michael@0: // Copyright 2013 The Chromium Authors. All rights reserved.
michael@0: // Use of this source code is governed by a BSD-style license that can be
michael@0: // found in the LICENSE file.
michael@0: //
michael@0: // This file defines utility functions for working with strings.
michael@0: 
michael@0: #ifndef BASE_STRINGS_STRING_UTIL_H_
michael@0: #define BASE_STRINGS_STRING_UTIL_H_
michael@0: 
michael@0: #include <ctype.h>
michael@0: #include <stdarg.h>   // va_list
michael@0: 
michael@0: #include <string>
michael@0: #include <vector>
michael@0: 
michael@0: #include "base/base_export.h"
michael@0: #include "base/basictypes.h"
michael@0: #include "base/compiler_specific.h"
michael@0: #include "base/strings/string16.h"
michael@0: #include "base/strings/string_piece.h"  // For implicit conversions.
michael@0: 
michael@0: // Safe standard library wrappers for all platforms.
michael@0: 
michael@0: namespace base {
michael@0: 
michael@0: // C standard-library functions like "strncasecmp" and "snprintf" that aren't
michael@0: // cross-platform are provided as "base::strncasecmp", and their prototypes
michael@0: // are listed below.  These functions are then implemented as inline calls
michael@0: // to the platform-specific equivalents in the platform-specific headers.
michael@0: 
michael@0: // Compares the two strings s1 and s2 without regard to case using
michael@0: // the current locale; returns 0 if they are equal, 1 if s1 > s2, and -1 if
michael@0: // s2 > s1 according to a lexicographic comparison.
michael@0: int strcasecmp(const char* s1, const char* s2);
michael@0: 
michael@0: // Compares up to count characters of s1 and s2 without regard to case using
michael@0: // the current locale; returns 0 if they are equal, 1 if s1 > s2, and -1 if
michael@0: // s2 > s1 according to a lexicographic comparison.
michael@0: int strncasecmp(const char* s1, const char* s2, size_t count);
michael@0: 
michael@0: // Same as strncmp but for char16 strings.
michael@0: int strncmp16(const char16* s1, const char16* s2, size_t count);
michael@0: 
michael@0: // Wrapper for vsnprintf that always null-terminates and always returns the
michael@0: // number of characters that would be in an untruncated formatted
michael@0: // string, even when truncation occurs.
michael@0: int vsnprintf(char* buffer, size_t size, const char* format, va_list arguments)
michael@0:     PRINTF_FORMAT(3, 0);
michael@0: 
michael@0: // vswprintf always null-terminates, but when truncation occurs, it will either
michael@0: // return -1 or the number of characters that would be in an untruncated
michael@0: // formatted string.  The actual return value depends on the underlying
michael@0: // C library's vswprintf implementation.
michael@0: int vswprintf(wchar_t* buffer, size_t size,
michael@0:               const wchar_t* format, va_list arguments)
michael@0:     WPRINTF_FORMAT(3, 0);
michael@0: 
michael@0: // Some of these implementations need to be inlined.
michael@0: 
michael@0: // We separate the declaration from the implementation of this inline
michael@0: // function just so the PRINTF_FORMAT works.
michael@0: inline int snprintf(char* buffer, size_t size, const char* format, ...)
michael@0:     PRINTF_FORMAT(3, 4);
michael@0: inline int snprintf(char* buffer, size_t size, const char* format, ...) {
michael@0:   va_list arguments;
michael@0:   va_start(arguments, format);
michael@0:   int result = vsnprintf(buffer, size, format, arguments);
michael@0:   va_end(arguments);
michael@0:   return result;
michael@0: }
michael@0: 
michael@0: // We separate the declaration from the implementation of this inline
michael@0: // function just so the WPRINTF_FORMAT works.
michael@0: inline int swprintf(wchar_t* buffer, size_t size, const wchar_t* format, ...)
michael@0:     WPRINTF_FORMAT(3, 4);
michael@0: inline int swprintf(wchar_t* buffer, size_t size, const wchar_t* format, ...) {
michael@0:   va_list arguments;
michael@0:   va_start(arguments, format);
michael@0:   int result = vswprintf(buffer, size, format, arguments);
michael@0:   va_end(arguments);
michael@0:   return result;
michael@0: }
michael@0: 
michael@0: // BSD-style safe and consistent string copy functions.
