Chris@102
|
1 /*=============================================================================
|
Chris@102
|
2 Copyright (c) 2001-2014 Joel de Guzman
|
Chris@102
|
3
|
Chris@102
|
4 Distributed under the Boost Software License, Version 1.0. (See accompanying
|
Chris@102
|
5 file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
Chris@102
|
6 ==============================================================================*/
|
Chris@102
|
7 #if !defined(BOOST_SPIRIT_X3_UNICODE_JAN_20_2012_1218AM)
|
Chris@102
|
8 #define BOOST_SPIRIT_X3_UNICODE_JAN_20_2012_1218AM
|
Chris@102
|
9
|
Chris@102
|
10 #if defined(_MSC_VER)
|
Chris@102
|
11 #pragma once
|
Chris@102
|
12 #endif
|
Chris@102
|
13
|
Chris@102
|
14 #include <boost/spirit/home/x3/char/char_parser.hpp>
|
Chris@102
|
15 #include <boost/spirit/home/x3/char/char.hpp>
|
Chris@102
|
16 #include <boost/spirit/home/x3/char/detail/cast_char.hpp>
|
Chris@102
|
17 #include <boost/spirit/home/support/char_encoding/unicode.hpp>
|
Chris@102
|
18
|
Chris@102
|
19 namespace boost { namespace spirit { namespace x3
|
Chris@102
|
20 {
|
Chris@102
|
21 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
22 // Unicode Major Categories
|
Chris@102
|
23 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
24 struct char_tag;
|
Chris@102
|
25 struct alnum_tag;
|
Chris@102
|
26 struct alpha_tag;
|
Chris@102
|
27 struct blank_tag;
|
Chris@102
|
28 struct cntrl_tag;
|
Chris@102
|
29 struct digit_tag;
|
Chris@102
|
30 struct graph_tag;
|
Chris@102
|
31 struct print_tag;
|
Chris@102
|
32 struct punct_tag;
|
Chris@102
|
33 struct space_tag;
|
Chris@102
|
34 struct xdigit_tag;
|
Chris@102
|
35 struct lower_tag;
|
Chris@102
|
36 struct upper_tag;
|
Chris@102
|
37
|
Chris@102
|
38 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
39 // Unicode Major Categories
|
Chris@102
|
40 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
41 struct letter_tag {};
|
Chris@102
|
42 struct mark_tag {};
|
Chris@102
|
43 struct number_tag {};
|
Chris@102
|
44 struct separator_tag {};
|
Chris@102
|
45 struct other_tag {};
|
Chris@102
|
46 struct punctuation_tag {};
|
Chris@102
|
47 struct symbol_tag {};
|
Chris@102
|
48
|
Chris@102
|
49 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
50 // Unicode General Categories
|
Chris@102
|
51 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
52 struct uppercase_letter_tag {};
|
Chris@102
|
53 struct lowercase_letter_tag {};
|
Chris@102
|
54 struct titlecase_letter_tag {};
|
Chris@102
|
55 struct modifier_letter_tag {};
|
Chris@102
|
56 struct other_letter_tag {};
|
Chris@102
|
57
|
Chris@102
|
58 struct nonspacing_mark_tag {};
|
Chris@102
|
59 struct enclosing_mark_tag {};
|
Chris@102
|
60 struct spacing_mark_tag {};
|
Chris@102
|
61
|
Chris@102
|
62 struct decimal_number_tag {};
|
Chris@102
|
63 struct letter_number_tag {};
|
Chris@102
|
64 struct other_number_tag {};
|
Chris@102
|
65
|
Chris@102
|
66 struct space_separator_tag {};
|
Chris@102
|
67 struct line_separator_tag {};
|
Chris@102
|
68 struct paragraph_separator_tag {};
|
Chris@102
|
69
|
Chris@102
|
70 struct control_tag {};
|
Chris@102
|
71 struct format_tag {};
|
Chris@102
|
72 struct private_use_tag {};
|
Chris@102
|
73 struct surrogate_tag {};
|
Chris@102
|
74 struct unassigned_tag {};
|
Chris@102
|
75
|
Chris@102
|
76 struct dash_punctuation_tag {};
|
Chris@102
|
77 struct open_punctuation_tag {};
|
Chris@102
|
78 struct close_punctuation_tag {};
|
Chris@102
|
79 struct connector_punctuation_tag {};
|
Chris@102
|
80 struct other_punctuation_tag {};
|
Chris@102
|
81 struct initial_punctuation_tag {};
|
Chris@102
|
82 struct final_punctuation_tag {};
|
Chris@102
|
83
|
Chris@102
|
84 struct math_symbol_tag {};
|
Chris@102
|
85 struct currency_symbol_tag {};
