xbmc/utils/CharsetConverter.h

   1 /*
   2  *  Copyright (C) 2005-2018 Team Kodi
   3  *  This file is part of Kodi - https://kodi.tv
   4  *
   5  *  SPDX-License-Identifier: GPL-2.0-or-later
   6  *  See LICENSES/README.md for more information.
   7  */
   8
   9 #pragma once
  10
  11 #include "settings/lib/ISettingCallback.h"
  12 #include "utils/GlobalsHandling.h"
  13
  14 #include <string>
  15 #include <utility>
  16 #include <vector>
  17
  18 class CSetting;
  19 struct StringSettingOption;
  20
  21 class CCharsetConverter : public ISettingCallback
  22 {
  23 public:
  24   CCharsetConverter();
  25
  26   void OnSettingChanged(const std::shared_ptr<const CSetting>& setting) override;
  27
  28   static void reset();
  29   static void resetSystemCharset();
  30   static void reinitCharsetsFromSettings(void);
  31
  32   static void clear();
  33
  34   /**
  35    * Convert UTF-8 string to UTF-32 string.
  36    * No RTL logical-visual transformation is performed.
  37    * @param utf8StringSrc       is source UTF-8 string to convert
  38    * @param utf32StringDst      is output UTF-32 string, empty on any error
  39    * @param failOnBadChar       if set to true function will fail on invalid character,
  40    *                            otherwise invalid character will be skipped
  41    * @return true on successful conversion, false on any error
  42    */
  43   static bool utf8ToUtf32(const std::string& utf8StringSrc, std::u32string& utf32StringDst, bool failOnBadChar = true);
  44   /**
  45    * Convert UTF-8 string to UTF-32 string.
  46    * No RTL logical-visual transformation is performed.
  47    * @param utf8StringSrc       is source UTF-8 string to convert
  48    * @param failOnBadChar       if set to true function will fail on invalid character,
  49    *                            otherwise invalid character will be skipped
  50    * @return converted string on successful conversion, empty string on any error
  51    */
  52   static std::u32string utf8ToUtf32(const std::string& utf8StringSrc, bool failOnBadChar = true);
  53   /**
  54    * Convert UTF-8 string to UTF-32 string.
  55    * RTL logical-visual transformation is optionally performed.
  56    * Use it for readable text, GUI strings etc.
  57    * @param utf8StringSrc       is source UTF-8 string to convert
  58    * @param utf32StringDst      is output UTF-32 string, empty on any error
  59    * @param bVisualBiDiFlip     allow RTL visual-logical transformation if set to true, must be set
  60    *                            to false is logical-visual transformation is already done
  61    * @param forceLTRReadingOrder        force LTR reading order
  62    * @param failOnBadChar       if set to true function will fail on invalid character,
  63    *                            otherwise invalid character will be skipped
  64    * @return true on successful conversion, false on any error
  65    */
  66   static bool utf8ToUtf32Visual(const std::string& utf8StringSrc, std::u32string& utf32StringDst, bool bVisualBiDiFlip = false, bool forceLTRReadingOrder = false, bool failOnBadChar = false);
  67   /**
  68    * Convert UTF-32 string to UTF-8 string.
  69    * No RTL visual-logical transformation is performed.
  70    * @param utf32StringSrc      is source UTF-32 string to convert
  71    * @param utf8StringDst       is output UTF-8 string, empty on any error
  72    * @param failOnBadChar       if set to true function will fail on invalid character,
  73    *                            otherwise invalid character will be skipped
  74    * @return true on successful conversion, false on any error
  75    */
  76   static bool utf32ToUtf8(const std::u32string& utf32StringSrc, std::string& utf8StringDst, bool failOnBadChar = false);
  77   /**
  78    * Convert UTF-32 string to UTF-8 string.
  79    * No RTL visual-logical transformation is performed.
  80    * @param utf32StringSrc      is source UTF-32 string to convert
  81    * @param failOnBadChar       if set to true function will fail on invalid character,
  82    *                            otherwise invalid character will be skipped
  83    * @return converted string on successful conversion, empty string on any error
  84    */
  85   static std::string utf32ToUtf8(const std::u32string& utf32StringSrc, bool failOnBadChar = false);
  86   /**
  87    * Convert UTF-32 string to wchar_t string (wstring).
  88    * No RTL visual-logical transformation is performed.
  89    * @param utf32StringSrc      is source UTF-32 string to convert
  90    * @param wStringDst          is output wchar_t string, empty on any error
  91    * @param failOnBadChar       if set to true function will fail on invalid character,
  92    *                            otherwise invalid character will be skipped
  93    * @return true on successful conversion, false on any error
  94    */
  95   static bool utf32ToW(const std::u32string& utf32StringSrc, std::wstring& wStringDst, bool failOnBadChar = false);
  96   /**
  97    * Perform logical to visual flip.
  98    * @param logicalStringSrc    is source string with logical characters order
  99    * @param visualStringDst     is output string with visual characters order, empty on any error
 100    * @param forceLTRReadingOrder        force LTR reading order
 101    * @param visualToLogicalMap    is output mapping of positions in the visual string to the logical string
 102    * @return true on success, false otherwise
 103    */
 104   static bool utf32logicalToVisualBiDi(const std::u32string& logicalStringSrc,
 105                                        std::u32string& visualStringDst,
 106                                        bool forceLTRReadingOrder = false,
 107                                        bool failOnBadString = false,
 108                                        int* visualToLogicalMap = nullptr);
 109   /**
 110    * Strictly convert wchar_t string (wstring) to UTF-32 string.
