]>
Commit | Line | Data |
---|---|---|
e3f6cbd9 VS |
1 | /////////////////////////////////////////////////////////////////////////////// |
2 | // Name: wx/unichar.h | |
3 | // Purpose: wxUniChar and wxUniCharRef classes | |
4 | // Author: Vaclav Slavik | |
5 | // Created: 2007-03-19 | |
6 | // RCS-ID: $Id$ | |
7 | // Copyright: (c) 2007 REA Elektronik GmbH | |
8 | // Licence: wxWindows licence | |
9 | /////////////////////////////////////////////////////////////////////////////// | |
10 | ||
11 | #ifndef _WX_UNICHAR_H_ | |
12 | #define _WX_UNICHAR_H_ | |
13 | ||
dbecee02 | 14 | #include "wx/defs.h" |
e3f6cbd9 | 15 | #include "wx/chartype.h" |
a7ea63e2 | 16 | #include "wx/stringimpl.h" |
a962cdf4 | 17 | |
b5dbe15d VS |
18 | class WXDLLIMPEXP_FWD_BASE wxUniCharRef; |
19 | class WXDLLIMPEXP_FWD_BASE wxStringIteratorNode; | |
e3f6cbd9 VS |
20 | |
21 | // This class represents single Unicode character. It can be converted to | |
22 | // and from char or wchar_t and implements commonly used character operations. | |
23 | class WXDLLIMPEXP_BASE wxUniChar | |
24 | { | |
25 | public: | |
26 | // NB: this is not wchar_t on purpose, it needs to represent the entire | |
27 | // Unicode code points range and wchar_t may be too small for that | |
28 | // (e.g. on Win32 where wchar_t* is encoded in UTF-16) | |
dbecee02 | 29 | typedef wxUint32 value_type; |
e3f6cbd9 VS |
30 | |
31 | wxUniChar() : m_value(0) {} | |
32 | ||
33 | // Create the character from 8bit character value encoded in the current | |
34 | // locale's charset. | |
35 | wxUniChar(char c) { m_value = From8bit(c); } | |
9f1b1b78 | 36 | wxUniChar(unsigned char c) { m_value = From8bit((char)c); } |
e3f6cbd9 VS |
37 | |
38 | // Create the character from a wchar_t character value. | |
c5cf8663 | 39 | #if wxWCHAR_T_IS_REAL_TYPE |
e3f6cbd9 | 40 | wxUniChar(wchar_t c) { m_value = c; } |
436bb9f5 | 41 | #endif |
e3f6cbd9 | 42 | |
e3f6cbd9 | 43 | wxUniChar(int c) { m_value = c; } |
21f0762a | 44 | wxUniChar(unsigned int c) { m_value = c; } |
4e5b3a4e | 45 | wxUniChar(long int c) { m_value = c; } |
21f0762a VS |
46 | wxUniChar(unsigned long int c) { m_value = c; } |
47 | wxUniChar(short int c) { m_value = c; } | |
48 | wxUniChar(unsigned short int c) { m_value = c; } | |
e3f6cbd9 VS |
49 | |
50 | wxUniChar(const wxUniCharRef& c); | |
51 | ||
52 | // Returns Unicode code point value of the character | |
d1b7ed67 | 53 | value_type GetValue() const { return m_value; } |
e3f6cbd9 | 54 | |
1fc10687 VS |
55 | #if wxUSE_UNICODE_UTF8 |
56 | // buffer for single UTF-8 character | |
57 | struct Utf8CharBuffer | |
58 | { | |
59 | char data[5]; | |
60 | operator const char*() const { return data; } | |
61 | }; | |
62 | ||
63 | // returns the character encoded as UTF-8 | |
64 | // (NB: implemented in stringops.cpp) | |
65 | Utf8CharBuffer AsUTF8() const; | |
66 | #endif // wxUSE_UNICODE_UTF8 | |
67 | ||
81727065 VS |
68 | // Returns true if the character is an ASCII character: |
69 | bool IsAscii() const { return m_value < 0x80; } | |
70 | ||
2a686bd3 VZ |
71 | // Conversions to char and wchar_t types: all of those are needed to be |
72 | // able to pass wxUniChars to verious standard narrow and wide character | |
