Dogcows Code - chaz/openbox/blob - otk/ustring.hh

   1 // -*- mode: C++; indent-tabs-mode: nil; c-basic-offset: 2; -*-
   2 #ifndef   __ustring_hh
   3 #define   __ustring_hh
   4
   5 /*! @file ustring.hh
   6   @brief Provides a simple UTF-8 encoded string
   7 */
   8
   9 extern "C" {
  10
  11 #ifdef HAVE_STDINT_H
  12 #  include <stdint.h>
  13 #else
  14 #  ifdef HAVE_SYS_TYPES_H
  15 #    include <sys/types.h>
  16 #  endif
  17 #endif
  18
  19 }
  20
  21 #include <string>
  22
  23 namespace otk {
  24
  25
  26 #ifdef HAVE_STDINT_H
  27 typedef uint32_t unichar;
  28 #else
  29 typedef u_int32_t unichar;
  30 #endif
  31
  32
  33 #ifndef DOXYGEN_IGNORE
  34
  35 unichar utf8_get_char(const char *p);
  36
  37 #endif // DOXYGEN_IGNORE
  38
  39 //! The iterator type for ustring
  40 /*!
  41   Note this is not a random access iterator but a bidirectional one, since all
  42   index operations need to iterate over the UTF-8 data.  Use std::advance() to
  43   move to a certain position.
  44   <p>
  45   A writeable iterator isn't provided because:  The number of bytes of the old
  46   UTF-8 character and the new one to write could be different. Therefore, any
  47   write operation would invalidate all other iterators pointing into the same
  48   string.
  49 */
  50
  51 template <class T>
  52 class ustring_Iterator
  53 {
  54 public:
  55   typedef std::bidirectional_iterator_tag   iterator_category;
  56   typedef unichar                           value_type;
  57   typedef std::string::difference_type      difference_type;
  58   //typedef value_type                        reference;
  59   typedef void                              pointer;
  60
  61   inline ustring_Iterator() {}
  62   inline ustring_Iterator(const ustring_Iterator<std::string::iterator>&
  63                           other) : _pos(other.base()) {}
  64
  65
  66   inline value_type operator*() const {
  67     // get an iterator to the internal string
  68     std::string::const_iterator pos = _pos;
  69     return utf8_get_char(&(*pos));
  70   }
  71
  72
  73   inline ustring_Iterator<T> &     operator++() {
  74     pos_ += g_utf8_skip[static_cast<unsigned char>(*pos_)];
  75     return *this;
  76   }
  77   inline ustring_Iterator<T> &     operator--() {
  78     do { --_pos; } while((*_pos & '\xC0') == '\x80');
  79     return *this;
  80   }
  81
  82   explicit inline ustring_Iterator(T pos) : _pos(pos) {}
  83   inline T base() const { return _pos; }
  84
  85 private:
  86   T _pos;
  87 };
  88
  89
  90 //! This class provides a simple wrapper to a std::string that can be encoded
  91 //! as UTF-8. The ustring::utf() member specifies if the given string is UTF-8
  92 //! encoded. ustrings default to specifying UTF-8 encoding.
  93 /*!
  94   This class does <b>not</b> handle extended 8-bit ASCII charsets like
  95   ISO-8859-1.
  96   <p>
  97   More info on Unicode and UTF-8 can be found here:
  98   http://www.cl.cam.ac.uk/~mgk25/unicode.html
  99   <p>
 100   This does not subclass std::string, because std::string was intended to be a
 101   final class.  For instance, it does not have a virtual destructor.
 102 */
 103 class ustring {
 104   std::string _string;
 105   bool _utf8;
 106
 107 public:
 108   typedef std::string::size_type                        size_type;
 109   typedef std::string::difference_type                  difference_type;
 110
 111   typedef unichar                                       value_type;
 112   //typedef unichar &                                     reference;
 113   //typedef const unichar &                               const_reference;
 114
 115   //typedef ustring_Iterator<std::string::iterator>       iterator;
 116   //typedef ustring_Iterator<std::string::const_iterator> const_iterator;
 117
 118   static const size_type npos = std::string::npos;
 119
 120   ustring(bool utf8 = true);
 121   ~ustring();
 122
 123   // make new strings
 124
 125   ustring(const ustring& other);
 126   ustring& operator=(const ustring& other);
 127   ustring(const std::string& src, bool utf8 = true);
 128   ustring(const char* src, bool utf8 = true);
 129
 130   // append to the string
 131
 132   ustring& operator+=(const ustring& src);
 133   ustring& operator+=(const char* src);
 134   ustring& operator+=(char c);
 135
 136   // sizes
 137
 138   ustring::size_type size() const;
 139   ustring::size_type bytes() const;
 140   ustring::size_type capacity() const;
 141   ustring::size_type max_size() const;
 142   bool empty() const;
 143
 144   // erase substrings
 145
 146   void clear();
 147   ustring& erase(size_type i, size_type n=npos);
 148
 149   // change the string's size
 150
 151   void resize(size_type n, char c='\0');
 152
 153   // extract characters
 154
 155   // No reference return; use replace() to write characters.
 156   value_type operator[](size_type i) const;
 157
 158   // internal data
 159
 160   const char* data()  const;
 161   const char* c_str() const;
 162
 163   // encoding
 164
 165   bool utf8() const;
 166   void setUtf8(bool utf8);
 167 };
 168
 169 }
 170
 171 #endif // __ustring_hh