|
Chris@102
|
86 struct modifier_symbol_tag {};
|
Chris@102
|
87 struct other_symbol_tag {};
|
Chris@102
|
88
|
Chris@102
|
89 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
90 // Unicode Derived Categories
|
Chris@102
|
91 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
92 struct alphabetic_tag {};
|
Chris@102
|
93 struct uppercase_tag {};
|
Chris@102
|
94 struct lowercase_tag {};
|
Chris@102
|
95 struct white_space_tag {};
|
Chris@102
|
96 struct hex_digit_tag {};
|
Chris@102
|
97 struct noncharacter_code_point_tag {};
|
Chris@102
|
98 struct default_ignorable_code_point_tag {};
|
Chris@102
|
99
|
Chris@102
|
100 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
101 // Unicode Scripts
|
Chris@102
|
102 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
103 struct arabic_tag {};
|
Chris@102
|
104 struct imperial_aramaic_tag {};
|
Chris@102
|
105 struct armenian_tag {};
|
Chris@102
|
106 struct avestan_tag {};
|
Chris@102
|
107 struct balinese_tag {};
|
Chris@102
|
108 struct bamum_tag {};
|
Chris@102
|
109 struct bengali_tag {};
|
Chris@102
|
110 struct bopomofo_tag {};
|
Chris@102
|
111 struct braille_tag {};
|
Chris@102
|
112 struct buginese_tag {};
|
Chris@102
|
113 struct buhid_tag {};
|
Chris@102
|
114 struct canadian_aboriginal_tag {};
|
Chris@102
|
115 struct carian_tag {};
|
Chris@102
|
116 struct cham_tag {};
|
Chris@102
|
117 struct cherokee_tag {};
|
Chris@102
|
118 struct coptic_tag {};
|
Chris@102
|
119 struct cypriot_tag {};
|
Chris@102
|
120 struct cyrillic_tag {};
|
Chris@102
|
121 struct devanagari_tag {};
|
Chris@102
|
122 struct deseret_tag {};
|
Chris@102
|
123 struct egyptian_hieroglyphs_tag {};
|
Chris@102
|
124 struct ethiopic_tag {};
|
Chris@102
|
125 struct georgian_tag {};
|
Chris@102
|
126 struct glagolitic_tag {};
|
Chris@102
|
127 struct gothic_tag {};
|
Chris@102
|
128 struct greek_tag {};
|
Chris@102
|
129 struct gujarati_tag {};
|
Chris@102
|
130 struct gurmukhi_tag {};
|
Chris@102
|
131 struct hangul_tag {};
|
Chris@102
|
132 struct han_tag {};
|
Chris@102
|
133 struct hanunoo_tag {};
|
Chris@102
|
134 struct hebrew_tag {};
|
Chris@102
|
135 struct hiragana_tag {};
|
Chris@102
|
136 struct katakana_or_hiragana_tag {};
|
Chris@102
|
137 struct old_italic_tag {};
|
Chris@102
|
138 struct javanese_tag {};
|
Chris@102
|
139 struct kayah_li_tag {};
|
Chris@102
|
140 struct katakana_tag {};
|
Chris@102
|
141 struct kharoshthi_tag {};
|
Chris@102
|
142 struct khmer_tag {};
|
Chris@102
|
143 struct kannada_tag {};
|
Chris@102
|
144 struct kaithi_tag {};
|
Chris@102
|
145 struct tai_tham_tag {};
|
Chris@102
|
146 struct lao_tag {};
|
Chris@102
|
147 struct latin_tag {};
|
Chris@102
|
148 struct lepcha_tag {};
|
Chris@102
|
149 struct limbu_tag {};
|
Chris@102
|
150 struct linear_b_tag {};
|
Chris@102
|
151 struct lisu_tag {};
|
Chris@102
|
152 struct lycian_tag {};
|
Chris@102
|
153 struct lydian_tag {};
|
Chris@102
|
154 struct malayalam_tag {};
|
Chris@102
|
155 struct mongolian_tag {};
|
Chris@102
|
156 struct meetei_mayek_tag {};
|
Chris@102
|
157 struct myanmar_tag {};
|
Chris@102
|
158 struct nko_tag {};
|
Chris@102
|
159 struct ogham_tag {};
|
Chris@102
|
160 struct ol_chiki_tag {};
|
Chris@102
|
161 struct old_turkic_tag {};
|
Chris@102
|
162 struct oriya_tag {};
|
Chris@102
|
163 struct osmanya_tag {};
|
Chris@102
|
164 struct phags_pa_tag {};
|
Chris@102
|
165 struct inscriptional_pahlavi_tag {};
|
Chris@102
|
166 struct phoenician_tag {};
|
Chris@102
|
167 struct inscriptional_parthian_tag {};
|
Chris@102
|
168 struct rejang_tag {};
|
Chris@102
|
169 struct runic_tag {};
|
Chris@102
|
170 struct samaritan_tag {};
|
Chris@102
|
171 struct old_south_arabian_tag {};
|
Chris@102
|
172 struct saurashtra_tag {};
