| 1 | /* | 
|---|
| 2 | * synergy -- mouse and keyboard sharing utility | 
|---|
| 3 | * Copyright (C) 2002 Chris Schoeneman | 
|---|
| 4 | * | 
|---|
| 5 | * This package is free software; you can redistribute it and/or | 
|---|
| 6 | * modify it under the terms of the GNU General Public License | 
|---|
| 7 | * found in the file COPYING that should have accompanied this file. | 
|---|
| 8 | * | 
|---|
| 9 | * This package is distributed in the hope that it will be useful, | 
|---|
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | 
|---|
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the | 
|---|
| 12 | * GNU General Public License for more details. | 
|---|
| 13 | */ | 
|---|
| 14 |  | 
|---|
| 15 | #ifndef CUNICODE_H | 
|---|
| 16 | #define CUNICODE_H | 
|---|
| 17 |  | 
|---|
| 18 | #include "CString.h" | 
|---|
| 19 | #include "BasicTypes.h" | 
|---|
| 20 |  | 
|---|
| 21 | //! Unicode utility functions | 
|---|
| 22 | /*! | 
|---|
| 23 | This class provides functions for converting between various Unicode | 
|---|
| 24 | encodings and the current locale encoding. | 
|---|
| 25 | */ | 
|---|
| 26 | class CUnicode { | 
|---|
| 27 | public: | 
|---|
| 28 | //! @name accessors | 
|---|
| 29 | //@{ | 
|---|
| 30 |  | 
|---|
| 31 | //! Test UTF-8 string for validity | 
|---|
| 32 | /*! | 
|---|
| 33 | Returns true iff the string contains a valid sequence of UTF-8 | 
|---|
| 34 | encoded characters. | 
|---|
| 35 | */ | 
|---|
| 36 | static bool                     isUTF8(const CString&); | 
|---|
| 37 |  | 
|---|
| 38 | //! Convert from UTF-8 to UCS-2 encoding | 
|---|
| 39 | /*! | 
|---|
| 40 | Convert from UTF-8 to UCS-2.  If errors is not NULL then *errors | 
|---|
| 41 | is set to true iff any character could not be encoded in UCS-2. | 
|---|
| 42 | Decoding errors do not set *errors. | 
|---|
| 43 | */ | 
|---|
| 44 | static CString          UTF8ToUCS2(const CString&, bool* errors = NULL); | 
|---|
| 45 |  | 
|---|
| 46 | //! Convert from UTF-8 to UCS-4 encoding | 
|---|
| 47 | /*! | 
|---|
| 48 | Convert from UTF-8 to UCS-4.  If errors is not NULL then *errors | 
|---|
| 49 | is set to true iff any character could not be encoded in UCS-4. | 
|---|
| 50 | Decoding errors do not set *errors. | 
|---|
| 51 | */ | 
|---|
| 52 | static CString          UTF8ToUCS4(const CString&, bool* errors = NULL); | 
|---|
| 53 |  | 
|---|
| 54 | //! Convert from UTF-8 to UTF-16 encoding | 
|---|
| 55 | /*! | 
|---|
| 56 | Convert from UTF-8 to UTF-16.  If errors is not NULL then *errors | 
|---|
| 57 | is set to true iff any character could not be encoded in UTF-16. | 
|---|
| 58 | Decoding errors do not set *errors. | 
|---|
| 59 | */ | 
|---|
| 60 | static CString          UTF8ToUTF16(const CString&, bool* errors = NULL); | 
|---|
| 61 |  | 
|---|
| 62 | //! Convert from UTF-8 to UTF-32 encoding | 
|---|
| 63 | /*! | 
|---|
| 64 | Convert from UTF-8 to UTF-32.  If errors is not NULL then *errors | 
|---|
| 65 | is set to true iff any character could not be encoded in UTF-32. | 
|---|
| 66 | Decoding errors do not set *errors. | 
|---|
| 67 | */ | 
|---|
| 68 | static CString          UTF8ToUTF32(const CString&, bool* errors = NULL); | 
|---|
| 69 |  | 
|---|
| 70 | //! Convert from UTF-8 to the current locale encoding | 
|---|
| 71 | /*! | 
|---|
| 72 | Convert from UTF-8 to the current locale encoding.  If errors is not | 
|---|
| 73 | NULL then *errors is set to true iff any character could not be encoded. | 
|---|
| 74 | Decoding errors do not set *errors. | 
|---|
| 75 | */ | 
