]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
1 | /*============================================================================= |
2 | Copyright (c) 2001-2011 Hartmut Kaiser | |
3 | Copyright (c) 2001-2011 Joel de Guzman | |
4 | ||
5 | Distributed under the Boost Software License, Version 1.0. (See accompanying | |
6 | file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt) | |
7 | =============================================================================*/ | |
8 | #if !defined(BOOST_SPIRIT_UNICODE_1_JANUARY_12_2010_0728PM) | |
9 | #define BOOST_SPIRIT_UNICODE_1_JANUARY_12_2010_0728PM | |
10 | ||
11 | #if defined(_MSC_VER) | |
12 | #pragma once | |
13 | #endif | |
14 | ||
15 | #include <boost/cstdint.hpp> | |
16 | #include <boost/spirit/home/support/char_encoding/unicode/query.hpp> | |
17 | ||
18 | namespace boost { namespace spirit { namespace char_encoding | |
19 | { | |
20 | /////////////////////////////////////////////////////////////////////////// | |
21 | // Test characters for specified conditions (using iso8859-1) | |
22 | /////////////////////////////////////////////////////////////////////////// | |
23 | struct unicode | |
24 | { | |
25 | typedef ::boost::uint32_t char_type; | |
26 | ||
27 | /////////////////////////////////////////////////////////////////////////// | |
28 | // Posix stuff | |
29 | /////////////////////////////////////////////////////////////////////////// | |
30 | static bool | |
31 | isascii_(char_type ch) | |
32 | { | |
33 | return 0 == (ch & ~0x7f); | |
34 | } | |
35 | ||
36 | static bool | |
37 | ischar(char_type ch) | |
38 | { | |
39 | // unicode code points in the range 0x00 to 0x10FFFF | |
40 | return ch <= 0x10FFFF; | |
41 | } | |
42 | ||
43 | static bool | |
44 | isalnum(char_type ch) | |
45 | { | |
46 | return ucd::is_alphanumeric(ch); | |
47 | } | |
48 | ||
49 | static bool | |
50 | isalpha(char_type ch) | |
51 | { | |
52 | return ucd::is_alphabetic(ch); | |
53 | } | |
54 | ||
55 | static bool | |
56 | isdigit(char_type ch) | |
57 | { | |
58 | return ucd::is_decimal_number(ch); | |
59 | } | |
60 | ||
61 | static bool | |
62 | isxdigit(char_type ch) | |
63 | { | |
64 | return ucd::is_hex_digit(ch); | |
65 | } | |
66 | ||
67 | static bool | |
68 | iscntrl(char_type ch) | |
69 | { | |
70 | return ucd::is_control(ch); | |
71 | } | |
72 | ||
73 | static bool | |
74 | isgraph(char_type ch) | |
75 | { | |
76 | return ucd::is_graph(ch); | |
77 | } | |
78 | ||
79 | static bool | |
80 | islower(char_type ch) | |
81 | { | |
82 | return ucd::is_lowercase(ch); | |
83 | } | |
84 | ||
85 | static bool | |
86 | isprint(char_type ch) | |
87 | { | |
88 | return ucd::is_print(ch); | |
89 | } | |
90 | ||
91 | static bool | |
92 | ispunct(char_type ch) | |
93 | { | |
94 | return ucd::is_punctuation(ch); | |
95 | } | |
96 | ||
97 | static bool | |
98 | isspace(char_type ch) | |
99 | { | |
100 | return ucd::is_white_space(ch); | |
101 | } | |
102 | ||