|
Chris@102
|
173 struct shavian_tag {};
|
Chris@102
|
174 struct sinhala_tag {};
|
Chris@102
|
175 struct sundanese_tag {};
|
Chris@102
|
176 struct syloti_nagri_tag {};
|
Chris@102
|
177 struct syriac_tag {};
|
Chris@102
|
178 struct tagbanwa_tag {};
|
Chris@102
|
179 struct tai_le_tag {};
|
Chris@102
|
180 struct new_tai_lue_tag {};
|
Chris@102
|
181 struct tamil_tag {};
|
Chris@102
|
182 struct tai_viet_tag {};
|
Chris@102
|
183 struct telugu_tag {};
|
Chris@102
|
184 struct tifinagh_tag {};
|
Chris@102
|
185 struct tagalog_tag {};
|
Chris@102
|
186 struct thaana_tag {};
|
Chris@102
|
187 struct thai_tag {};
|
Chris@102
|
188 struct tibetan_tag {};
|
Chris@102
|
189 struct ugaritic_tag {};
|
Chris@102
|
190 struct vai_tag {};
|
Chris@102
|
191 struct old_persian_tag {};
|
Chris@102
|
192 struct cuneiform_tag {};
|
Chris@102
|
193 struct yi_tag {};
|
Chris@102
|
194 struct inherited_tag {};
|
Chris@102
|
195 struct common_tag {};
|
Chris@102
|
196 struct unknown_tag {};
|
Chris@102
|
197
|
Chris@102
|
198 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
199 struct unicode_char_class_base
|
Chris@102
|
200 {
|
Chris@102
|
201 typedef char_encoding::unicode encoding;
|
Chris@102
|
202 typedef char_encoding::unicode::char_type char_type;
|
Chris@102
|
203
|
Chris@102
|
204 #define BOOST_SPIRIT_X3_BASIC_CLASSIFY(name) \
|
Chris@102
|
205 template <typename Char> \
|
Chris@102
|
206 static bool \
|
Chris@102
|
207 is(name##_tag, Char ch) \
|
Chris@102
|
208 { \
|
Chris@102
|
209 return encoding::is ##name \
|
Chris@102
|
210 BOOST_PREVENT_MACRO_SUBSTITUTION \
|
Chris@102
|
211 (detail::cast_char<char_type>(ch)); \
|
Chris@102
|
212 } \
|
Chris@102
|
213 /***/
|
Chris@102
|
214
|
Chris@102
|
215 #define BOOST_SPIRIT_X3_CLASSIFY(name) \
|
Chris@102
|
216 template <typename Char> \
|
Chris@102
|
217 static bool \
|
Chris@102
|
218 is(name##_tag, Char ch) \
|
Chris@102
|
219 { \
|
Chris@102
|
220 return encoding::is_##name \
|
Chris@102
|
221 BOOST_PREVENT_MACRO_SUBSTITUTION \
|
Chris@102
|
222 (detail::cast_char<char_type>(ch)); \
|
Chris@102
|
223 } \
|
Chris@102
|
224 /***/
|
Chris@102
|
225
|
Chris@102
|
226
|
Chris@102
|
227 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
228 // Unicode Major Categories
|
Chris@102
|
229 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
230 BOOST_SPIRIT_X3_BASIC_CLASSIFY(char)
|
Chris@102
|
231 BOOST_SPIRIT_X3_BASIC_CLASSIFY(alnum)
|
Chris@102
|
232 BOOST_SPIRIT_X3_BASIC_CLASSIFY(alpha)
|
Chris@102
|
233 BOOST_SPIRIT_X3_BASIC_CLASSIFY(digit)
|
Chris@102
|
234 BOOST_SPIRIT_X3_BASIC_CLASSIFY(xdigit)
|
Chris@102
|
235 BOOST_SPIRIT_X3_BASIC_CLASSIFY(cntrl)
|
Chris@102
|
236 BOOST_SPIRIT_X3_BASIC_CLASSIFY(graph)
|
Chris@102
|
237 BOOST_SPIRIT_X3_BASIC_CLASSIFY(lower)
|
Chris@102
|
238 BOOST_SPIRIT_X3_BASIC_CLASSIFY(print)
|
Chris@102
|
239 BOOST_SPIRIT_X3_BASIC_CLASSIFY(punct)
|
Chris@102
|
240 BOOST_SPIRIT_X3_BASIC_CLASSIFY(space)
|
Chris@102
|
241 BOOST_SPIRIT_X3_BASIC_CLASSIFY(blank)
|
Chris@102
|
242 BOOST_SPIRIT_X3_BASIC_CLASSIFY(upper)
|
Chris@102
|
243
|
Chris@102
|
244 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
245 // Unicode Major Categories
|
Chris@102
|
246 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
247 BOOST_SPIRIT_X3_CLASSIFY(letter)
|
Chris@102
|
248 BOOST_SPIRIT_X3_CLASSIFY(mark)
|
Chris@102
|
249 BOOST_SPIRIT_X3_CLASSIFY(number)
|
Chris@102
|
250 BOOST_SPIRIT_X3_CLASSIFY(separator)
|
Chris@102
|
251 BOOST_SPIRIT_X3_CLASSIFY(other)
|
Chris@102
|
252 BOOST_SPIRIT_X3_CLASSIFY(punctuation)
|
Chris@102
|
253 BOOST_SPIRIT_X3_CLASSIFY(symbol)
|
Chris@102
|
254
|
Chris@102
|
255 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
256 // Unicode General Categories
|
Chris@102
|
