include/wx/stringops.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        wx/stringops.h
   3 // Purpose:     implementation of wxString primitive operations
   4 // Author:      Vaclav Slavik
   5 // Modified by:
   6 // Created:     2007-04-16
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 2007 REA Elektronik GmbH
   9 // Licence:     wxWindows licence
  10 ///////////////////////////////////////////////////////////////////////////////
  11
  12 #ifndef _WX_WXSTRINGOPS_H__
  13 #define _WX_WXSTRINGOPS_H__
  14
  15 #include "wx/chartype.h"
  16 #include "wx/stringimpl.h"
  17 #include "wx/unichar.h"
  18 #include "wx/buffer.h"
  19
  20 // This header contains wxStringOperations "namespace" class that implements
  21 // elementary operations on string data as static methods; wxString methods and
  22 // iterators are implemented in terms of it. Two implementations are available,
  23 // one for UTF-8 encoded char* string and one for "raw" wchar_t* strings (or
  24 // char* in ANSI build).
  25
  26 // FIXME-UTF8: only wchar after we remove ANSI build
  27 #if wxUSE_UNICODE_WCHAR || !wxUSE_UNICODE
  28 struct WXDLLIMPEXP_BASE wxStringOperationsWchar
  29 {
  30     // moves the iterator to the next Unicode character
  31     static void IncIter(wxStringImpl::iterator& i) { ++i; }
  32     static void IncIter(wxStringImpl::const_iterator& i) { ++i; }
  33
  34     // moves the iterator to the previous Unicode character
  35     static void DecIter(wxStringImpl::iterator& i) { --i; }
  36     static void DecIter(wxStringImpl::const_iterator& i) { --i; }
  37
  38     // moves the iterator by n Unicode characters
  39     static wxStringImpl::iterator AddToIter(const wxStringImpl::iterator& i, int n)
  40         { return i + n; }
  41     static wxStringImpl::const_iterator AddToIter(const wxStringImpl::const_iterator& i, int n)
  42         { return i + n; }
  43     static const wxChar* AddToIter(const wxChar *i, int n)
  44         { return i + n; }
  45
  46     // returns distance of the two iterators in Unicode characters
  47     static int DiffIters(const wxStringImpl::iterator& i1,
  48                          const wxStringImpl::iterator& i2)
  49         { return i1 - i2; }
  50     static int DiffIters(const wxStringImpl::const_iterator& i1,
  51                          const wxStringImpl::const_iterator& i2)
  52         { return i1 - i2; }
  53
  54     // encodes the character to a form used to represent it in internal
  55     // representation (returns a string in UTF8 version)
  56     static wxChar EncodeChar(const wxUniChar& ch) { return (wxChar)ch; }
  57
  58     static wxUniChar DecodeChar(const wxStringImpl::const_iterator& i)
  59         { return *i; }
  60 };
  61 #endif // wxUSE_UNICODE_WCHAR || !wxUSE_UNICODE
  62
  63
  64 #if wxUSE_UNICODE_UTF8
  65 struct WXDLLIMPEXP_BASE wxStringOperationsUtf8
  66 {
  67     // checks correctness of UTF-8 sequence
  68     static bool IsValidUtf8String(const char *c);
  69 #ifdef __WXDEBUG__
  70     static bool IsValidUtf8LeadByte(unsigned char c);
  71 #endif
  72
  73     // table of offsets to skip forward when iterating over UTF-8 sequence
  74     static unsigned char ms_utf8IterTable[256];
  75
  76
  77     template<typename Iterator>
  78     static void IncIter(Iterator& i)
  79     {
  80         wxASSERT( IsValidUtf8LeadByte(*i) );
  81         i += ms_utf8IterTable[(unsigned char)*i];
  82     }
  83
  84     template<typename Iterator>
  85     static void DecIter(Iterator& i)
  86     {
  87         wxASSERT( IsValidUtf8LeadByte(*i) );
  88
  89         // Non-lead bytes are all in the 0x80..0xBF range (i.e. 10xxxxxx in
  90         // binary), so we just have to go back until we hit a byte that is
  91         // either < 0x80 (i.e. 0xxxxxxx in binary) or 0xC0..0xFF (11xxxxxx in
  92         // binary; this includes some invalid values, but we can ignore it
  93         // here, because we assume valid UTF-8 input for the purpose of
  94         // efficient implementation).
  95         --i;
  96         while ( ((*i) & 0xC0) == 0x80 /* 2 highest bits are '10' */ )
  97             --i;
  98     }
  99
 100     template<typename Iterator>
 101     static Iterator AddToIter(const Iterator& i, int n)
 102     {
 103         Iterator out(i);
 104
 105         if ( n > 0 )
 106         {
 107             for ( int j = 0; j < n; ++j )
 108                 IncIter(out);
 109         }
 110         else if ( n < 0 )
 111         {
 112             for ( int j = 0; j > n; --j )
 113                 DecIter(out);
 114         }
 115
 116         return out;
 117     }
 118
 119     template<typename Iterator>
 120     static int DiffIters(Iterator i1, Iterator i2)
 121     {
 122         int dist = 0;
 123
 124         if ( i1 < i2 )
 125         {
 126             while ( i1 != i2 )
 127             {
 128                 IncIter(i1);
 129                 dist--;
 130             }
 131         }
 132         else if ( i2 < i1 )
 133         {
 134             while ( i2 != i1 )
 135             {
 136                 IncIter(i2);
 137                 dist++;
 138             }
 139         }
 140
 141         return dist;
 142     }
 143
 144     // buffer for single UTF-8 character
 145     struct Utf8CharBuffer
 146     {
 147         char data[5];
 148         operator const char*() const { return data; }
 149     };
 150
 151     // encodes the character as UTF-8:
 152     static Utf8CharBuffer EncodeChar(const wxUniChar& ch);
 153
 154     // returns n copies of ch encoded in UTF-8 string
 155     static wxCharBuffer EncodeNChars(size_t n, const wxUniChar& ch);
 156
 157     // returns the length of UTF-8 encoding of the character with lead byte 'c'
 158     static size_t GetUtf8CharLength(char c)
 159     {
 160         wxASSERT( IsValidUtf8LeadByte(c) );
 161         return ms_utf8IterTable[(unsigned char)c];
 162     }
 163
 164     // decodes single UTF-8 character from UTF-8 string
 165     static wxUniChar DecodeChar(wxStringImpl::const_iterator i);
 166 };
 167 #endif // wxUSE_UNICODE_UTF8
 168
 169
 170 #if wxUSE_UNICODE_UTF8
 171 typedef wxStringOperationsUtf8 wxStringOperations;
 172 #else
 173 typedef wxStringOperationsWchar wxStringOperations;
 174 #endif
 175
 176 #endif  // _WX_WXSTRINGOPS_H_