michael@0: // Copies |src| to |dst|, where |dst_size| is the total allocated size of |dst|.
michael@0: // Copies at most |dst_size|-1 characters, and always NULL terminates |dst|, as
michael@0: // long as |dst_size| is not 0.  Returns the length of |src| in characters.
michael@0: // If the return value is >= dst_size, then the output was truncated.
michael@0: // NOTE: All sizes are in number of characters, NOT in bytes.
michael@0: BASE_EXPORT size_t strlcpy(char* dst, const char* src, size_t dst_size);
michael@0: BASE_EXPORT size_t wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size);
michael@0: 
michael@0: // Scan a wprintf format string to determine whether it's portable across a
michael@0: // variety of systems.  This function only checks that the conversion
michael@0: // specifiers used by the format string are supported and have the same meaning
michael@0: // on a variety of systems.  It doesn't check for other errors that might occur
michael@0: // within a format string.
michael@0: //
michael@0: // Nonportable conversion specifiers for wprintf are:
michael@0: //  - 's' and 'c' without an 'l' length modifier.  %s and %c operate on char
michael@0: //     data on all systems except Windows, which treat them as wchar_t data.
michael@0: //     Use %ls and %lc for wchar_t data instead.
michael@0: //  - 'S' and 'C', which operate on wchar_t data on all systems except Windows,
michael@0: //     which treat them as char data.  Use %ls and %lc for wchar_t data
michael@0: //     instead.
michael@0: //  - 'F', which is not identified by Windows wprintf documentation.
michael@0: //  - 'D', 'O', and 'U', which are deprecated and not available on all systems.
michael@0: //     Use %ld, %lo, and %lu instead.
michael@0: //
michael@0: // Note that there is no portable conversion specifier for char data when
michael@0: // working with wprintf.
michael@0: //
michael@0: // This function is intended to be called from base::vswprintf.
michael@0: BASE_EXPORT bool IsWprintfFormatPortable(const wchar_t* format);
michael@0: 
michael@0: // ASCII-specific tolower.  The standard library's tolower is locale sensitive,
michael@0: // so we don't want to use it here.
michael@0: template <class Char> inline Char ToLowerASCII(Char c) {
michael@0:   return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c;
michael@0: }
michael@0: 
michael@0: // ASCII-specific toupper.  The standard library's toupper is locale sensitive,
michael@0: // so we don't want to use it here.
michael@0: template <class Char> inline Char ToUpperASCII(Char c) {
michael@0:   return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c;
michael@0: }
michael@0: 
michael@0: // Function objects to aid in comparing/searching strings.
michael@0: 
michael@0: template<typename Char> struct CaseInsensitiveCompare {
michael@0:  public:
michael@0:   bool operator()(Char x, Char y) const {
michael@0:     // TODO(darin): Do we really want to do locale sensitive comparisons here?
michael@0:     // See http://crbug.com/24917
michael@0:     return tolower(x) == tolower(y);
michael@0:   }
michael@0: };
michael@0: 
michael@0: template<typename Char> struct CaseInsensitiveCompareASCII {
michael@0:  public:
michael@0:   bool operator()(Char x, Char y) const {
michael@0:     return ToLowerASCII(x) == ToLowerASCII(y);
michael@0:   }
michael@0: };
michael@0: 
michael@0: }  // namespace base
michael@0: 
michael@0: #if defined(OS_WIN)
michael@0: #include "base/strings/string_util_win.h"
michael@0: #elif defined(OS_POSIX)
michael@0: #include "base/strings/string_util_posix.h"
michael@0: #else
michael@0: #error Define string operations appropriately for your platform
michael@0: #endif
michael@0: 
michael@0: // These threadsafe functions return references to globally unique empty
michael@0: // strings.
michael@0: //
michael@0: // DO NOT USE THESE AS A GENERAL-PURPOSE SUBSTITUTE FOR DEFAULT CONSTRUCTORS.
michael@0: // There is only one case where you should use these: functions which need to
michael@0: // return a string by reference (e.g. as a class member accessor), and don't
michael@0: // have an empty string to use (e.g. in an error case).  These should not be
michael@0: // used as initializers, function arguments, or return values for functions
michael@0: // which return by value or outparam.