73 | // functions | |
e3f6cbd9 | 74 | operator char() const { return To8bit(m_value); } |
21f0762a | 75 | operator unsigned char() const { return (unsigned char)To8bit(m_value); } |
c5cf8663 | 76 | #if wxWCHAR_T_IS_REAL_TYPE |
d4e59879 | 77 | operator wchar_t() const { return (wchar_t)m_value; } |
21f0762a | 78 | #endif |
d4e59879 VS |
79 | operator int() const { return (int)m_value; } |
80 | operator unsigned int() const { return (unsigned int)m_value; } | |
81 | operator long int() const { return (long int)m_value; } | |
82 | operator unsigned long int() const { return (unsigned long)m_value; } | |
83 | operator short int() const { return (short int)m_value; } | |
84 | operator unsigned short int() const { return (unsigned short int)m_value; } | |
e3f6cbd9 VS |
85 | |
86 | // We need this operator for the "*p" part of expressions like "for ( | |
87 | // const_iterator p = begin() + nStart; *p; ++p )". In this case, | |
88 | // compilation would fail without it because the conversion to bool would | |
89 | // be ambiguous (there are all these int types conversions...). (And adding | |
90 | // operator unspecified_bool_type() would only makes the ambiguity worse.) | |
91 | operator bool() const { return m_value != 0; } | |
92 | bool operator!() const { return !((bool)*this); } | |
8019a673 VZ |
93 | |
94 | // And this one is needed by some (not all, but not using ifdefs makes the | |
95 | // code easier) compilers to parse "str[0] && *p" successfully | |
e3f6cbd9 | 96 | bool operator&&(bool v) const { return (bool)*this && v; } |
e3f6cbd9 VS |
97 | |
98 | // Assignment operators: | |
99 | wxUniChar& operator=(const wxUniChar& c) { m_value = c.m_value; return *this; } | |
c2d122aa | 100 | wxUniChar& operator=(const wxUniCharRef& c); |
e3f6cbd9 | 101 | wxUniChar& operator=(char c) { m_value = From8bit(c); return *this; } |
db18be57 | 102 | wxUniChar& operator=(unsigned char c) { m_value = From8bit((char)c); return *this; } |
c5cf8663 | 103 | #if wxWCHAR_T_IS_REAL_TYPE |
e3f6cbd9 | 104 | wxUniChar& operator=(wchar_t c) { m_value = c; return *this; } |
21f0762a | 105 | #endif |
171d6132 | 106 | wxUniChar& operator=(int c) { m_value = c; return *this; } |
21f0762a | 107 | wxUniChar& operator=(unsigned int c) { m_value = c; return *this; } |
4e5b3a4e | 108 | wxUniChar& operator=(long int c) { m_value = c; return *this; } |
21f0762a VS |
109 | wxUniChar& operator=(unsigned long int c) { m_value = c; return *this; } |
110 | wxUniChar& operator=(short int c) { m_value = c; return *this; } | |
111 | wxUniChar& operator=(unsigned short int c) { m_value = c; return *this; } | |
e3f6cbd9 | 112 | |
fb52c2b6 | 113 | // Comparison operators: |
e3f6cbd9 | 114 | |
fb52c2b6 | 115 | // define the given comparison operator for all the types |
98c4eb39 | 116 | #define wxDEFINE_UNICHAR_OPERATOR(op) \ |
fb52c2b6 VZ |
117 | bool operator op(const wxUniChar& c) const { return m_value op c.m_value; }\ |
118 | bool operator op(char c) const { return m_value op From8bit(c); } \ | |
db18be57 | 119 | bool operator op(unsigned char c) const { return m_value op From8bit((char)c); } \ |
