1 | // -*- C++ -*- |
2 | //===----------------------------------------------------------------------===// |
3 | // |
4 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
5 | // See https://llvm.org/LICENSE.txt for license information. |
6 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
7 | // |
8 | //===----------------------------------------------------------------------===// |
9 | |
10 | #ifndef _LIBCPP___FORMAT_UNICODE_H |
11 | #define _LIBCPP___FORMAT_UNICODE_H |
12 | |
13 | #include <__assert> |
14 | #include <__bit/countl.h> |
15 | #include <__concepts/same_as.h> |
16 | #include <__config> |
17 | #include <__format/extended_grapheme_cluster_table.h> |
18 | #include <__format/indic_conjunct_break_table.h> |
19 | #include <__iterator/concepts.h> |
20 | #include <__iterator/readable_traits.h> // iter_value_t |
21 | #include <__utility/unreachable.h> |
22 | #include <string_view> |
23 | |
24 | #if !defined(_LIBCPP_HAS_NO_PRAGMA_SYSTEM_HEADER) |
25 | # pragma GCC system_header |
26 | #endif |
27 | |
28 | _LIBCPP_BEGIN_NAMESPACE_STD |
29 | |
30 | #if _LIBCPP_STD_VER >= 20 |
31 | |
32 | namespace __unicode { |
33 | |
34 | // Helper struct for the result of a consume operation. |
35 | // |
36 | // The status value for a correct code point is 0. This allows a valid value to |
37 | // be used without masking. |
38 | // When the decoding fails it know the number of code units affected. For the |
39 | // current use-cases that value is not needed, therefore it is not stored. |
40 | // The escape routine needs the number of code units for both a valid and |
41 | // invalid character and keeps track of it itself. Doing it in this result |
42 | // unconditionally would give some overhead when the value is unneeded. |
43 | struct __consume_result { |
44 | // When __status == __ok it contains the decoded code point. |
45 | // Else it contains the replacement character U+FFFD |
46 | char32_t __code_point : 31; |
47 | |
48 | enum : char32_t { |
49 | // Consumed a well-formed code point. |
50 | __ok = 0, |
51 | // Encountered invalid UTF-8 |
52 | __error = 1 |
53 | } __status : 1 {__ok}; |
54 | }; |
55 | static_assert(sizeof(__consume_result) == sizeof(char32_t)); |
56 | |
57 | # ifndef _LIBCPP_HAS_NO_UNICODE |
58 | |
59 | /// Implements the grapheme cluster boundary rules |
60 | /// |
61 | /// These rules are used to implement format's width estimation as stated in |
62 | /// [format.string.std]/11 |
63 | /// |
64 | /// The Standard refers to UAX \#29 for Unicode 12.0.0 |
65 | /// https://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundary_Rules |
66 | /// |
67 | /// The data tables used are |
68 | /// https://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakProperty.txt |
69 | /// https://www.unicode.org/Public/UCD/latest/ucd/emoji/emoji-data.txt |
70 | /// https://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakTest.txt (for testing only) |
71 | |
72 | inline constexpr char32_t __replacement_character = U'\ufffd'; |
73 | |
74 | // The error of a consume operation. |
75 | // |
76 | // This sets the code point to the replacement character. This code point does |
77 | // not participate in the grapheme clustering, so grapheme clustering code can |
78 | // ignore the error status and always use the code point. |
79 | inline constexpr __consume_result __consume_result_error{__replacement_character, __consume_result::__error}; |
80 | |
81 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool __is_high_surrogate(char32_t __value) { |
82 | return __value >= 0xd800 && __value <= 0xdbff; |
83 | } |