michael@0: BASE_EXPORT const std::string& EmptyString();
michael@0: BASE_EXPORT const std::wstring& EmptyWString();
michael@0: BASE_EXPORT const string16& EmptyString16();
michael@0: 
michael@0: BASE_EXPORT extern const wchar_t kWhitespaceWide[];
michael@0: BASE_EXPORT extern const char16 kWhitespaceUTF16[];
michael@0: BASE_EXPORT extern const char kWhitespaceASCII[];
michael@0: 
michael@0: BASE_EXPORT extern const char kUtf8ByteOrderMark[];
michael@0: 
michael@0: // Removes characters in |remove_chars| from anywhere in |input|.  Returns true
michael@0: // if any characters were removed.  |remove_chars| must be null-terminated.
michael@0: // NOTE: Safe to use the same variable for both |input| and |output|.
michael@0: BASE_EXPORT bool RemoveChars(const string16& input,
michael@0:                              const char16 remove_chars[],
michael@0:                              string16* output);
michael@0: BASE_EXPORT bool RemoveChars(const std::string& input,
michael@0:                              const char remove_chars[],
michael@0:                              std::string* output);
michael@0: 
michael@0: // Replaces characters in |replace_chars| from anywhere in |input| with
michael@0: // |replace_with|.  Each character in |replace_chars| will be replaced with
michael@0: // the |replace_with| string.  Returns true if any characters were replaced.
michael@0: // |replace_chars| must be null-terminated.
michael@0: // NOTE: Safe to use the same variable for both |input| and |output|.
michael@0: BASE_EXPORT bool ReplaceChars(const string16& input,
michael@0:                               const char16 replace_chars[],
michael@0:                               const string16& replace_with,
michael@0:                               string16* output);
michael@0: BASE_EXPORT bool ReplaceChars(const std::string& input,
michael@0:                               const char replace_chars[],
michael@0:                               const std::string& replace_with,
michael@0:                               std::string* output);
michael@0: 
michael@0: // Removes characters in |trim_chars| from the beginning and end of |input|.
michael@0: // |trim_chars| must be null-terminated.
michael@0: // NOTE: Safe to use the same variable for both |input| and |output|.
michael@0: BASE_EXPORT bool TrimString(const std::wstring& input,
michael@0:                             const wchar_t trim_chars[],
michael@0:                             std::wstring* output);
michael@0: BASE_EXPORT bool TrimString(const string16& input,
michael@0:                             const char16 trim_chars[],
michael@0:                             string16* output);
michael@0: BASE_EXPORT bool TrimString(const std::string& input,
michael@0:                             const char trim_chars[],
michael@0:                             std::string* output);
michael@0: 
michael@0: // Truncates a string to the nearest UTF-8 character that will leave
michael@0: // the string less than or equal to the specified byte size.
michael@0: BASE_EXPORT void TruncateUTF8ToByteSize(const std::string& input,
michael@0:                                         const size_t byte_size,
michael@0:                                         std::string* output);
michael@0: 
michael@0: // Trims any whitespace from either end of the input string.  Returns where
michael@0: // whitespace was found.
michael@0: // The non-wide version has two functions:
michael@0: // * TrimWhitespaceASCII()
michael@0: //   This function is for ASCII strings and only looks for ASCII whitespace;
michael@0: // Please choose the best one according to your usage.
michael@0: // NOTE: Safe to use the same variable for both input and output.
michael@0: enum TrimPositions {
michael@0:   TRIM_NONE     = 0,
michael@0:   TRIM_LEADING  = 1 << 0,
michael@0:   TRIM_TRAILING = 1 << 1,
michael@0:   TRIM_ALL      = TRIM_LEADING | TRIM_TRAILING,
michael@0: };
michael@0: BASE_EXPORT TrimPositions TrimWhitespace(const string16& input,
michael@0:                                          TrimPositions positions,
michael@0:                                          string16* output);
michael@0: BASE_EXPORT TrimPositions TrimWhitespaceASCII(const std::string& input,
michael@0:                                               TrimPositions positions,
michael@0:                                               std::string* output);
michael@0: 
michael@0: // Deprecated. This function is only for backward compatibility and calls
michael@0: // TrimWhitespaceASCII().
michael@0: BASE_EXPORT TrimPositions TrimWhitespace(const std::string& input,
michael@0:                                          TrimPositions positions,
michael@0:                                          std::string* output);
michael@0: 
michael@0: // Searches  for CR or LF characters.  Removes all contiguous whitespace
michael@0: // strings that contain them.  This is useful when trying to deal with text
michael@0: // copied from terminals.
