlaceholders(const string16& format_string, 474 const string16& a, 475 size_t* offset); 476 477 } // namespace base 478 479 #if defined(OS_WIN) 480 #include "base/strings/string_util_win.h" 481 #elif defined(OS_POSIX) 482 #include "base/strings/string_util_posix.h" 483 #else 484 #error Define string operations appropriately for your platform 485 #endif 486 487 #endif // BASE_STRINGS_STRING_UTIL_H_ 488