21f0762a | 120 | wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return m_value op (value_type)c; } ) \ |
db18be57 | 121 | bool operator op(int c) const { return m_value op (value_type)c; } \ |
21f0762a VS |
122 | bool operator op(unsigned int c) const { return m_value op (value_type)c; } \ |
123 | bool operator op(short int c) const { return m_value op (value_type)c; } \ | |
124 | bool operator op(unsigned short int c) const { return m_value op (value_type)c; } \ | |
4e5b3a4e | 125 | bool operator op(long int c) const { return m_value op (value_type)c; } \ |
21f0762a | 126 | bool operator op(unsigned long int c) const { return m_value op (value_type)c; } |
e3f6cbd9 | 127 | |
fb52c2b6 | 128 | wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHAR_OPERATOR) |
e3f6cbd9 | 129 | |
fb52c2b6 | 130 | #undef wxDEFINE_UNICHAR_OPERATOR |
e3f6cbd9 | 131 | |
fb52c2b6 | 132 | // this is needed for expressions like 'Z'-c |
e3f6cbd9 VS |
133 | int operator-(const wxUniChar& c) const { return m_value - c.m_value; } |
134 | int operator-(char c) const { return m_value - From8bit(c); } | |
db18be57 | 135 | int operator-(unsigned char c) const { return m_value - From8bit((char)c); } |
d1b7ed67 | 136 | int operator-(wchar_t c) const { return m_value - (value_type)c; } |
21f0762a | 137 | |
e3f6cbd9 VS |
138 | |
139 | private: | |
d1b7ed67 VS |
140 | static value_type From8bit(char c); |
141 | static char To8bit(value_type c); | |
e3f6cbd9 VS |
142 | |
143 | private: | |
d1b7ed67 | 144 | value_type m_value; |
e3f6cbd9 VS |
145 | }; |
146 | ||
147 | ||
148 | // Writeable reference to a character in wxString. | |
149 | // | |
150 | // This class can be used in the same way wxChar is used, except that changing | |
151 | // its value updates the underlying string object. | |
152 | class WXDLLIMPEXP_BASE wxUniCharRef | |
153 | { | |
154 | private: | |
a962cdf4 VS |
155 | typedef wxStringImpl::iterator iterator; |
156 | ||
e3f6cbd9 | 157 | // create the reference |
81727065 | 158 | #if wxUSE_UNICODE_UTF8 |
b0c4d5d7 | 159 | wxUniCharRef(wxStringIteratorNode& node, iterator pos) : m_node(node), m_pos(pos) {} |
81727065 | 160 | #else |
a962cdf4 | 161 | wxUniCharRef(iterator pos) : m_pos(pos) {} |
81727065 | 162 | #endif |
e3f6cbd9 VS |
163 | |
164 | public: | |
165 | // NB: we have to make this public, because we don't have wxString | |
166 | // declaration available here and so can't declare wxString::iterator | |
167 | // as friend; so at least don't use a ctor but a static function | |
168 | // that must be used explicitly (this is more than using 'explicit' | |
169 | // keyword on ctor!): | |
81727065 | 170 | #if wxUSE_UNICODE_UTF8 |
b0c4d5d7 VS |
171 | static wxUniCharRef CreateForString(wxStringIteratorNode& node, iterator pos) |
172 | { return wxUniCharRef(node, pos); } | |
81727065 | 173 | #else |
a962cdf4 | 174 | static wxUniCharRef CreateForString(iterator pos) |
e3f6cbd9 | 175 | { return wxUniCharRef(pos); } |
81727065 | 176 | #endif |
e3f6cbd9 | 177 | |
d1b7ed67 | 178 | wxUniChar::value_type GetValue() const { return UniChar().GetValue(); } |
1fc10687 VS |
179 | |
180 | #if wxUSE_UNICODE_UTF8 | |