|---|
| 76 | static CString          UTF8ToText(const CString&, bool* errors = NULL); | 
|---|
| 77 |  | 
|---|
| 78 | //! Convert from UCS-2 to UTF-8 | 
|---|
| 79 | /*! | 
|---|
| 80 | Convert from UCS-2 to UTF-8.  If errors is not NULL then *errors is | 
|---|
| 81 | set to true iff any character could not be decoded. | 
|---|
| 82 | */ | 
|---|
| 83 | static CString          UCS2ToUTF8(const CString&, bool* errors = NULL); | 
|---|
| 84 |  | 
|---|
| 85 | //! Convert from UCS-4 to UTF-8 | 
|---|
| 86 | /*! | 
|---|
| 87 | Convert from UCS-4 to UTF-8.  If errors is not NULL then *errors is | 
|---|
| 88 | set to true iff any character could not be decoded. | 
|---|
| 89 | */ | 
|---|
| 90 | static CString          UCS4ToUTF8(const CString&, bool* errors = NULL); | 
|---|
| 91 |  | 
|---|
| 92 | //! Convert from UTF-16 to UTF-8 | 
|---|
| 93 | /*! | 
|---|
| 94 | Convert from UTF-16 to UTF-8.  If errors is not NULL then *errors is | 
|---|
| 95 | set to true iff any character could not be decoded. | 
|---|
| 96 | */ | 
|---|
| 97 | static CString          UTF16ToUTF8(const CString&, bool* errors = NULL); | 
|---|
| 98 |  | 
|---|
| 99 | //! Convert from UTF-32 to UTF-8 | 
|---|
| 100 | /*! | 
|---|
| 101 | Convert from UTF-32 to UTF-8.  If errors is not NULL then *errors is | 
|---|
| 102 | set to true iff any character could not be decoded. | 
|---|
| 103 | */ | 
|---|
| 104 | static CString          UTF32ToUTF8(const CString&, bool* errors = NULL); | 
|---|
| 105 |  | 
|---|
| 106 | //! Convert from the current locale encoding to UTF-8 | 
|---|
| 107 | /*! | 
|---|
| 108 | Convert from the current locale encoding to UTF-8.  If errors is not | 
|---|
| 109 | NULL then *errors is set to true iff any character could not be decoded. | 
|---|
| 110 | */ | 
|---|
| 111 | static CString          textToUTF8(const CString&, bool* errors = NULL); | 
|---|
| 112 |  | 
|---|
| 113 | //@} | 
|---|
| 114 |  | 
|---|
| 115 | private: | 
|---|
| 116 | // convert UTF8 to wchar_t string (using whatever encoding is native | 
|---|
| 117 | // to the platform).  caller must delete[] the returned string.  the | 
|---|
| 118 | // string is *not* nul terminated;  the length (in characters) is | 
|---|
| 119 | // returned in size. | 
|---|
| 120 | static wchar_t*         UTF8ToWideChar(const CString&, | 
|---|
| 121 | UInt32& size, bool* errors); | 
|---|
| 122 |  | 
|---|
| 123 | // convert nul terminated wchar_t string (in platform's native | 
|---|
| 124 | // encoding) to UTF8. | 
|---|
| 125 | static CString          wideCharToUTF8(const wchar_t*, | 
|---|
| 126 | UInt32 size, bool* errors); | 
|---|
| 127 |  | 
|---|
| 128 | // internal conversion to UTF8 | 
|---|
| 129 | static CString          doUCS2ToUTF8(const UInt8* src, UInt32 n, bool* errors); | 
|---|
| 130 | static CString          doUCS4ToUTF8(const UInt8* src, UInt32 n, bool* errors); | 
|---|
| 131 | static CString          doUTF16ToUTF8(const UInt8* src, UInt32 n, bool* errors); | 
|---|
| 132 | static CString          doUTF32ToUTF8(const UInt8* src, UInt32 n, bool* errors); | 
|---|
| 133 |  | 
|---|
| 134 | // convert characters to/from UTF8 | 
|---|
| 135 | static UInt32           fromUTF8(const UInt8*& src, UInt32& size); | 
|---|
| 136 | static void                     toUTF8(CString& dst, UInt32 c, bool* errors); | 
|---|
| 137 |  | 
|---|
| 138 | private: | 
|---|
| 139 | static UInt32           s_invalid; | 
|---|
| 140 | static UInt32           s_replacement; | 
|---|
| 141 | }; | 
|---|
| 142 |  | 
|---|
| 143 | #endif | 
|---|