257 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
258 BOOST_SPIRIT_X3_CLASSIFY(uppercase_letter)
|
Chris@102
|
259 BOOST_SPIRIT_X3_CLASSIFY(lowercase_letter)
|
Chris@102
|
260 BOOST_SPIRIT_X3_CLASSIFY(titlecase_letter)
|
Chris@102
|
261 BOOST_SPIRIT_X3_CLASSIFY(modifier_letter)
|
Chris@102
|
262 BOOST_SPIRIT_X3_CLASSIFY(other_letter)
|
Chris@102
|
263
|
Chris@102
|
264 BOOST_SPIRIT_X3_CLASSIFY(nonspacing_mark)
|
Chris@102
|
265 BOOST_SPIRIT_X3_CLASSIFY(enclosing_mark)
|
Chris@102
|
266 BOOST_SPIRIT_X3_CLASSIFY(spacing_mark)
|
Chris@102
|
267
|
Chris@102
|
268 BOOST_SPIRIT_X3_CLASSIFY(decimal_number)
|
Chris@102
|
269 BOOST_SPIRIT_X3_CLASSIFY(letter_number)
|
Chris@102
|
270 BOOST_SPIRIT_X3_CLASSIFY(other_number)
|
Chris@102
|
271
|
Chris@102
|
272 BOOST_SPIRIT_X3_CLASSIFY(space_separator)
|
Chris@102
|
273 BOOST_SPIRIT_X3_CLASSIFY(line_separator)
|
Chris@102
|
274 BOOST_SPIRIT_X3_CLASSIFY(paragraph_separator)
|
Chris@102
|
275
|
Chris@102
|
276 BOOST_SPIRIT_X3_CLASSIFY(control)
|
Chris@102
|
277 BOOST_SPIRIT_X3_CLASSIFY(format)
|
Chris@102
|
278 BOOST_SPIRIT_X3_CLASSIFY(private_use)
|
Chris@102
|
279 BOOST_SPIRIT_X3_CLASSIFY(surrogate)
|
Chris@102
|
280 BOOST_SPIRIT_X3_CLASSIFY(unassigned)
|
Chris@102
|
281
|
Chris@102
|
282 BOOST_SPIRIT_X3_CLASSIFY(dash_punctuation)
|
Chris@102
|
283 BOOST_SPIRIT_X3_CLASSIFY(open_punctuation)
|
Chris@102
|
284 BOOST_SPIRIT_X3_CLASSIFY(close_punctuation)
|
Chris@102
|
285 BOOST_SPIRIT_X3_CLASSIFY(connector_punctuation)
|
Chris@102
|
286 BOOST_SPIRIT_X3_CLASSIFY(other_punctuation)
|
Chris@102
|
287 BOOST_SPIRIT_X3_CLASSIFY(initial_punctuation)
|
Chris@102
|
288 BOOST_SPIRIT_X3_CLASSIFY(final_punctuation)
|
Chris@102
|
289
|
Chris@102
|
290 BOOST_SPIRIT_X3_CLASSIFY(math_symbol)
|
Chris@102
|
291 BOOST_SPIRIT_X3_CLASSIFY(currency_symbol)
|
Chris@102
|
292 BOOST_SPIRIT_X3_CLASSIFY(modifier_symbol)
|
Chris@102
|
293 BOOST_SPIRIT_X3_CLASSIFY(other_symbol)
|
Chris@102
|
294
|
Chris@102
|
295 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
296 // Unicode Derived Categories
|
Chris@102
|
297 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
298 BOOST_SPIRIT_X3_CLASSIFY(alphabetic)
|
Chris@102
|
299 BOOST_SPIRIT_X3_CLASSIFY(uppercase)
|
Chris@102
|
300 BOOST_SPIRIT_X3_CLASSIFY(lowercase)
|
Chris@102
|
301 BOOST_SPIRIT_X3_CLASSIFY(white_space)
|
Chris@102
|
302 BOOST_SPIRIT_X3_CLASSIFY(hex_digit)
|
Chris@102
|
303 BOOST_SPIRIT_X3_CLASSIFY(noncharacter_code_point)
|
Chris@102
|
304 BOOST_SPIRIT_X3_CLASSIFY(default_ignorable_code_point)
|
Chris@102
|
305
|
Chris@102
|
306 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
307 // Unicode Scripts
|
Chris@102
|
308 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
309 BOOST_SPIRIT_X3_CLASSIFY(arabic)
|
Chris@102
|
310 BOOST_SPIRIT_X3_CLASSIFY(imperial_aramaic)
|
Chris@102
|
311 BOOST_SPIRIT_X3_CLASSIFY(armenian)
|
Chris@102
|
312 BOOST_SPIRIT_X3_CLASSIFY(avestan)
|
Chris@102
|
313 BOOST_SPIRIT_X3_CLASSIFY(balinese)
|
Chris@102
|
314 BOOST_SPIRIT_X3_CLASSIFY(bamum)
|
Chris@102
|
315 BOOST_SPIRIT_X3_CLASSIFY(bengali)
|
Chris@102
|
316 BOOST_SPIRIT_X3_CLASSIFY(bopomofo)
|
Chris@102
|
317 BOOST_SPIRIT_X3_CLASSIFY(braille)
|
Chris@102
|
318 BOOST_SPIRIT_X3_CLASSIFY(buginese)
|
Chris@102
|
319 BOOST_SPIRIT_X3_CLASSIFY(buhid)
|
Chris@102
|
320 BOOST_SPIRIT_X3_CLASSIFY(canadian_aboriginal)
|
Chris@102
|
321 BOOST_SPIRIT_X3_CLASSIFY(carian)
|
Chris@102
|
322 BOOST_SPIRIT_X3_CLASSIFY(cham)
|
Chris@102
|
323 BOOST_SPIRIT_X3_CLASSIFY(cherokee)
|
Chris@102
|
324 BOOST_SPIRIT_X3_CLASSIFY(coptic)
|
Chris@102
|
325 BOOST_SPIRIT_X3_CLASSIFY(cypriot)
|
Chris@102
|
326 BOOST_SPIRIT_X3_CLASSIFY(cyrillic)
|
Chris@102
|
327 BOOST_SPIRIT_X3_CLASSIFY(devanagari)
|
Chris@102
|
328 BOOST_SPIRIT_X3_CLASSIFY(deseret)
