michael@0: // Copyright (c) 2006-2008 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_STRING_UTIL_H_
michael@0: #define BASE_STRING_UTIL_H_
michael@0: 
michael@0: #include <stdarg.h>   // va_list
michael@0: #include <ctype.h>
michael@0: 
michael@0: #include <string>
michael@0: #include <vector>
michael@0: 
michael@0: #include "base/basictypes.h"
michael@0: #include "base/string16.h"
michael@0: #include "base/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: // Compare 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: // Compare 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: // 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: 
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: 
michael@0: // Some of these implementations need to be inlined.
michael@0: 
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: 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: size_t strlcpy(char* dst, const char* src, size_t dst_size);
michael@0: 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: bool IsWprintfFormatPortable(const wchar_t* format);
michael@0: 
michael@0: }  // namespace base
michael@0: 
michael@0: #if defined(OS_WIN)
michael@0: #include "base/string_util_win.h"
michael@0: #elif defined(OS_POSIX)
michael@0: #include "base/string_util_posix.h"
michael@0: #else
michael@0: #error Define string operations appropriately for your platform
michael@0: #endif
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: // * TrimWhitespaceUTF8()
michael@0: //   This function is for UTF-8 strings and looks for Unicode 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: TrimPositions TrimWhitespace(const std::wstring& input,
michael@0:                              TrimPositions positions,
michael@0:                              std::wstring* output);
michael@0: 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: 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: std::wstring CollapseWhitespace(const std::wstring& text,
michael@0:                                 bool trim_sequences_with_line_breaks);
michael@0: 
michael@0: // These convert between ASCII (7-bit) and Wide/UTF16 strings.
michael@0: std::string WideToASCII(const std::wstring& wide);
michael@0: std::wstring ASCIIToWide(const std::string& ascii);
michael@0: std::string UTF16ToASCII(const string16& utf16);
michael@0: string16 ASCIIToUTF16(const std::string& ascii);
michael@0: 
michael@0: // These convert between UTF-8, -16, and -32 strings. They are potentially slow,
michael@0: // so avoid unnecessary conversions. The low-level versions return a boolean
michael@0: // indicating whether the conversion was 100% valid. In this case, it will still
michael@0: // do the best it can and put the result in the output buffer. The versions that
michael@0: // return strings ignore this error and just return the best conversion
michael@0: // possible.
michael@0: bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output);
michael@0: std::string WideToUTF8(const std::wstring& wide);
michael@0: bool UTF8ToWide(const char* src, size_t src_len, std::wstring* output);
michael@0: std::wstring UTF8ToWide(const ::StringPiece& utf8);
michael@0: 
michael@0: bool IsStringASCII(const std::wstring& str);
michael@0: bool IsStringASCII(const std::string& str);
michael@0: bool IsStringASCII(const string16& str);
michael@0: 
michael@0: // Specialized string-conversion functions.
michael@0: std::string IntToString(int value);
michael@0: std::wstring IntToWString(int value);
michael@0: std::string UintToString(unsigned int value);
michael@0: std::wstring UintToWString(unsigned int value);
michael@0: std::string Int64ToString(int64_t value);
michael@0: std::wstring Int64ToWString(int64_t value);
michael@0: std::string Uint64ToString(uint64_t value);
michael@0: std::wstring Uint64ToWString(uint64_t value);
michael@0: // The DoubleToString methods convert the double to a string format that
michael@0: // ignores the locale.  If you want to use locale specific formatting, use ICU.
michael@0: std::string DoubleToString(double value);
michael@0: std::wstring DoubleToWString(double value);
michael@0: 
michael@0: // Perform a best-effort conversion of the input string to a numeric type,
michael@0: // setting |*output| to the result of the conversion.  Returns true for
michael@0: // "perfect" conversions; returns false in the following cases:
michael@0: //  - Overflow/underflow.  |*output| will be set to the maximum value supported
michael@0: //    by the data type.
michael@0: //  - Trailing characters in the string after parsing the number.  |*output|
michael@0: //    will be set to the value of the number that was parsed.
michael@0: //  - No characters parseable as a number at the beginning of the string.
michael@0: //    |*output| will be set to 0.
michael@0: //  - Empty string.  |*output| will be set to 0.
michael@0: bool StringToInt(const std::string& input, int* output);
michael@0: bool StringToInt(const string16& input, int* output);
michael@0: bool StringToInt64(const std::string& input, int64_t* output);
michael@0: bool StringToInt64(const string16& input, int64_t* output);
michael@0: 
michael@0: // Convenience forms of the above, when the caller is uninterested in the
michael@0: // boolean return value.  These return only the |*output| value from the
michael@0: // above conversions: a best-effort conversion when possible, otherwise, 0.
michael@0: int StringToInt(const std::string& value);
michael@0: int StringToInt(const string16& value);
michael@0: int64_t StringToInt64(const std::string& value);
michael@0: int64_t StringToInt64(const string16& value);
michael@0: 
michael@0: // Return a C++ string given printf-like input.
michael@0: std::string StringPrintf(const char* format, ...);
michael@0: std::wstring StringPrintf(const wchar_t* format, ...);
michael@0: 
michael@0: // Store result into a supplied string and return it
michael@0: const std::string& SStringPrintf(std::string* dst, const char* format, ...);
michael@0: const std::wstring& SStringPrintf(std::wstring* dst,
michael@0:                                   const wchar_t* format, ...);
michael@0: 
michael@0: // Append result to a supplied string
michael@0: void StringAppendF(std::string* dst, const char* format, ...);
michael@0: void StringAppendF(std::wstring* dst, const wchar_t* format, ...);
michael@0: 
michael@0: //-----------------------------------------------------------------------------
michael@0: 
michael@0: // Splits |str| into a vector of strings delimited by |s|. Append the results
michael@0: // into |r| as they appear. If several instances of |s| are contiguous, or if
michael@0: // |str| begins with or ends with |s|, then an empty string is inserted.
michael@0: //
michael@0: // Every substring is trimmed of any leading or trailing white space.
michael@0: void SplitString(const std::wstring& str,
michael@0:                  wchar_t s,
michael@0:                  std::vector<std::wstring>* r);
michael@0: void SplitString(const std::string& str,
michael@0:                  char s,
michael@0:                  std::vector<std::string>* r);
michael@0: 
michael@0: #endif  // BASE_STRING_UTIL_H_