103 | static bool | |
104 | isblank BOOST_PREVENT_MACRO_SUBSTITUTION (char_type ch) | |
105 | { | |
106 | return ucd::is_blank(ch); | |
107 | } | |
108 | ||
109 | static bool | |
110 | isupper(char_type ch) | |
111 | { | |
112 | return ucd::is_uppercase(ch); | |
113 | } | |
114 | ||
115 | /////////////////////////////////////////////////////////////////////////// | |
116 | // Simple character conversions | |
117 | /////////////////////////////////////////////////////////////////////////// | |
118 | ||
119 | static char_type | |
120 | tolower(char_type ch) | |
121 | { | |
122 | return ucd::to_lowercase(ch); | |
123 | } | |
124 | ||
125 | static char_type | |
126 | toupper(char_type ch) | |
127 | { | |
128 | return ucd::to_uppercase(ch); | |
129 | } | |
130 | ||
131 | static ::boost::uint32_t | |
132 | toucs4(char_type ch) | |
133 | { | |
134 | return ch; | |
135 | } | |
136 | ||
137 | /////////////////////////////////////////////////////////////////////////// | |
138 | // Major Categories | |
139 | /////////////////////////////////////////////////////////////////////////// | |
140 | #define BOOST_SPIRIT_MAJOR_CATEGORY(name) \ | |
141 | static bool \ | |
142 | is_##name(char_type ch) \ | |
143 | { \ | |
144 | return ucd::get_major_category(ch) == ucd::properties::name; \ | |
145 | } \ | |
146 | /***/ | |
147 | ||
148 | BOOST_SPIRIT_MAJOR_CATEGORY(letter) | |
149 | BOOST_SPIRIT_MAJOR_CATEGORY(mark) | |
150 | BOOST_SPIRIT_MAJOR_CATEGORY(number) | |
151 | BOOST_SPIRIT_MAJOR_CATEGORY(separator) | |
152 | BOOST_SPIRIT_MAJOR_CATEGORY(other) | |
153 | BOOST_SPIRIT_MAJOR_CATEGORY(punctuation) | |
154 | BOOST_SPIRIT_MAJOR_CATEGORY(symbol) | |
155 | ||
156 | /////////////////////////////////////////////////////////////////////////// | |
157 | // General Categories | |
158 | /////////////////////////////////////////////////////////////////////////// | |
159 | #define BOOST_SPIRIT_CATEGORY(name) \ | |
160 | static bool \ | |
161 | is_##name(char_type ch) \ | |
162 | { \ | |
163 | return ucd::get_category(ch) == ucd::properties::name; \ | |
164 | } \ | |
165 | /***/ | |
166 | ||
167 | BOOST_SPIRIT_CATEGORY(uppercase_letter) | |
168 | BOOST_SPIRIT_CATEGORY(lowercase_letter) | |
169 | BOOST_SPIRIT_CATEGORY(titlecase_letter) | |
170 | BOOST_SPIRIT_CATEGORY(modifier_letter) | |
171 | BOOST_SPIRIT_CATEGORY(other_letter) | |
172 | ||
173 | BOOST_SPIRIT_CATEGORY(nonspacing_mark) | |
174 | BOOST_SPIRIT_CATEGORY(enclosing_mark) | |
175 | BOOST_SPIRIT_CATEGORY(spacing_mark) | |
176 | ||
177 | BOOST_SPIRIT_CATEGORY(decimal_number) | |
178 | BOOST_SPIRIT_CATEGORY(letter_number) | |
179 | BOOST_SPIRIT_CATEGORY(other_number) | |
180 | ||
181 | BOOST_SPIRIT_CATEGORY(space_separator) | |
182 | BOOST_SPIRIT_CATEGORY(line_separator) | |
183 | BOOST_SPIRIT_CATEGORY(paragraph_separator) | |
184 | ||
185 | BOOST_SPIRIT_CATEGORY(control) | |
186 | BOOST_SPIRIT_CATEGORY(format) | |
187 | BOOST_SPIRIT_CATEGORY(private_use) | |