84 | |
85 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool __is_low_surrogate(char32_t __value) { |
86 | return __value >= 0xdc00 && __value <= 0xdfff; |
87 | } |
88 | |
89 | // https://www.unicode.org/glossary/#surrogate_code_point |
90 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI inline constexpr bool __is_surrogate(char32_t __value) { |
91 | return __value >= 0xd800 && __value <= 0xdfff; |
92 | } |
93 | |
94 | // https://www.unicode.org/glossary/#code_point |
95 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI inline constexpr bool __is_code_point(char32_t __value) { |
96 | return __value <= 0x10ffff; |
97 | } |
98 | |
99 | // https://www.unicode.org/glossary/#unicode_scalar_value |
100 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI inline constexpr bool __is_scalar_value(char32_t __value) { |
101 | return __unicode::__is_code_point(__value) && !__unicode::__is_surrogate(__value); |
102 | } |
103 | |
104 | template <contiguous_iterator _Iterator> |
105 | requires same_as<iter_value_t<_Iterator>, char> |
106 | _LIBCPP_HIDE_FROM_ABI constexpr bool __is_continuation(_Iterator __char, int __count) { |
107 | do { |
108 | if ((*__char & 0b1100'0000) != 0b1000'0000) |
109 | return false; |
110 | --__count; |
111 | ++__char; |
112 | } while (__count); |
113 | return true; |
114 | } |
115 | |
116 | /// Helper class to extract a code unit from a Unicode character range. |
117 | /// |
118 | /// The stored range is a view. There are multiple specialization for different |
119 | /// character types. |
120 | template <class _CharT> |
121 | class __code_point_view; |
122 | |
123 | /// UTF-8 specialization. |
124 | template <> |
125 | class __code_point_view<char> { |
126 | using _Iterator = basic_string_view<char>::const_iterator; |
127 | |
128 | public: |
129 | _LIBCPP_HIDE_FROM_ABI constexpr explicit __code_point_view(_Iterator __first, _Iterator __last) |
130 | : __first_(__first), __last_(__last) {} |
131 | |
132 | _LIBCPP_HIDE_FROM_ABI constexpr bool __at_end() const noexcept { return __first_ == __last_; } |
133 | _LIBCPP_HIDE_FROM_ABI constexpr _Iterator __position() const noexcept { return __first_; } |
134 | |
135 | // https://www.unicode.org/versions/latest/ch03.pdf#G7404 |
136 | // Based on Table 3-7, Well-Formed UTF-8 Byte Sequences |
137 | // |
138 | // Code Points First Byte Second Byte Third Byte Fourth Byte Remarks |
139 | // U+0000..U+007F 00..7F U+0000..U+007F 1 code unit range |
140 | // C0..C1 80..BF invalid overlong encoding |
141 | // U+0080..U+07FF C2..DF 80..BF U+0080..U+07FF 2 code unit range |
142 | // E0 80..9F 80..BF invalid overlong encoding |
143 | // U+0800..U+0FFF E0 A0..BF 80..BF U+0800..U+FFFF 3 code unit range |
144 | // U+1000..U+CFFF E1..EC 80..BF 80..BF |
145 | // U+D000..U+D7FF ED 80..9F 80..BF |
146 | // U+D800..U+DFFF ED A0..BF 80..BF invalid encoding of surrogate code point |
147 | // U+E000..U+FFFF EE..EF 80..BF 80..BF |
148 | // F0 80..8F 80..BF 80..BF invalid overlong encoding |
149 | // U+10000..U+3FFFF F0 90..BF 80..BF 80..BF U+10000..U+10FFFF 4 code unit range |
150 | // U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF |
151 | // U+100000..U+10FFFF F4 80..8F 80..BF 80..BF |
152 | // F4 90..BF 80..BF 80..BF U+110000.. invalid code point range |
153 | // |
154 | // Unlike other parsers, these invalid entries are tested after decoding. |
155 | // - The parser always needs to consume these code units |
156 | // - The code is optimized for well-formed UTF-8 |
157 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr __consume_result __consume() noexcept { |