michael@0: // Returns |text|, with the following three transformations:
michael@0: // (1) Leading and trailing whitespace is trimmed.
michael@0: // (2) If |trim_sequences_with_line_breaks| is true, any other whitespace
michael@0: //     sequences containing a CR or LF are trimmed.
michael@0: // (3) All other whitespace sequences are converted to single spaces.
michael@0: BASE_EXPORT std::wstring CollapseWhitespace(
michael@0:     const std::wstring& text,
michael@0:     bool trim_sequences_with_line_breaks);
michael@0: BASE_EXPORT string16 CollapseWhitespace(
michael@0:     const string16& text,
michael@0:     bool trim_sequences_with_line_breaks);
michael@0: BASE_EXPORT std::string CollapseWhitespaceASCII(
michael@0:     const std::string& text,
michael@0:     bool trim_sequences_with_line_breaks);
michael@0: 
michael@0: // Returns true if the passed string is empty or contains only white-space
michael@0: // characters.
michael@0: BASE_EXPORT bool ContainsOnlyWhitespaceASCII(const std::string& str);
michael@0: BASE_EXPORT bool ContainsOnlyWhitespace(const string16& str);
michael@0: 
michael@0: // Returns true if |input| is empty or contains only characters found in
michael@0: // |characters|.
michael@0: BASE_EXPORT bool ContainsOnlyChars(const std::wstring& input,
michael@0:                                    const std::wstring& characters);
michael@0: BASE_EXPORT bool ContainsOnlyChars(const string16& input,
michael@0:                                    const string16& characters);
michael@0: BASE_EXPORT bool ContainsOnlyChars(const std::string& input,
michael@0:                                    const std::string& characters);
michael@0: 
michael@0: // Converts to 7-bit ASCII by truncating. The result must be known to be ASCII
michael@0: // beforehand.
michael@0: BASE_EXPORT std::string WideToASCII(const std::wstring& wide);
michael@0: BASE_EXPORT std::string UTF16ToASCII(const string16& utf16);
michael@0: 
michael@0: // Converts the given wide string to the corresponding Latin1. This will fail
michael@0: // (return false) if any characters are more than 255.
michael@0: BASE_EXPORT bool WideToLatin1(const std::wstring& wide, std::string* latin1);
michael@0: 
michael@0: // Returns true if the specified string matches the criteria. How can a wide
michael@0: // string be 8-bit or UTF8? It contains only characters that are < 256 (in the
michael@0: // first case) or characters that use only 8-bits and whose 8-bit
michael@0: // representation looks like a UTF-8 string (the second case).
michael@0: //
michael@0: // Note that IsStringUTF8 checks not only if the input is structurally
michael@0: // valid but also if it doesn't contain any non-character codepoint
michael@0: // (e.g. U+FFFE). It's done on purpose because all the existing callers want
michael@0: // to have the maximum 'discriminating' power from other encodings. If
michael@0: // there's a use case for just checking the structural validity, we have to
michael@0: // add a new function for that.
michael@0: BASE_EXPORT bool IsStringUTF8(const std::string& str);
michael@0: BASE_EXPORT bool IsStringASCII(const std::wstring& str);
michael@0: BASE_EXPORT bool IsStringASCII(const string16& str);
michael@0: 
michael@0: // Converts the elements of the given string.  This version uses a pointer to
michael@0: // clearly differentiate it from the non-pointer variant.
michael@0: template <class str> inline void StringToLowerASCII(str* s) {
michael@0:   for (typename str::iterator i = s->begin(); i != s->end(); ++i)
michael@0:     *i = base::ToLowerASCII(*i);
michael@0: }
michael@0: 
michael@0: template <class str> inline str StringToLowerASCII(const str& s) {
michael@0:   // for std::string and std::wstring
michael@0:   str output(s);
michael@0:   StringToLowerASCII(&output);
michael@0:   return output;
michael@0: }
michael@0: 
michael@0: // Converts the elements of the given string.  This version uses a pointer to
michael@0: // clearly differentiate it from the non-pointer variant.