|
Chris@102
|
329 BOOST_SPIRIT_X3_CLASSIFY(egyptian_hieroglyphs)
|
Chris@102
|
330 BOOST_SPIRIT_X3_CLASSIFY(ethiopic)
|
Chris@102
|
331 BOOST_SPIRIT_X3_CLASSIFY(georgian)
|
Chris@102
|
332 BOOST_SPIRIT_X3_CLASSIFY(glagolitic)
|
Chris@102
|
333 BOOST_SPIRIT_X3_CLASSIFY(gothic)
|
Chris@102
|
334 BOOST_SPIRIT_X3_CLASSIFY(greek)
|
Chris@102
|
335 BOOST_SPIRIT_X3_CLASSIFY(gujarati)
|
Chris@102
|
336 BOOST_SPIRIT_X3_CLASSIFY(gurmukhi)
|
Chris@102
|
337 BOOST_SPIRIT_X3_CLASSIFY(hangul)
|
Chris@102
|
338 BOOST_SPIRIT_X3_CLASSIFY(han)
|
Chris@102
|
339 BOOST_SPIRIT_X3_CLASSIFY(hanunoo)
|
Chris@102
|
340 BOOST_SPIRIT_X3_CLASSIFY(hebrew)
|
Chris@102
|
341 BOOST_SPIRIT_X3_CLASSIFY(hiragana)
|
Chris@102
|
342 BOOST_SPIRIT_X3_CLASSIFY(katakana_or_hiragana)
|
Chris@102
|
343 BOOST_SPIRIT_X3_CLASSIFY(old_italic)
|
Chris@102
|
344 BOOST_SPIRIT_X3_CLASSIFY(javanese)
|
Chris@102
|
345 BOOST_SPIRIT_X3_CLASSIFY(kayah_li)
|
Chris@102
|
346 BOOST_SPIRIT_X3_CLASSIFY(katakana)
|
Chris@102
|
347 BOOST_SPIRIT_X3_CLASSIFY(kharoshthi)
|
Chris@102
|
348 BOOST_SPIRIT_X3_CLASSIFY(khmer)
|
Chris@102
|
349 BOOST_SPIRIT_X3_CLASSIFY(kannada)
|
Chris@102
|
350 BOOST_SPIRIT_X3_CLASSIFY(kaithi)
|
Chris@102
|
351 BOOST_SPIRIT_X3_CLASSIFY(tai_tham)
|
Chris@102
|
352 BOOST_SPIRIT_X3_CLASSIFY(lao)
|
Chris@102
|
353 BOOST_SPIRIT_X3_CLASSIFY(latin)
|
Chris@102
|
354 BOOST_SPIRIT_X3_CLASSIFY(lepcha)
|
Chris@102
|
355 BOOST_SPIRIT_X3_CLASSIFY(limbu)
|
Chris@102
|
356 BOOST_SPIRIT_X3_CLASSIFY(linear_b)
|
Chris@102
|
357 BOOST_SPIRIT_X3_CLASSIFY(lisu)
|
Chris@102
|
358 BOOST_SPIRIT_X3_CLASSIFY(lycian)
|
Chris@102
|
359 BOOST_SPIRIT_X3_CLASSIFY(lydian)
|
Chris@102
|
360 BOOST_SPIRIT_X3_CLASSIFY(malayalam)
|
Chris@102
|
361 BOOST_SPIRIT_X3_CLASSIFY(mongolian)
|
Chris@102
|
362 BOOST_SPIRIT_X3_CLASSIFY(meetei_mayek)
|
Chris@102
|
363 BOOST_SPIRIT_X3_CLASSIFY(myanmar)
|
Chris@102
|
364 BOOST_SPIRIT_X3_CLASSIFY(nko)
|
Chris@102
|
365 BOOST_SPIRIT_X3_CLASSIFY(ogham)
|
Chris@102
|
366 BOOST_SPIRIT_X3_CLASSIFY(ol_chiki)
|
Chris@102
|
367 BOOST_SPIRIT_X3_CLASSIFY(old_turkic)
|
Chris@102
|
368 BOOST_SPIRIT_X3_CLASSIFY(oriya)
|
Chris@102
|
369 BOOST_SPIRIT_X3_CLASSIFY(osmanya)
|
Chris@102
|
370 BOOST_SPIRIT_X3_CLASSIFY(phags_pa)
|
Chris@102
|
371 BOOST_SPIRIT_X3_CLASSIFY(inscriptional_pahlavi)
|
Chris@102
|
372 BOOST_SPIRIT_X3_CLASSIFY(phoenician)
|
Chris@102
|
373 BOOST_SPIRIT_X3_CLASSIFY(inscriptional_parthian)
|
Chris@102
|
374 BOOST_SPIRIT_X3_CLASSIFY(rejang)
|
Chris@102
|
375 BOOST_SPIRIT_X3_CLASSIFY(runic)
|
Chris@102
|
376 BOOST_SPIRIT_X3_CLASSIFY(samaritan)
|
Chris@102
|
377 BOOST_SPIRIT_X3_CLASSIFY(old_south_arabian)
|
Chris@102
|
378 BOOST_SPIRIT_X3_CLASSIFY(saurashtra)
|
Chris@102
|
379 BOOST_SPIRIT_X3_CLASSIFY(shavian)
|
Chris@102
|
380 BOOST_SPIRIT_X3_CLASSIFY(sinhala)
|
Chris@102
|
381 BOOST_SPIRIT_X3_CLASSIFY(sundanese)
|
Chris@102
|
382 BOOST_SPIRIT_X3_CLASSIFY(syloti_nagri)
|
Chris@102
|
383 BOOST_SPIRIT_X3_CLASSIFY(syriac)
|
Chris@102
|
384 BOOST_SPIRIT_X3_CLASSIFY(tagbanwa)
|
Chris@102
|
385 BOOST_SPIRIT_X3_CLASSIFY(tai_le)
|
Chris@102
|
386 BOOST_SPIRIT_X3_CLASSIFY(new_tai_lue)
|
Chris@102
|
387 BOOST_SPIRIT_X3_CLASSIFY(tamil)
|
Chris@102
|
388 BOOST_SPIRIT_X3_CLASSIFY(tai_viet)
|
Chris@102
|
389 BOOST_SPIRIT_X3_CLASSIFY(telugu)
|
Chris@102
|
390 BOOST_SPIRIT_X3_CLASSIFY(tifinagh)
|
Chris@102
|
391 BOOST_SPIRIT_X3_CLASSIFY(tagalog)
|
Chris@102
|
392 BOOST_SPIRIT_X3_CLASSIFY(thaana)
|
Chris@102
|
393 BOOST_SPIRIT_X3_CLASSIFY(thai)
|
Chris@102
|
394 BOOST_SPIRIT_X3_CLASSIFY(tibetan)
|
Chris@102
|
395 BOOST_SPIRIT_X3_CLASSIFY(ugaritic)
|
Chris@102
|
396 BOOST_SPIRIT_X3_CLASSIFY(vai)
|
Chris@102
|
397 BOOST_SPIRIT_X3_CLASSIFY(old_persian)
|
Chris@102
|
398 BOOST_SPIRIT_X3_CLASSIFY(cuneiform)
|
Chris@102
|