158 | _LIBCPP_ASSERT_INTERNAL(__first_ != __last_, "can't move beyond the end of input" ); |
159 | |
160 | // Based on the number of leading 1 bits the number of code units in the |
161 | // code point can be determined. See |
162 | // https://en.wikipedia.org/wiki/UTF-8#Encoding |
163 | switch (std::countl_one(static_cast<unsigned char>(*__first_))) { |
164 | case 0: |
165 | return {static_cast<unsigned char>(*__first_++)}; |
166 | |
167 | case 2: { |
168 | if (__last_ - __first_ < 2 || !__unicode::__is_continuation(__first_ + 1, 1)) [[unlikely]] |
169 | break; |
170 | |
171 | char32_t __value = static_cast<unsigned char>(*__first_++) & 0x1f; |
172 | __value <<= 6; |
173 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
174 | |
175 | // These values should be encoded in 1 UTF-8 code unit. |
176 | if (__value < 0x0080) [[unlikely]] |
177 | return __consume_result_error; |
178 | |
179 | return {__value}; |
180 | } |
181 | |
182 | case 3: { |
183 | if (__last_ - __first_ < 3 || !__unicode::__is_continuation(__first_ + 1, 2)) [[unlikely]] |
184 | break; |
185 | |
186 | char32_t __value = static_cast<unsigned char>(*__first_++) & 0x0f; |
187 | __value <<= 6; |
188 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
189 | __value <<= 6; |
190 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
191 | |
192 | // These values should be encoded in 1 or 2 UTF-8 code units. |
193 | if (__value < 0x0800) [[unlikely]] |
194 | return __consume_result_error; |
195 | |
196 | // A surrogate value is always encoded in 3 UTF-8 code units. |
197 | if (__unicode::__is_surrogate(__value)) [[unlikely]] |
198 | return __consume_result_error; |
199 | |
200 | return {__value}; |
201 | } |
202 | |
203 | case 4: { |
204 | if (__last_ - __first_ < 4 || !__unicode::__is_continuation(__first_ + 1, 3)) [[unlikely]] |
205 | break; |
206 | |
207 | char32_t __value = static_cast<unsigned char>(*__first_++) & 0x07; |
208 | __value <<= 6; |
209 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
210 | __value <<= 6; |
211 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
212 | __value <<= 6; |
213 | __value |= static_cast<unsigned char>(*__first_++) & 0x3f; |
214 | |
215 | // These values should be encoded in 1, 2, or 3 UTF-8 code units. |
216 | if (__value < 0x10000) [[unlikely]] |
217 | return __consume_result_error; |
218 | |
219 | // A value too large is always encoded in 4 UTF-8 code units. |
220 | if (!__unicode::__is_code_point(__value)) [[unlikely]] |
221 | return __consume_result_error; |
222 | |
223 | return {__value}; |
224 | } |
225 | } |
226 | // An invalid number of leading ones can be garbage or a code unit in the |
227 | // middle of a code point. By consuming one code unit the parser may get |
228 | // "in sync" after a few code units. |
229 | ++__first_; |
230 | return __consume_result_error; |
231 | } |
232 | |
233 | private: |
234 | _Iterator __first_; |
235 | _Iterator __last_; |
236 | }; |
237 | |
238 | # ifndef _LIBCPP_HAS_NO_WIDE_CHARACTERS |
239 | _LIBCPP_HIDE_FROM_ABI constexpr bool __is_surrogate_pair_high(wchar_t __value) { |
240 | return __value >= 0xd800 && __value <= 0xdbff; |
241 | } |
242 | |
243 | _LIBCPP_HIDE_FROM_ABI constexpr bool __is_surrogate_pair_low(wchar_t __value) { |
244 | return __value >= 0xdc00 && __value <= 0xdfff; |
245 | } |
246 | |
247 | /// This specialization depends on the size of wchar_t |
248 | /// - 2 UTF-16 (for example Windows and AIX) |
249 | /// - 4 UTF-32 (for example Linux) |
250 | template <> |
251 | class __code_point_view<wchar_t> { |