188 | BOOST_SPIRIT_CATEGORY(surrogate) | |
189 | BOOST_SPIRIT_CATEGORY(unassigned) | |
190 | ||
191 | BOOST_SPIRIT_CATEGORY(dash_punctuation) | |
192 | BOOST_SPIRIT_CATEGORY(open_punctuation) | |
193 | BOOST_SPIRIT_CATEGORY(close_punctuation) | |
194 | BOOST_SPIRIT_CATEGORY(connector_punctuation) | |
195 | BOOST_SPIRIT_CATEGORY(other_punctuation) | |
196 | BOOST_SPIRIT_CATEGORY(initial_punctuation) | |
197 | BOOST_SPIRIT_CATEGORY(final_punctuation) | |
198 | ||
199 | BOOST_SPIRIT_CATEGORY(math_symbol) | |
200 | BOOST_SPIRIT_CATEGORY(currency_symbol) | |
201 | BOOST_SPIRIT_CATEGORY(modifier_symbol) | |
202 | BOOST_SPIRIT_CATEGORY(other_symbol) | |
203 | ||
204 | /////////////////////////////////////////////////////////////////////////// | |
205 | // Derived Categories | |
206 | /////////////////////////////////////////////////////////////////////////// | |
207 | #define BOOST_SPIRIT_DERIVED_CATEGORY(name) \ | |
208 | static bool \ | |
209 | is_##name(char_type ch) \ | |
210 | { \ | |
211 | return ucd::is_##name(ch); \ | |
212 | } \ | |
213 | /***/ | |
214 | ||
215 | BOOST_SPIRIT_DERIVED_CATEGORY(alphabetic) | |
216 | BOOST_SPIRIT_DERIVED_CATEGORY(uppercase) | |
217 | BOOST_SPIRIT_DERIVED_CATEGORY(lowercase) | |
218 | BOOST_SPIRIT_DERIVED_CATEGORY(white_space) | |
219 | BOOST_SPIRIT_DERIVED_CATEGORY(hex_digit) | |
220 | BOOST_SPIRIT_DERIVED_CATEGORY(noncharacter_code_point) | |
221 | BOOST_SPIRIT_DERIVED_CATEGORY(default_ignorable_code_point) | |
222 | ||
223 | /////////////////////////////////////////////////////////////////////////// | |
224 | // Scripts | |
225 | /////////////////////////////////////////////////////////////////////////// | |
226 | #define BOOST_SPIRIT_SCRIPT(name) \ | |
227 | static bool \ | |
228 | is_##name(char_type ch) \ | |
229 | { \ | |
230 | return ucd::get_script(ch) == ucd::properties::name; \ | |
231 | } \ | |
232 | /***/ | |
233 | ||
234 | BOOST_SPIRIT_SCRIPT(arabic) | |
235 | BOOST_SPIRIT_SCRIPT(imperial_aramaic) | |
236 | BOOST_SPIRIT_SCRIPT(armenian) | |
237 | BOOST_SPIRIT_SCRIPT(avestan) | |
238 | BOOST_SPIRIT_SCRIPT(balinese) | |
239 | BOOST_SPIRIT_SCRIPT(bamum) | |
240 | BOOST_SPIRIT_SCRIPT(bengali) | |
241 | BOOST_SPIRIT_SCRIPT(bopomofo) | |
242 | BOOST_SPIRIT_SCRIPT(braille) | |
243 | BOOST_SPIRIT_SCRIPT(buginese) | |
244 | BOOST_SPIRIT_SCRIPT(buhid) | |
245 | BOOST_SPIRIT_SCRIPT(canadian_aboriginal) | |
246 | BOOST_SPIRIT_SCRIPT(carian) | |
247 | BOOST_SPIRIT_SCRIPT(cham) | |
248 | BOOST_SPIRIT_SCRIPT(cherokee) | |
249 | BOOST_SPIRIT_SCRIPT(coptic) | |
250 | BOOST_SPIRIT_SCRIPT(cypriot) | |
251 | BOOST_SPIRIT_SCRIPT(cyrillic) | |
252 | BOOST_SPIRIT_SCRIPT(devanagari) | |
253 | BOOST_SPIRIT_SCRIPT(deseret) | |
254 | BOOST_SPIRIT_SCRIPT(egyptian_hieroglyphs) | |
255 | BOOST_SPIRIT_SCRIPT(ethiopic) | |
256 | BOOST_SPIRIT_SCRIPT(georgian) | |
257 | BOOST_SPIRIT_SCRIPT(glagolitic) | |
258 | BOOST_SPIRIT_SCRIPT(gothic) | |
259 | BOOST_SPIRIT_SCRIPT(greek) | |