michael@0: template <class str> inline void StringToUpperASCII(str* s) {
michael@0:   for (typename str::iterator i = s->begin(); i != s->end(); ++i)
michael@0:     *i = base::ToUpperASCII(*i);
michael@0: }
michael@0: 
michael@0: template <class str> inline str StringToUpperASCII(const str& s) {
michael@0:   // for std::string and std::wstring
michael@0:   str output(s);
michael@0:   StringToUpperASCII(&output);
michael@0:   return output;
michael@0: }
michael@0: 
michael@0: // Compare the lower-case form of the given string against the given ASCII
michael@0: // string.  This is useful for doing checking if an input string matches some
michael@0: // token, and it is optimized to avoid intermediate string copies.  This API is
michael@0: // borrowed from the equivalent APIs in Mozilla.
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const std::string& a, const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const std::wstring& a, const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const string16& a, const char* b);
michael@0: 
michael@0: // Same thing, but with string iterators instead.
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(std::string::const_iterator a_begin,
michael@0:                                       std::string::const_iterator a_end,
michael@0:                                       const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(std::wstring::const_iterator a_begin,
michael@0:                                       std::wstring::const_iterator a_end,
michael@0:                                       const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(string16::const_iterator a_begin,
michael@0:                                       string16::const_iterator a_end,
michael@0:                                       const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const char* a_begin,
michael@0:                                       const char* a_end,
michael@0:                                       const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const wchar_t* a_begin,
michael@0:                                       const wchar_t* a_end,
michael@0:                                       const char* b);
michael@0: BASE_EXPORT bool LowerCaseEqualsASCII(const char16* a_begin,
michael@0:                                       const char16* a_end,
michael@0:                                       const char* b);
michael@0: 
michael@0: // Performs a case-sensitive string compare. The behavior is undefined if both
michael@0: // strings are not ASCII.
michael@0: BASE_EXPORT bool EqualsASCII(const string16& a, const base::StringPiece& b);
michael@0: 
michael@0: // Returns true if str starts with search, or false otherwise.
michael@0: BASE_EXPORT bool StartsWithASCII(const std::string& str,
michael@0:                                  const std::string& search,
michael@0:                                  bool case_sensitive);
michael@0: BASE_EXPORT bool StartsWith(const std::wstring& str,
michael@0:                             const std::wstring& search,
michael@0:                             bool case_sensitive);
michael@0: BASE_EXPORT bool StartsWith(const string16& str,
michael@0:                             const string16& search,
michael@0:                             bool case_sensitive);
michael@0: 
michael@0: // Returns true if str ends with search, or false otherwise.
michael@0: BASE_EXPORT bool EndsWith(const std::string& str,
michael@0:                           const std::string& search,
michael@0:                           bool case_sensitive);
michael@0: BASE_EXPORT bool EndsWith(const std::wstring& str,
michael@0:                           const std::wstring& search,
michael@0:                           bool case_sensitive);
michael@0: BASE_EXPORT bool EndsWith(const string16& str,
michael@0:                           const string16& search,
michael@0:                           bool case_sensitive);
michael@0: 
michael@0: 
michael@0: // Determines the type of ASCII character, independent of locale (the C
michael@0: // library versions will change based on locale).
michael@0: template <typename Char>
michael@0: inline bool IsAsciiWhitespace(Char c) {
michael@0:   return c == ' ' || c == '\r' || c == '\n' || c == '\t';
michael@0: }
michael@0: template <typename Char>
michael@0: inline bool IsAsciiAlpha(Char c) {
michael@0:   return ((c >= 'A') && (c <= 'Z')) || ((c >= 'a') && (c <= 'z'));
michael@0: }
michael@0: template <typename Char>
michael@0: inline bool IsAsciiDigit(Char c) {
michael@0:   return c >= '0' && c <= '9';
michael@0: }
michael@0: 
michael@0: template <typename Char>
michael@0: inline bool IsHexDigit(Char c) {
michael@0:   return (c >= '0' && c <= '9') ||
michael@0:          (c >= 'A' && c <= 'F') ||
michael@0:          (c >= 'a' && c <= 'f');
michael@0: }
michael@0: 
michael@0: template <typename Char>
michael@0: inline Char HexDigitToInt(Char c) {
michael@0:   DCHECK(IsHexDigit(c));
michael@0:   if (c >= '0' && c <= '9')
michael@0:     return c - '0';
michael@0:   if (c >= 'A' && c <= 'F')
michael@0:     return c - 'A' + 10;
michael@0:   if (c >= 'a' && c <= 'f')
michael@0:     return c - 'a' + 10;
michael@0:   return 0;
michael@0: }
michael@0: 
michael@0: // Returns true if it's a whitespace character.