252 | using _Iterator = typename basic_string_view<wchar_t>::const_iterator; |
253 | |
254 | public: |
255 | static_assert(sizeof(wchar_t) == 2 || sizeof(wchar_t) == 4, "sizeof(wchar_t) has a not implemented value" ); |
256 | |
257 | _LIBCPP_HIDE_FROM_ABI constexpr explicit __code_point_view(_Iterator __first, _Iterator __last) |
258 | : __first_(__first), __last_(__last) {} |
259 | |
260 | _LIBCPP_HIDE_FROM_ABI constexpr _Iterator __position() const noexcept { return __first_; } |
261 | _LIBCPP_HIDE_FROM_ABI constexpr bool __at_end() const noexcept { return __first_ == __last_; } |
262 | |
263 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr __consume_result __consume() noexcept { |
264 | _LIBCPP_ASSERT_INTERNAL(__first_ != __last_, "can't move beyond the end of input" ); |
265 | |
266 | char32_t __value = static_cast<char32_t>(*__first_++); |
267 | if constexpr (sizeof(wchar_t) == 2) { |
268 | if (__unicode::__is_low_surrogate(__value)) [[unlikely]] |
269 | return __consume_result_error; |
270 | |
271 | if (__unicode::__is_high_surrogate(__value)) { |
272 | if (__first_ == __last_ || !__unicode::__is_low_surrogate(static_cast<char32_t>(*__first_))) [[unlikely]] |
273 | return __consume_result_error; |
274 | |
275 | __value -= 0xd800; |
276 | __value <<= 10; |
277 | __value += static_cast<char32_t>(*__first_++) - 0xdc00; |
278 | __value += 0x10000; |
279 | |
280 | if (!__unicode::__is_code_point(__value)) [[unlikely]] |
281 | return __consume_result_error; |
282 | } |
283 | } else { |
284 | if (!__unicode::__is_scalar_value(__value)) [[unlikely]] |
285 | return __consume_result_error; |
286 | } |
287 | |
288 | return {__value}; |
289 | } |
290 | |
291 | private: |
292 | _Iterator __first_; |
293 | _Iterator __last_; |
294 | }; |
295 | # endif // _LIBCPP_HAS_NO_WIDE_CHARACTERS |
296 | |
297 | // State machine to implement the Extended Grapheme Cluster Boundary |
298 | // |
299 | // The exact rules may change between Unicode versions. |
300 | // This implements the extended rules see |
301 | // https://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundaries |
302 | class __extended_grapheme_cluster_break { |
303 | using __EGC_property = __extended_grapheme_custer_property_boundary::__property; |
304 | using __inCB_property = __indic_conjunct_break::__property; |
305 | |
306 | public: |
307 | _LIBCPP_HIDE_FROM_ABI constexpr explicit __extended_grapheme_cluster_break(char32_t __first_code_point) |
308 | : __prev_code_point_(__first_code_point), |
309 | __prev_property_(__extended_grapheme_custer_property_boundary::__get_property(__first_code_point)) { |
310 | // Initializes the active rule. |
311 | if (__prev_property_ == __EGC_property::__Extended_Pictographic) |
312 | __active_rule_ = __rule::__GB11_emoji; |
313 | else if (__prev_property_ == __EGC_property::__Regional_Indicator) |
314 | __active_rule_ = __rule::__GB12_GB13_regional_indicator; |
315 | else if (__indic_conjunct_break::__get_property(__first_code_point) == __inCB_property::__Consonant) |
316 | __active_rule_ = __rule::__GB9c_indic_conjunct_break; |
317 | } |
318 | |
319 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool operator()(char32_t __next_code_point) { |
320 | __EGC_property __next_property = __extended_grapheme_custer_property_boundary::__get_property(__next_code_point); |
321 | bool __result = __evaluate(__next_code_point, __next_property); |
322 | __prev_code_point_ = __next_code_point; |
323 | __prev_property_ = __next_property; |
324 | return __result; |
325 | } |
326 | |
327 | // The code point whose break propery are considered during the next |