181 | wxUniChar::Utf8CharBuffer AsUTF8() const { return UniChar().AsUTF8(); } | |
182 | #endif // wxUSE_UNICODE_UTF8 | |
183 | ||
bc5443ba | 184 | bool IsAscii() const { return UniChar().IsAscii(); } |
e3f6cbd9 VS |
185 | |
186 | // Assignment operators: | |
81727065 VS |
187 | #if wxUSE_UNICODE_UTF8 |
188 | wxUniCharRef& operator=(const wxUniChar& c); | |
189 | #else | |
190 | wxUniCharRef& operator=(const wxUniChar& c) { *m_pos = c; return *this; } | |
191 | #endif | |
e3f6cbd9 | 192 | |
81727065 VS |
193 | wxUniCharRef& operator=(const wxUniCharRef& c) |
194 | { return *this = c.UniChar(); } | |
e3f6cbd9 VS |
195 | |
196 | wxUniCharRef& operator=(char c) { return *this = wxUniChar(c); } | |
db18be57 | 197 | wxUniCharRef& operator=(unsigned char c) { return *this = wxUniChar(c); } |
c5cf8663 | 198 | #if wxWCHAR_T_IS_REAL_TYPE |
e3f6cbd9 | 199 | wxUniCharRef& operator=(wchar_t c) { return *this = wxUniChar(c); } |
21f0762a | 200 | #endif |
171d6132 | 201 | wxUniCharRef& operator=(int c) { return *this = wxUniChar(c); } |
21f0762a VS |
202 | wxUniCharRef& operator=(unsigned int c) { return *this = wxUniChar(c); } |
203 | wxUniCharRef& operator=(short int c) { return *this = wxUniChar(c); } | |
204 | wxUniCharRef& operator=(unsigned short int c) { return *this = wxUniChar(c); } | |
4e5b3a4e | 205 | wxUniCharRef& operator=(long int c) { return *this = wxUniChar(c); } |
21f0762a | 206 | wxUniCharRef& operator=(unsigned long int c) { return *this = wxUniChar(c); } |
e3f6cbd9 | 207 | |
2a686bd3 | 208 | // Conversions to the same types as wxUniChar is convertible too: |
e3f6cbd9 | 209 | operator char() const { return UniChar(); } |
2a686bd3 | 210 | operator unsigned char() const { return UniChar(); } |
c5cf8663 | 211 | #if wxWCHAR_T_IS_REAL_TYPE |
436bb9f5 | 212 | operator wchar_t() const { return UniChar(); } |
c961c0cf | 213 | #endif |
21f0762a | 214 | operator int() const { return UniChar(); } |
153bc803 | 215 | operator unsigned int() const { return UniChar(); } |
21f0762a VS |
216 | operator short int() const { return UniChar(); } |
217 | operator unsigned short int() const { return UniChar(); } | |
218 | operator long int() const { return UniChar(); } | |
219 | operator unsigned long int() const { return UniChar(); } | |
e3f6cbd9 VS |
220 | |
221 | // see wxUniChar::operator bool etc. for explanation | |
222 | operator bool() const { return (bool)UniChar(); } | |
223 | bool operator!() const { return !UniChar(); } | |
e3f6cbd9 | 224 | bool operator&&(bool v) const { return UniChar() && v; } |
e3f6cbd9 | 225 | |
fb52c2b6 | 226 | // Comparison operators: |
98c4eb39 | 227 | #define wxDEFINE_UNICHARREF_OPERATOR(op) \ |
fb52c2b6 VZ |
228 | bool operator op(const wxUniCharRef& c) const { return UniChar() op c.UniChar(); }\ |
229 | bool operator op(const wxUniChar& c) const { return UniChar() op c; } \ | |
230 | bool operator op(char c) const { return UniChar() op c; } \ | |
db18be57 | 231 | bool operator op(unsigned char c) const { return UniChar() op c; } \ |
21f0762a | 232 | wxIF_WCHAR_T_TYPE( bool operator op(wchar_t c) const { return UniChar() op c; } ) \ |
db18be57 | 233 | bool operator op(int c) const { return UniChar() op c; } \ |