michael@0: inline bool IsWhitespace(wchar_t c) {
michael@0:   return wcschr(kWhitespaceWide, c) != NULL;
michael@0: }
michael@0: 
michael@0: // Return a byte string in human-readable format with a unit suffix. Not
michael@0: // appropriate for use in any UI; use of FormatBytes and friends in ui/base is
michael@0: // highly recommended instead. TODO(avi): Figure out how to get callers to use
michael@0: // FormatBytes instead; remove this.
michael@0: BASE_EXPORT string16 FormatBytesUnlocalized(int64 bytes);
michael@0: 
michael@0: // Starting at |start_offset| (usually 0), replace the first instance of
michael@0: // |find_this| with |replace_with|.
michael@0: BASE_EXPORT void ReplaceFirstSubstringAfterOffset(
michael@0:     string16* str,
michael@0:     string16::size_type start_offset,
michael@0:     const string16& find_this,
michael@0:     const string16& replace_with);
michael@0: BASE_EXPORT void ReplaceFirstSubstringAfterOffset(
michael@0:     std::string* str,
michael@0:     std::string::size_type start_offset,
michael@0:     const std::string& find_this,
michael@0:     const std::string& replace_with);
michael@0: 
michael@0: // Starting at |start_offset| (usually 0), look through |str| and replace all
michael@0: // instances of |find_this| with |replace_with|.
michael@0: //
michael@0: // This does entire substrings; use std::replace in <algorithm> for single
michael@0: // characters, for example:
michael@0: //   std::replace(str.begin(), str.end(), 'a', 'b');
michael@0: BASE_EXPORT void ReplaceSubstringsAfterOffset(
michael@0:     string16* str,
michael@0:     string16::size_type start_offset,
michael@0:     const string16& find_this,
michael@0:     const string16& replace_with);
michael@0: BASE_EXPORT void ReplaceSubstringsAfterOffset(
michael@0:     std::string* str,
michael@0:     std::string::size_type start_offset,
michael@0:     const std::string& find_this,
michael@0:     const std::string& replace_with);
michael@0: 
michael@0: // Reserves enough memory in |str| to accommodate |length_with_null| characters,
michael@0: // sets the size of |str| to |length_with_null - 1| characters, and returns a
michael@0: // pointer to the underlying contiguous array of characters.  This is typically
michael@0: // used when calling a function that writes results into a character array, but
michael@0: // the caller wants the data to be managed by a string-like object.  It is
michael@0: // convenient in that is can be used inline in the call, and fast in that it
michael@0: // avoids copying the results of the call from a char* into a string.
michael@0: //
michael@0: // |length_with_null| must be at least 2, since otherwise the underlying string
michael@0: // would have size 0, and trying to access &((*str)[0]) in that case can result
michael@0: // in a number of problems.
michael@0: //
michael@0: // Internally, this takes linear time because the resize() call 0-fills the
michael@0: // underlying array for potentially all
michael@0: // (|length_with_null - 1| * sizeof(string_type::value_type)) bytes.  Ideally we
michael@0: // could avoid this aspect of the resize() call, as we expect the caller to
michael@0: // immediately write over this memory, but there is no other way to set the size
michael@0: // of the string, and not doing that will mean people who access |str| rather
michael@0: // than str.c_str() will get back a string of whatever size |str| had on entry
michael@0: // to this function (probably 0).
michael@0: template <class string_type>
michael@0: inline typename string_type::value_type* WriteInto(string_type* str,
michael@0:                                                    size_t length_with_null) {
michael@0:   DCHECK_GT(length_with_null, 1u);
michael@0:   str->reserve(length_with_null);
michael@0:   str->resize(length_with_null - 1);
michael@0:   return &((*str)[0]);
michael@0: }
michael@0: 
michael@0: //-----------------------------------------------------------------------------
michael@0: 
michael@0: // Splits a string into its fields delimited by any of the characters in
michael@0: // |delimiters|.  Each field is added to the |tokens| vector.  Returns the
michael@0: // number of tokens found.