399 BOOST_SPIRIT_X3_CLASSIFY(yi)
|
Chris@102
|
400 BOOST_SPIRIT_X3_CLASSIFY(inherited)
|
Chris@102
|
401 BOOST_SPIRIT_X3_CLASSIFY(common)
|
Chris@102
|
402 BOOST_SPIRIT_X3_CLASSIFY(unknown)
|
Chris@102
|
403
|
Chris@102
|
404 #undef BOOST_SPIRIT_X3_BASIC_CLASSIFY
|
Chris@102
|
405 #undef BOOST_SPIRIT_X3_CLASSIFY
|
Chris@102
|
406 };
|
Chris@102
|
407
|
Chris@102
|
408 template <typename Tag>
|
Chris@102
|
409 struct unicode_char_class
|
Chris@102
|
410 : char_parser<unicode_char_class<Tag>>
|
Chris@102
|
411 {
|
Chris@102
|
412 typedef char_encoding::unicode encoding;
|
Chris@102
|
413 typedef Tag tag;
|
Chris@102
|
414 typedef typename encoding::char_type char_type;
|
Chris@102
|
415 typedef char_type attribute_type;
|
Chris@102
|
416 static bool const has_attribute = true;
|
Chris@102
|
417
|
Chris@102
|
418 template <typename Char, typename Context>
|
Chris@102
|
419 bool test(Char ch, Context const&) const
|
Chris@102
|
420 {
|
Chris@102
|
421 return ((sizeof(Char) <= sizeof(char_type)) || encoding::ischar(ch))
|
Chris@102
|
422 && unicode_char_class_base::is(tag(), ch);
|
Chris@102
|
423 }
|
Chris@102
|
424 };
|
Chris@102
|
425
|
Chris@102
|
426 #define BOOST_SPIRIT_X3_CHAR_CLASS(name) \
|
Chris@102
|
427 typedef unicode_char_class<name##_tag> name##_type; \
|
Chris@102
|
428 name##_type const name = name##_type(); \
|
Chris@102
|
429 /***/
|
Chris@102
|
430
|
Chris@102
|
431 namespace unicode
|
Chris@102
|
432 {
|
Chris@102
|
433 typedef any_char<char_encoding::unicode> char_type;
|
Chris@102
|
434 char_type const char_ = char_type();
|
Chris@102
|
435
|
Chris@102
|
436 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
437 // Unicode Major Categories
|
Chris@102
|
438 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
439 BOOST_SPIRIT_X3_CHAR_CLASS(alnum)
|
Chris@102
|
440 BOOST_SPIRIT_X3_CHAR_CLASS(alpha)
|
Chris@102
|
441 BOOST_SPIRIT_X3_CHAR_CLASS(digit)
|
Chris@102
|
442 BOOST_SPIRIT_X3_CHAR_CLASS(xdigit)
|
Chris@102
|
443 BOOST_SPIRIT_X3_CHAR_CLASS(cntrl)
|
Chris@102
|
444 BOOST_SPIRIT_X3_CHAR_CLASS(graph)
|
Chris@102
|
445 BOOST_SPIRIT_X3_CHAR_CLASS(lower)
|
Chris@102
|
446 BOOST_SPIRIT_X3_CHAR_CLASS(print)
|
Chris@102
|
447 BOOST_SPIRIT_X3_CHAR_CLASS(punct)
|
Chris@102
|
448 BOOST_SPIRIT_X3_CHAR_CLASS(space)
|
Chris@102
|
449 BOOST_SPIRIT_X3_CHAR_CLASS(blank)
|
Chris@102
|
450 BOOST_SPIRIT_X3_CHAR_CLASS(upper)
|
Chris@102
|
451
|
Chris@102
|
452 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
453 // Unicode Major Categories
|
Chris@102
|
454 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
455 BOOST_SPIRIT_X3_CHAR_CLASS(letter)
|
Chris@102
|
456 BOOST_SPIRIT_X3_CHAR_CLASS(mark)
|
Chris@102
|
457 BOOST_SPIRIT_X3_CHAR_CLASS(number)
|
Chris@102
|
458 BOOST_SPIRIT_X3_CHAR_CLASS(separator)
|
Chris@102
|
459 BOOST_SPIRIT_X3_CHAR_CLASS(other)
|
Chris@102
|
460 BOOST_SPIRIT_X3_CHAR_CLASS(punctuation)
|
Chris@102
|
461 BOOST_SPIRIT_X3_CHAR_CLASS(symbol)
|
Chris@102
|
462
|
Chris@102
|
463 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
464 // Unicode General Categories
|
Chris@102
|
465 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
466 BOOST_SPIRIT_X3_CHAR_CLASS(uppercase_letter)
|
Chris@102
|
467 BOOST_SPIRIT_X3_CHAR_CLASS(lowercase_letter)
|
Chris@102
|
468 BOOST_SPIRIT_X3_CHAR_CLASS(titlecase_letter)
|
Chris@102
|
469 BOOST_SPIRIT_X3_CHAR_CLASS(modifier_letter)
|
Chris@102
|
470 BOOST_SPIRIT_X3_CHAR_CLASS(other_letter)
|
Chris@102
|
471
|
Chris@102
|
472 BOOST_SPIRIT_X3_CHAR_CLASS(nonspacing_mark)
|
Chris@102
|
473 BOOST_SPIRIT_X3_CHAR_CLASS(enclosing_mark)
|
Chris@102
|
474 BOOST_SPIRIT_X3_CHAR_CLASS(spacing_mark)
|
Chris@102
|
475
|
Chris@102
|
476 BOOST_SPIRIT_X3_CHAR_CLASS(decimal_number)
|
Chris@102
|
477 BOOST_SPIRIT_X3_CHAR_CLASS(letter_number)
