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