328 | // evaluation cyle. |
329 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr char32_t __current_code_point() const { return __prev_code_point_; } |
330 | |
331 | private: |
332 | // The naming of the identifiers matches the Unicode standard. |
333 | // NOLINTBEGIN(readability-identifier-naming) |
334 | |
335 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool |
336 | __evaluate(char32_t __next_code_point, __EGC_property __next_property) { |
337 | switch (__active_rule_) { |
338 | case __rule::__none: |
339 | return __evaluate_none(__next_code_point, __next_property); |
340 | case __rule::__GB9c_indic_conjunct_break: |
341 | return __evaluate_GB9c_indic_conjunct_break(__next_code_point, __next_property); |
342 | case __rule::__GB11_emoji: |
343 | return __evaluate_GB11_emoji(__next_code_point, __next_property); |
344 | case __rule::__GB12_GB13_regional_indicator: |
345 | return __evaluate_GB12_GB13_regional_indicator(__next_code_point, __next_property); |
346 | } |
347 | __libcpp_unreachable(); |
348 | } |
349 | |
350 | _LIBCPP_HIDE_FROM_ABI constexpr bool __evaluate_none(char32_t __next_code_point, __EGC_property __next_property) { |
351 | // *** Break at the start and end of text, unless the text is empty. *** |
352 | |
353 | _LIBCPP_ASSERT_INTERNAL(__prev_property_ != __EGC_property::__sot, "should be handled in the constructor" ); // GB1 |
354 | _LIBCPP_ASSERT_INTERNAL(__prev_property_ != __EGC_property::__eot, "should be handled by our caller" ); // GB2 |
355 | |
356 | // *** Do not break between a CR and LF. Otherwise, break before and after controls. *** |
357 | if (__prev_property_ == __EGC_property::__CR && __next_property == __EGC_property::__LF) // GB3 |
358 | return false; |
359 | |
360 | if (__prev_property_ == __EGC_property::__Control || __prev_property_ == __EGC_property::__CR || |
361 | __prev_property_ == __EGC_property::__LF) // GB4 |
362 | return true; |
363 | |
364 | if (__next_property == __EGC_property::__Control || __next_property == __EGC_property::__CR || |
365 | __next_property == __EGC_property::__LF) // GB5 |
366 | return true; |
367 | |
368 | // *** Do not break Hangul syllable sequences. *** |
369 | if (__prev_property_ == __EGC_property::__L && |
370 | (__next_property == __EGC_property::__L || __next_property == __EGC_property::__V || |
371 | __next_property == __EGC_property::__LV || __next_property == __EGC_property::__LVT)) // GB6 |
372 | return false; |
373 | |
374 | if ((__prev_property_ == __EGC_property::__LV || __prev_property_ == __EGC_property::__V) && |
375 | (__next_property == __EGC_property::__V || __next_property == __EGC_property::__T)) // GB7 |
376 | return false; |
377 | |
378 | if ((__prev_property_ == __EGC_property::__LVT || __prev_property_ == __EGC_property::__T) && |
379 | __next_property == __EGC_property::__T) // GB8 |
380 | return false; |
381 | |
382 | // *** Do not break before extending characters or ZWJ. *** |
383 | if (__next_property == __EGC_property::__Extend || __next_property == __EGC_property::__ZWJ) |
384 | return false; // GB9 |
385 | |
386 | // *** Do not break before SpacingMarks, or after Prepend characters. *** |
387 | if (__next_property == __EGC_property::__SpacingMark) // GB9a |
388 | return false; |
389 | |
390 | if (__prev_property_ == __EGC_property::__Prepend) // GB9b |
391 | return false; |
392 | |
393 | // *** Do not break within certain combinations with Indic_Conjunct_Break (InCB)=Linker. *** |
394 | if (__indic_conjunct_break::__get_property(__next_code_point) == __inCB_property::__Consonant) { |