|
Chris@102
|
478 BOOST_SPIRIT_X3_CHAR_CLASS(other_number)
|
Chris@102
|
479
|
Chris@102
|
480 BOOST_SPIRIT_X3_CHAR_CLASS(space_separator)
|
Chris@102
|
481 BOOST_SPIRIT_X3_CHAR_CLASS(line_separator)
|
Chris@102
|
482 BOOST_SPIRIT_X3_CHAR_CLASS(paragraph_separator)
|
Chris@102
|
483
|
Chris@102
|
484 BOOST_SPIRIT_X3_CHAR_CLASS(control)
|
Chris@102
|
485 BOOST_SPIRIT_X3_CHAR_CLASS(format)
|
Chris@102
|
486 BOOST_SPIRIT_X3_CHAR_CLASS(private_use)
|
Chris@102
|
487 BOOST_SPIRIT_X3_CHAR_CLASS(surrogate)
|
Chris@102
|
488 BOOST_SPIRIT_X3_CHAR_CLASS(unassigned)
|
Chris@102
|
489
|
Chris@102
|
490 BOOST_SPIRIT_X3_CHAR_CLASS(dash_punctuation)
|
Chris@102
|
491 BOOST_SPIRIT_X3_CHAR_CLASS(open_punctuation)
|
Chris@102
|
492 BOOST_SPIRIT_X3_CHAR_CLASS(close_punctuation)
|
Chris@102
|
493 BOOST_SPIRIT_X3_CHAR_CLASS(connector_punctuation)
|
Chris@102
|
494 BOOST_SPIRIT_X3_CHAR_CLASS(other_punctuation)
|
Chris@102
|
495 BOOST_SPIRIT_X3_CHAR_CLASS(initial_punctuation)
|
Chris@102
|
496 BOOST_SPIRIT_X3_CHAR_CLASS(final_punctuation)
|
Chris@102
|
497
|
Chris@102
|
498 BOOST_SPIRIT_X3_CHAR_CLASS(math_symbol)
|
Chris@102
|
499 BOOST_SPIRIT_X3_CHAR_CLASS(currency_symbol)
|
Chris@102
|
500 BOOST_SPIRIT_X3_CHAR_CLASS(modifier_symbol)
|
Chris@102
|
501 BOOST_SPIRIT_X3_CHAR_CLASS(other_symbol)
|
Chris@102
|
502
|
Chris@102
|
503 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
504 // Unicode Derived Categories
|
Chris@102
|
505 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
506 BOOST_SPIRIT_X3_CHAR_CLASS(alphabetic)
|
Chris@102
|
507 BOOST_SPIRIT_X3_CHAR_CLASS(uppercase)
|
Chris@102
|
508 BOOST_SPIRIT_X3_CHAR_CLASS(lowercase)
|
Chris@102
|
509 BOOST_SPIRIT_X3_CHAR_CLASS(white_space)
|
Chris@102
|
510 BOOST_SPIRIT_X3_CHAR_CLASS(hex_digit)
|
Chris@102
|
511 BOOST_SPIRIT_X3_CHAR_CLASS(noncharacter_code_point)
|
Chris@102
|
512 BOOST_SPIRIT_X3_CHAR_CLASS(default_ignorable_code_point)
|
Chris@102
|
513
|
Chris@102
|
514 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
515 // Unicode Scripts
|
Chris@102
|
516 ///////////////////////////////////////////////////////////////////////////
|
Chris@102
|
517 BOOST_SPIRIT_X3_CHAR_CLASS(arabic)
|
Chris@102
|
518 BOOST_SPIRIT_X3_CHAR_CLASS(imperial_aramaic)
|
Chris@102
|
519 BOOST_SPIRIT_X3_CHAR_CLASS(armenian)
|
Chris@102
|
520 BOOST_SPIRIT_X3_CHAR_CLASS(avestan)
|
Chris@102
|
521 BOOST_SPIRIT_X3_CHAR_CLASS(balinese)
|
Chris@102
|
522 BOOST_SPIRIT_X3_CHAR_CLASS(bamum)
|
Chris@102
|
523 BOOST_SPIRIT_X3_CHAR_CLASS(bengali)
|
Chris@102
|
524 BOOST_SPIRIT_X3_CHAR_CLASS(bopomofo)
|
Chris@102
|
525 BOOST_SPIRIT_X3_CHAR_CLASS(braille)
|
Chris@102
|
526 BOOST_SPIRIT_X3_CHAR_CLASS(buginese)
|
Chris@102
|
527 BOOST_SPIRIT_X3_CHAR_CLASS(buhid)
|
Chris@102
|
528 BOOST_SPIRIT_X3_CHAR_CLASS(canadian_aboriginal)
|
Chris@102
|
529 BOOST_SPIRIT_X3_CHAR_CLASS(carian)
|
Chris@102
|
530 BOOST_SPIRIT_X3_CHAR_CLASS(cham)
|
Chris@102
|
531 BOOST_SPIRIT_X3_CHAR_CLASS(cherokee)
|
Chris@102
|
532 BOOST_SPIRIT_X3_CHAR_CLASS(coptic)
|
Chris@102
|
533 BOOST_SPIRIT_X3_CHAR_CLASS(cypriot)
|
Chris@102
|
534 BOOST_SPIRIT_X3_CHAR_CLASS(cyrillic)
|
Chris@102
|
535 BOOST_SPIRIT_X3_CHAR_CLASS(devanagari)
|
Chris@102
|
536 BOOST_SPIRIT_X3_CHAR_CLASS(deseret)
|
Chris@102
|
537 BOOST_SPIRIT_X3_CHAR_CLASS(egyptian_hieroglyphs)
|
Chris@102
|
538 BOOST_SPIRIT_X3_CHAR_CLASS(ethiopic)
|
Chris@102
|
539 BOOST_SPIRIT_X3_CHAR_CLASS(georgian)
|
Chris@102
|
540 BOOST_SPIRIT_X3_CHAR_CLASS(glagolitic)
|
Chris@102
|
541 BOOST_SPIRIT_X3_CHAR_CLASS(gothic)
|
Chris@102
|
542 BOOST_SPIRIT_X3_CHAR_CLASS(greek)
|
Chris@102
|
543 BOOST_SPIRIT_X3_CHAR_CLASS(gujarati)
|
Chris@102
|
544 BOOST_SPIRIT_X3_CHAR_CLASS(gurmukhi)
|
Chris@102
|
545 BOOST_SPIRIT_X3_CHAR_CLASS(hangul)
|