21f0762a VS |
234 | bool operator op(unsigned int c) const { return UniChar() op c; } \ |
235 | bool operator op(short int c) const { return UniChar() op c; } \ | |
236 | bool operator op(unsigned short int c) const { return UniChar() op c; } \ | |
4e5b3a4e | 237 | bool operator op(long int c) const { return UniChar() op c; } \ |
21f0762a | 238 | bool operator op(unsigned long int c) const { return UniChar() op c; } |
e3f6cbd9 | 239 | |
fb52c2b6 | 240 | wxFOR_ALL_COMPARISONS(wxDEFINE_UNICHARREF_OPERATOR) |
e3f6cbd9 | 241 | |
fb52c2b6 | 242 | #undef wxDEFINE_UNICHARREF_OPERATOR |
e3f6cbd9 VS |
243 | |
244 | // for expressions like c-'A': | |
245 | int operator-(const wxUniCharRef& c) const { return UniChar() - c.UniChar(); } | |
246 | int operator-(const wxUniChar& c) const { return UniChar() - c; } | |
247 | int operator-(char c) const { return UniChar() - c; } | |
db18be57 | 248 | int operator-(unsigned char c) const { return UniChar() - c; } |
e3f6cbd9 | 249 | int operator-(wchar_t c) const { return UniChar() - c; } |
e3f6cbd9 VS |
250 | |
251 | private: | |
81727065 | 252 | #if wxUSE_UNICODE_UTF8 |
467175ab | 253 | wxUniChar UniChar() const; |
81727065 | 254 | #else |
467175ab | 255 | wxUniChar UniChar() const { return *m_pos; } |
81727065 VS |
256 | #endif |
257 | ||
b5dbe15d | 258 | friend class WXDLLIMPEXP_FWD_BASE wxUniChar; |
e3f6cbd9 VS |
259 | |
260 | private: | |
81727065 VS |
261 | // reference to the string and pointer to the character in string |
262 | #if wxUSE_UNICODE_UTF8 | |
b0c4d5d7 | 263 | wxStringIteratorNode& m_node; |
81727065 | 264 | #endif |
a962cdf4 | 265 | iterator m_pos; |
e3f6cbd9 VS |
266 | }; |
267 | ||
268 | inline wxUniChar::wxUniChar(const wxUniCharRef& c) | |
269 | { | |
270 | m_value = c.UniChar().m_value; | |
271 | } | |
272 | ||
c2d122aa VS |
273 | inline wxUniChar& wxUniChar::operator=(const wxUniCharRef& c) |
274 | { | |
275 | m_value = c.UniChar().m_value; | |
276 | return *this; | |
277 | } | |
278 | ||
fb52c2b6 VZ |
279 | // Comparison operators for the case when wxUniChar(Ref) is the second operand |
280 | // implemented in terms of member comparison functions | |
e3f6cbd9 | 281 | |
fb52c2b6 | 282 | #define wxCMP_REVERSE(c1, c2, op) c2 op c1 |
e3f6cbd9 | 283 | |
fb52c2b6 VZ |
284 | wxDEFINE_COMPARISONS(char, const wxUniChar&, wxCMP_REVERSE) |
285 | wxDEFINE_COMPARISONS(char, const wxUniCharRef&, wxCMP_REVERSE) | |
e3f6cbd9 | 286 | |
fb52c2b6 VZ |
287 | wxDEFINE_COMPARISONS(wchar_t, const wxUniChar&, wxCMP_REVERSE) |
288 | wxDEFINE_COMPARISONS(wchar_t, const wxUniCharRef&, wxCMP_REVERSE) | |
e3f6cbd9 | 289 | |
fb52c2b6 | 290 | wxDEFINE_COMPARISONS(const wxUniChar&, const wxUniCharRef&, wxCMP_REVERSE) |
e3f6cbd9 | 291 | |
fb52c2b6 | 292 | #undef wxCMP_REVERSE |
e3f6cbd9 VS |
293 | |
294 | // for expressions like c-'A': | |
295 | inline int operator-(char c1, const wxUniCharRef& c2) { return -(c2 - c1); } | |
e3f6cbd9 | 296 | inline int operator-(const wxUniChar& c1, const wxUniCharRef& c2) { return -(c2 - c1); } |
436bb9f5 | 297 | inline int operator-(wchar_t c1, const wxUniCharRef& c2) { return -(c2 - c1); } |
e3f6cbd9 VS |
298 | |
299 | #endif /* _WX_UNICHAR_H_ */ |