260 | BOOST_SPIRIT_SCRIPT(gujarati) | |
261 | BOOST_SPIRIT_SCRIPT(gurmukhi) | |
262 | BOOST_SPIRIT_SCRIPT(hangul) | |
263 | BOOST_SPIRIT_SCRIPT(han) | |
264 | BOOST_SPIRIT_SCRIPT(hanunoo) | |
265 | BOOST_SPIRIT_SCRIPT(hebrew) | |
266 | BOOST_SPIRIT_SCRIPT(hiragana) | |
267 | BOOST_SPIRIT_SCRIPT(katakana_or_hiragana) | |
268 | BOOST_SPIRIT_SCRIPT(old_italic) | |
269 | BOOST_SPIRIT_SCRIPT(javanese) | |
270 | BOOST_SPIRIT_SCRIPT(kayah_li) | |
271 | BOOST_SPIRIT_SCRIPT(katakana) | |
272 | BOOST_SPIRIT_SCRIPT(kharoshthi) | |
273 | BOOST_SPIRIT_SCRIPT(khmer) | |
274 | BOOST_SPIRIT_SCRIPT(kannada) | |
275 | BOOST_SPIRIT_SCRIPT(kaithi) | |
276 | BOOST_SPIRIT_SCRIPT(tai_tham) | |
277 | BOOST_SPIRIT_SCRIPT(lao) | |
278 | BOOST_SPIRIT_SCRIPT(latin) | |
279 | BOOST_SPIRIT_SCRIPT(lepcha) | |
280 | BOOST_SPIRIT_SCRIPT(limbu) | |
281 | BOOST_SPIRIT_SCRIPT(linear_b) | |
282 | BOOST_SPIRIT_SCRIPT(lisu) | |
283 | BOOST_SPIRIT_SCRIPT(lycian) | |
284 | BOOST_SPIRIT_SCRIPT(lydian) | |
285 | BOOST_SPIRIT_SCRIPT(malayalam) | |
286 | BOOST_SPIRIT_SCRIPT(mongolian) | |
287 | BOOST_SPIRIT_SCRIPT(meetei_mayek) | |
288 | BOOST_SPIRIT_SCRIPT(myanmar) | |
289 | BOOST_SPIRIT_SCRIPT(nko) | |
290 | BOOST_SPIRIT_SCRIPT(ogham) | |
291 | BOOST_SPIRIT_SCRIPT(ol_chiki) | |
292 | BOOST_SPIRIT_SCRIPT(old_turkic) | |
293 | BOOST_SPIRIT_SCRIPT(oriya) | |
294 | BOOST_SPIRIT_SCRIPT(osmanya) | |
295 | BOOST_SPIRIT_SCRIPT(phags_pa) | |
296 | BOOST_SPIRIT_SCRIPT(inscriptional_pahlavi) | |
297 | BOOST_SPIRIT_SCRIPT(phoenician) | |
298 | BOOST_SPIRIT_SCRIPT(inscriptional_parthian) | |
299 | BOOST_SPIRIT_SCRIPT(rejang) | |
300 | BOOST_SPIRIT_SCRIPT(runic) | |
301 | BOOST_SPIRIT_SCRIPT(samaritan) | |
302 | BOOST_SPIRIT_SCRIPT(old_south_arabian) | |
303 | BOOST_SPIRIT_SCRIPT(saurashtra) | |
304 | BOOST_SPIRIT_SCRIPT(shavian) | |
305 | BOOST_SPIRIT_SCRIPT(sinhala) | |
306 | BOOST_SPIRIT_SCRIPT(sundanese) | |
307 | BOOST_SPIRIT_SCRIPT(syloti_nagri) | |
308 | BOOST_SPIRIT_SCRIPT(syriac) | |
309 | BOOST_SPIRIT_SCRIPT(tagbanwa) | |
310 | BOOST_SPIRIT_SCRIPT(tai_le) | |
311 | BOOST_SPIRIT_SCRIPT(new_tai_lue) | |
312 | BOOST_SPIRIT_SCRIPT(tamil) | |
313 | BOOST_SPIRIT_SCRIPT(tai_viet) | |
314 | BOOST_SPIRIT_SCRIPT(telugu) | |
315 | BOOST_SPIRIT_SCRIPT(tifinagh) | |
316 | BOOST_SPIRIT_SCRIPT(tagalog) | |
317 | BOOST_SPIRIT_SCRIPT(thaana) | |
318 | BOOST_SPIRIT_SCRIPT(thai) | |
319 | BOOST_SPIRIT_SCRIPT(tibetan) | |
320 | BOOST_SPIRIT_SCRIPT(ugaritic) | |
321 | BOOST_SPIRIT_SCRIPT(vai) | |
322 | BOOST_SPIRIT_SCRIPT(old_persian) | |
323 | BOOST_SPIRIT_SCRIPT(cuneiform) | |
324 | BOOST_SPIRIT_SCRIPT(yi) | |
325 | BOOST_SPIRIT_SCRIPT(inherited) | |
326 | BOOST_SPIRIT_SCRIPT(common) | |
327 | BOOST_SPIRIT_SCRIPT(unknown) | |
328 | ||
329 | #undef BOOST_SPIRIT_MAJOR_CATEGORY | |
330 | #undef BOOST_SPIRIT_CATEGORY | |
331 | #undef BOOST_SPIRIT_DERIVED_CATEGORY | |
332 | #undef BOOST_SPIRIT_SCRIPT | |
333 | ||
334 | }; | |
335 | ||
336 | }}} | |
337 | ||
338 | #endif | |
339 |