395 | __active_rule_ = __rule::__GB9c_indic_conjunct_break; |
396 | __GB9c_indic_conjunct_break_state_ = __GB9c_indic_conjunct_break_state::__Consonant; |
397 | return true; |
398 | } |
399 | |
400 | // *** Do not break within emoji modifier sequences or emoji zwj sequences. *** |
401 | if (__next_property == __EGC_property::__Extended_Pictographic) { |
402 | __active_rule_ = __rule::__GB11_emoji; |
403 | __GB11_emoji_state_ = __GB11_emoji_state::__Extended_Pictographic; |
404 | return true; |
405 | } |
406 | |
407 | // *** Do not break within emoji flag sequences *** |
408 | |
409 | // That is, do not break between regional indicator (RI) symbols if there |
410 | // is an odd number of RI characters before the break point. |
411 | if (__next_property == __EGC_property::__Regional_Indicator) { // GB12 + GB13 |
412 | __active_rule_ = __rule::__GB12_GB13_regional_indicator; |
413 | return true; |
414 | } |
415 | |
416 | // *** Otherwise, break everywhere. *** |
417 | return true; // GB999 |
418 | } |
419 | |
420 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool |
421 | __evaluate_GB9c_indic_conjunct_break(char32_t __next_code_point, __EGC_property __next_property) { |
422 | __inCB_property __break = __indic_conjunct_break::__get_property(__next_code_point); |
423 | if (__break == __inCB_property::__none) { |
424 | __active_rule_ = __rule::__none; |
425 | return __evaluate_none(__next_code_point, __next_property); |
426 | } |
427 | |
428 | switch (__GB9c_indic_conjunct_break_state_) { |
429 | case __GB9c_indic_conjunct_break_state::__Consonant: |
430 | if (__break == __inCB_property::__Extend) { |
431 | return false; |
432 | } |
433 | if (__break == __inCB_property::__Linker) { |
434 | __GB9c_indic_conjunct_break_state_ = __GB9c_indic_conjunct_break_state::__Linker; |
435 | return false; |
436 | } |
437 | __active_rule_ = __rule::__none; |
438 | return __evaluate_none(__next_code_point, __next_property); |
439 | |
440 | case __GB9c_indic_conjunct_break_state::__Linker: |
441 | if (__break == __inCB_property::__Extend) { |
442 | return false; |
443 | } |
444 | if (__break == __inCB_property::__Linker) { |
445 | return false; |
446 | } |
447 | if (__break == __inCB_property::__Consonant) { |
448 | __GB9c_indic_conjunct_break_state_ = __GB9c_indic_conjunct_break_state::__Consonant; |
449 | return false; |
450 | } |
451 | __active_rule_ = __rule::__none; |
452 | return __evaluate_none(__next_code_point, __next_property); |
453 | } |
454 | __libcpp_unreachable(); |
455 | } |
456 | |
457 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool |
458 | __evaluate_GB11_emoji(char32_t __next_code_point, __EGC_property __next_property) { |
459 | switch (__GB11_emoji_state_) { |
460 | case __GB11_emoji_state::__Extended_Pictographic: |
461 | if (__next_property == __EGC_property::__Extend) { |
462 | __GB11_emoji_state_ = __GB11_emoji_state::__Extend; |
463 | return false; |
464 | } |
465 | [[fallthrough]]; |
466 | case __GB11_emoji_state::__Extend: |
467 | if (__next_property == __EGC_property::__ZWJ) { |
468 | __GB11_emoji_state_ = __GB11_emoji_state::__ZWJ; |
469 | return false; |
470 | } |
471 | if (__next_property == __EGC_property::__Extend) |
472 | return false; |
473 | __active_rule_ = __rule::__none; |
474 | return __evaluate_none(__next_code_point, __next_property); |
475 | |
476 | case __GB11_emoji_state::__ZWJ: |
477 | if (__next_property == __EGC_property::__Extended_Pictographic) { |
478 | __GB11_emoji_state_ = __GB11_emoji_state::__Extended_Pictographic; |
479 | return false; |
480 | } |