michael@0: BASE_EXPORT size_t Tokenize(const std::wstring& str,
michael@0:                             const std::wstring& delimiters,
michael@0:                             std::vector<std::wstring>* tokens);
michael@0: BASE_EXPORT size_t Tokenize(const string16& str,
michael@0:                             const string16& delimiters,
michael@0:                             std::vector<string16>* tokens);
michael@0: BASE_EXPORT size_t Tokenize(const std::string& str,
michael@0:                             const std::string& delimiters,
michael@0:                             std::vector<std::string>* tokens);
michael@0: BASE_EXPORT size_t Tokenize(const base::StringPiece& str,
michael@0:                             const base::StringPiece& delimiters,
michael@0:                             std::vector<base::StringPiece>* tokens);
michael@0: 
michael@0: // Does the opposite of SplitString().
michael@0: BASE_EXPORT string16 JoinString(const std::vector<string16>& parts, char16 s);
michael@0: BASE_EXPORT std::string JoinString(
michael@0:     const std::vector<std::string>& parts, char s);
michael@0: 
michael@0: // Join |parts| using |separator|.
michael@0: BASE_EXPORT std::string JoinString(
michael@0:     const std::vector<std::string>& parts,
michael@0:     const std::string& separator);
michael@0: BASE_EXPORT string16 JoinString(
michael@0:     const std::vector<string16>& parts,
michael@0:     const string16& separator);
michael@0: 
michael@0: // Replace $1-$2-$3..$9 in the format string with |a|-|b|-|c|..|i| respectively.
michael@0: // Additionally, any number of consecutive '$' characters is replaced by that
michael@0: // number less one. Eg $$->$, $$$->$$, etc. The offsets parameter here can be
michael@0: // NULL. This only allows you to use up to nine replacements.
michael@0: BASE_EXPORT string16 ReplaceStringPlaceholders(
michael@0:     const string16& format_string,
michael@0:     const std::vector<string16>& subst,
michael@0:     std::vector<size_t>* offsets);
michael@0: 
michael@0: BASE_EXPORT std::string ReplaceStringPlaceholders(
michael@0:     const base::StringPiece& format_string,
michael@0:     const std::vector<std::string>& subst,
michael@0:     std::vector<size_t>* offsets);
michael@0: 
michael@0: // Single-string shortcut for ReplaceStringHolders. |offset| may be NULL.
michael@0: BASE_EXPORT string16 ReplaceStringPlaceholders(const string16& format_string,
michael@0:                                                const string16& a,
michael@0:                                                size_t* offset);
michael@0: 
michael@0: // Returns true if the string passed in matches the pattern. The pattern
michael@0: // string can contain wildcards like * and ?
michael@0: // The backslash character (\) is an escape character for * and ?
michael@0: // We limit the patterns to having a max of 16 * or ? characters.
michael@0: // ? matches 0 or 1 character, while * matches 0 or more characters.
michael@0: BASE_EXPORT bool MatchPattern(const base::StringPiece& string,
michael@0:                               const base::StringPiece& pattern);
michael@0: BASE_EXPORT bool MatchPattern(const string16& string, const string16& pattern);
michael@0: 
michael@0: // Hack to convert any char-like type to its unsigned counterpart.
michael@0: // For example, it will convert char, signed char and unsigned char to unsigned
michael@0: // char.
michael@0: template<typename T>
michael@0: struct ToUnsigned {
michael@0:   typedef T Unsigned;
michael@0: };
michael@0: 
michael@0: template<>
michael@0: struct ToUnsigned<char> {
michael@0:   typedef unsigned char Unsigned;
michael@0: };
michael@0: template<>
michael@0: struct ToUnsigned<signed char> {
michael@0:   typedef unsigned char Unsigned;
michael@0: };
michael@0: template<>
michael@0: struct ToUnsigned<wchar_t> {
michael@0: #if defined(WCHAR_T_IS_UTF16)
michael@0:   typedef unsigned short Unsigned;
michael@0: #elif defined(WCHAR_T_IS_UTF32)
michael@0:   typedef uint32 Unsigned;
michael@0: #endif
michael@0: };
michael@0: template<>
michael@0: struct ToUnsigned<short> {
michael@0:   typedef unsigned short Unsigned;
michael@0: };
michael@0: 
michael@0: #endif  // BASE_STRINGS_STRING_UTIL_H_