 111    * No RTL visual-logical transformation is performed.
 112    * @param wStringSrc          is source wchar_t string to convert
 113    * @param utf32StringDst      is output UTF-32 string, empty on any error
 114    * @param failOnBadChar       if set to true function will fail on invalid character,
 115    *                            otherwise invalid character will be skipped
 116    * @return true on successful conversion, false on any error
 117    */
 118   static bool wToUtf32(const std::wstring& wStringSrc, std::u32string& utf32StringDst, bool failOnBadChar = false);
 119
 120   static bool utf8ToW(const std::string& utf8StringSrc, std::wstring& wStringDst,
 121                 bool bVisualBiDiFlip = true, bool forceLTRReadingOrder = false,
 122                 bool failOnBadChar = false);
 123
 124   static bool utf16LEtoW(const std::u16string& utf16String, std::wstring& wString);
 125
 126   static bool subtitleCharsetToUtf8(const std::string& stringSrc, std::string& utf8StringDst);
 127
 128   static bool utf8ToStringCharset(const std::string& utf8StringSrc, std::string& stringDst);
 129
 130   static bool utf8ToStringCharset(std::string& stringSrcDst);
 131   static bool utf8ToSystem(std::string& stringSrcDst, bool failOnBadChar = false);
 132   static bool systemToUtf8(const std::string& sysStringSrc, std::string& utf8StringDst, bool failOnBadChar = false);
 133
 134   static bool utf8To(const std::string& strDestCharset, const std::string& utf8StringSrc, std::string& stringDst);
 135   static bool utf8To(const std::string& strDestCharset, const std::string& utf8StringSrc, std::u16string& utf16StringDst);
 136   static bool utf8To(const std::string& strDestCharset, const std::string& utf8StringSrc, std::u32string& utf32StringDst);
 137
 138   static bool ToUtf8(const std::string& strSourceCharset, const std::string& stringSrc, std::string& utf8StringDst, bool failOnBadChar = false);
 139
 140   static bool wToUTF8(const std::wstring& wStringSrc, std::string& utf8StringDst, bool failOnBadChar = false);
 141
 142   /*!
 143    *  \brief Convert UTF-16BE (u16string) string to UTF-8 string.
 144    *  No RTL visual-logical transformation is performed.
 145    *  \param utf16StringSrc Is source UTF-16BE u16string string to convert
 146    *  \param utf8StringDst Is output UTF-8 string, empty on any error
 147    *  \return True on successful conversion, false on any error
 148    */
 149   static bool utf16BEtoUTF8(const std::u16string& utf16StringSrc, std::string& utf8StringDst);
 150
 151   /*!
 152    *  \brief Convert UTF-16BE (string) string to UTF-8 string.
 153    *  No RTL visual-logical transformation is performed.
 154    *  \param utf16StringSrc Is source UTF-16BE string to convert
 155    *  \param utf8StringDst Is output UTF-8 string, empty on any error
 156    *  \return True on successful conversion, false on any error
 157    */
 158   static bool utf16BEtoUTF8(const std::string& utf16StringSrc, std::string& utf8StringDst);
 159
 160   static bool utf16LEtoUTF8(const std::u16string& utf16StringSrc, std::string& utf8StringDst);
 161   static bool ucs2ToUTF8(const std::u16string& ucs2StringSrc, std::string& utf8StringDst);
 162
 163   /*!
 164    *  \brief Convert Macintosh (string) string to UTF-8 string.
 165    *  No RTL visual-logical transformation is performed.
 166    *  \param macStringSrc Is source Macintosh string to convert
 167    *  \param utf8StringDst Is output UTF-8 string, empty on any error
 168    *  \return True on successful conversion, false on any error
 169    */
 170   static bool MacintoshToUTF8(const std::string& macStringSrc, std::string& utf8StringDst);
 171
 172   static bool utf8logicalToVisualBiDi(const std::string& utf8StringSrc, std::string& utf8StringDst, bool failOnBadString = false);
 173
 174   /**
 175    * Check if a string has RTL direction.
 176    * @param utf8StringSrc the string
 177    * @return true if the string is RTL, otherwise false
 178    */
 179   static bool utf8IsRTLBidiDirection(const std::string& utf8String);
 180
 181   static bool utf32ToStringCharset(const std::u32string& utf32StringSrc, std::string& stringDst);
 182
 183   static std::vector<std::string> getCharsetLabels();
 184   static std::string getCharsetLabelByName(const std::string& charsetName);
 185   static std::string getCharsetNameByLabel(const std::string& charsetLabel);
 186
 187   static bool unknownToUTF8(std::string& stringSrcDst);
 188   static bool unknownToUTF8(const std::string& stringSrc, std::string& utf8StringDst, bool failOnBadChar = false);
 189
 190   static bool toW(const std::string& stringSrc, std::wstring& wStringDst, const std::string& enc);
 191   static bool fromW(const std::wstring& wStringSrc, std::string& stringDst, const std::string& enc);
 192
 193   static void SettingOptionsCharsetsFiller(const std::shared_ptr<const CSetting>& setting,
 194                                            std::vector<StringSettingOption>& list,
 195                                            std::string& current,
 196                                            void* data);
 197
 198 private:
 199   static void resetUserCharset(void);
 200   static void resetSubtitleCharset(void);
 201
 202   static const int m_Utf8CharMinSize, m_Utf8CharMaxSize;
 203   class CInnerConverter;
 204 };
 205
 206 XBMC_GLOBAL_REF(CCharsetConverter,g_charsetConverter);
 207 #define g_charsetConverter XBMC_GLOBAL_USE(CCharsetConverter)