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