Chris@102
|
546 BOOST_SPIRIT_X3_CHAR_CLASS(han)
|
Chris@102
|
547 BOOST_SPIRIT_X3_CHAR_CLASS(hanunoo)
|
Chris@102
|
548 BOOST_SPIRIT_X3_CHAR_CLASS(hebrew)
|
Chris@102
|
549 BOOST_SPIRIT_X3_CHAR_CLASS(hiragana)
|
Chris@102
|
550 BOOST_SPIRIT_X3_CHAR_CLASS(katakana_or_hiragana)
|
Chris@102
|
551 BOOST_SPIRIT_X3_CHAR_CLASS(old_italic)
|
Chris@102
|
552 BOOST_SPIRIT_X3_CHAR_CLASS(javanese)
|
Chris@102
|
553 BOOST_SPIRIT_X3_CHAR_CLASS(kayah_li)
|
Chris@102
|
554 BOOST_SPIRIT_X3_CHAR_CLASS(katakana)
|
Chris@102
|
555 BOOST_SPIRIT_X3_CHAR_CLASS(kharoshthi)
|
Chris@102
|
556 BOOST_SPIRIT_X3_CHAR_CLASS(khmer)
|
Chris@102
|
557 BOOST_SPIRIT_X3_CHAR_CLASS(kannada)
|
Chris@102
|
558 BOOST_SPIRIT_X3_CHAR_CLASS(kaithi)
|
Chris@102
|
559 BOOST_SPIRIT_X3_CHAR_CLASS(tai_tham)
|
Chris@102
|
560 BOOST_SPIRIT_X3_CHAR_CLASS(lao)
|
Chris@102
|
561 BOOST_SPIRIT_X3_CHAR_CLASS(latin)
|
Chris@102
|
562 BOOST_SPIRIT_X3_CHAR_CLASS(lepcha)
|
Chris@102
|
563 BOOST_SPIRIT_X3_CHAR_CLASS(limbu)
|
Chris@102
|
564 BOOST_SPIRIT_X3_CHAR_CLASS(linear_b)
|
Chris@102
|
565 BOOST_SPIRIT_X3_CHAR_CLASS(lisu)
|
Chris@102
|
566 BOOST_SPIRIT_X3_CHAR_CLASS(lycian)
|
Chris@102
|
567 BOOST_SPIRIT_X3_CHAR_CLASS(lydian)
|
Chris@102
|
568 BOOST_SPIRIT_X3_CHAR_CLASS(malayalam)
|
Chris@102
|
569 BOOST_SPIRIT_X3_CHAR_CLASS(mongolian)
|
Chris@102
|
570 BOOST_SPIRIT_X3_CHAR_CLASS(meetei_mayek)
|
Chris@102
|
571 BOOST_SPIRIT_X3_CHAR_CLASS(myanmar)
|
Chris@102
|
572 BOOST_SPIRIT_X3_CHAR_CLASS(nko)
|
Chris@102
|
573 BOOST_SPIRIT_X3_CHAR_CLASS(ogham)
|
Chris@102
|
574 BOOST_SPIRIT_X3_CHAR_CLASS(ol_chiki)
|
Chris@102
|
575 BOOST_SPIRIT_X3_CHAR_CLASS(old_turkic)
|
Chris@102
|
576 BOOST_SPIRIT_X3_CHAR_CLASS(oriya)
|
Chris@102
|
577 BOOST_SPIRIT_X3_CHAR_CLASS(osmanya)
|
Chris@102
|
578 BOOST_SPIRIT_X3_CHAR_CLASS(phags_pa)
|
Chris@102
|
579 BOOST_SPIRIT_X3_CHAR_CLASS(inscriptional_pahlavi)
|
Chris@102
|
580 BOOST_SPIRIT_X3_CHAR_CLASS(phoenician)
|
Chris@102
|
581 BOOST_SPIRIT_X3_CHAR_CLASS(inscriptional_parthian)
|
Chris@102
|
582 BOOST_SPIRIT_X3_CHAR_CLASS(rejang)
|
Chris@102
|
583 BOOST_SPIRIT_X3_CHAR_CLASS(runic)
|
Chris@102
|
584 BOOST_SPIRIT_X3_CHAR_CLASS(samaritan)
|
Chris@102
|
585 BOOST_SPIRIT_X3_CHAR_CLASS(old_south_arabian)
|
Chris@102
|
586 BOOST_SPIRIT_X3_CHAR_CLASS(saurashtra)
|
Chris@102
|
587 BOOST_SPIRIT_X3_CHAR_CLASS(shavian)
|
Chris@102
|
588 BOOST_SPIRIT_X3_CHAR_CLASS(sinhala)
|
Chris@102
|
589 BOOST_SPIRIT_X3_CHAR_CLASS(sundanese)
|
Chris@102
|
590 BOOST_SPIRIT_X3_CHAR_CLASS(syloti_nagri)
|
Chris@102
|
591 BOOST_SPIRIT_X3_CHAR_CLASS(syriac)
|
Chris@102
|
592 BOOST_SPIRIT_X3_CHAR_CLASS(tagbanwa)
|
Chris@102
|
593 BOOST_SPIRIT_X3_CHAR_CLASS(tai_le)
|
Chris@102
|
594 BOOST_SPIRIT_X3_CHAR_CLASS(new_tai_lue)
|
Chris@102
|
595 BOOST_SPIRIT_X3_CHAR_CLASS(tamil)
|
Chris@102
|
596 BOOST_SPIRIT_X3_CHAR_CLASS(tai_viet)
|
Chris@102
|
597 BOOST_SPIRIT_X3_CHAR_CLASS(telugu)
|
Chris@102
|
598 BOOST_SPIRIT_X3_CHAR_CLASS(tifinagh)
|
Chris@102
|
599 BOOST_SPIRIT_X3_CHAR_CLASS(tagalog)
|
Chris@102
|
600 BOOST_SPIRIT_X3_CHAR_CLASS(thaana)
|
Chris@102
|
601 BOOST_SPIRIT_X3_CHAR_CLASS(thai)
|
Chris@102
|
602 BOOST_SPIRIT_X3_CHAR_CLASS(tibetan)
|
Chris@102
|
603 BOOST_SPIRIT_X3_CHAR_CLASS(ugaritic)
|
Chris@102
|
604 BOOST_SPIRIT_X3_CHAR_CLASS(vai)
|
Chris@102
|
605 BOOST_SPIRIT_X3_CHAR_CLASS(old_persian)
|
Chris@102
|
606 BOOST_SPIRIT_X3_CHAR_CLASS(cuneiform)
|
Chris@102
|
607 BOOST_SPIRIT_X3_CHAR_CLASS(yi)
|
Chris@102
|
608 BOOST_SPIRIT_X3_CHAR_CLASS(inherited)
|
Chris@102
|
609 BOOST_SPIRIT_X3_CHAR_CLASS(common)
|
Chris@102
|
610 BOOST_SPIRIT_X3_CHAR_CLASS(unknown)
|
Chris@102
|
611 }
|
Chris@102
|
612
|
Chris@102
|
613 #undef BOOST_SPIRIT_X3_CHAR_CLASS
|
Chris@102
|
614
|
Chris@102
|
615 }}}
|
Chris@102
|
616
|
Chris@102
|
617 #endif
|