481 | __active_rule_ = __rule::__none; |
482 | return __evaluate_none(__next_code_point, __next_property); |
483 | } |
484 | __libcpp_unreachable(); |
485 | } |
486 | |
487 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr bool |
488 | __evaluate_GB12_GB13_regional_indicator(char32_t __next_code_point, __EGC_property __next_property) { |
489 | __active_rule_ = __rule::__none; |
490 | if (__next_property == __EGC_property::__Regional_Indicator) |
491 | return false; |
492 | return __evaluate_none(__next_code_point, __next_property); |
493 | } |
494 | |
495 | char32_t __prev_code_point_; |
496 | __EGC_property __prev_property_; |
497 | |
498 | enum class __rule { |
499 | __none, |
500 | __GB9c_indic_conjunct_break, |
501 | __GB11_emoji, |
502 | __GB12_GB13_regional_indicator, |
503 | }; |
504 | __rule __active_rule_ = __rule::__none; |
505 | |
506 | enum class __GB11_emoji_state { |
507 | __Extended_Pictographic, |
508 | __Extend, |
509 | __ZWJ, |
510 | }; |
511 | __GB11_emoji_state __GB11_emoji_state_ = __GB11_emoji_state::__Extended_Pictographic; |
512 | |
513 | enum class __GB9c_indic_conjunct_break_state { |
514 | __Consonant, |
515 | __Linker, |
516 | }; |
517 | |
518 | __GB9c_indic_conjunct_break_state __GB9c_indic_conjunct_break_state_ = __GB9c_indic_conjunct_break_state::__Consonant; |
519 | |
520 | // NOLINTEND(readability-identifier-naming) |
521 | }; |
522 | |
523 | /// Helper class to extract an extended grapheme cluster from a Unicode character range. |
524 | /// |
525 | /// This function is used to determine the column width of an extended grapheme |
526 | /// cluster. In order to do that only the first code point is evaluated. |
527 | /// Therefore only this code point is extracted. |
528 | template <class _CharT> |
529 | class __extended_grapheme_cluster_view { |
530 | using _Iterator = typename basic_string_view<_CharT>::const_iterator; |
531 | |
532 | public: |
533 | _LIBCPP_HIDE_FROM_ABI constexpr explicit __extended_grapheme_cluster_view(_Iterator __first, _Iterator __last) |
534 | : __code_point_view_(__first, __last), __at_break_(__code_point_view_.__consume().__code_point) {} |
535 | |
536 | struct __cluster { |
537 | /// The first code point of the extended grapheme cluster. |
538 | /// |
539 | /// The first code point is used to estimate the width of the extended |
540 | /// grapheme cluster. |
541 | char32_t __code_point_; |
542 | |
543 | /// Points one beyond the last code unit in the extended grapheme cluster. |
544 | /// |
545 | /// It's expected the caller has the start position and thus can determine |
546 | /// the code unit range of the extended grapheme cluster. |
547 | _Iterator __last_; |
548 | }; |
549 | |
550 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr __cluster __consume() { |
551 | char32_t __code_point = __at_break_.__current_code_point(); |
552 | _Iterator __position = __code_point_view_.__position(); |
553 | while (!__code_point_view_.__at_end()) { |
554 | if (__at_break_(__code_point_view_.__consume().__code_point)) |
555 | break; |
556 | __position = __code_point_view_.__position(); |
557 | } |
558 | return {__code_point, __position}; |
559 | } |
560 | |
561 | private: |
562 | __code_point_view<_CharT> __code_point_view_; |
563 | __extended_grapheme_cluster_break __at_break_; |
564 | }; |
565 | |
566 | template <contiguous_iterator _Iterator> |
567 | __extended_grapheme_cluster_view(_Iterator, _Iterator) -> __extended_grapheme_cluster_view<iter_value_t<_Iterator>>; |
568 | |
569 | # else // _LIBCPP_HAS_NO_UNICODE |
570 | |
571 | // For ASCII every character is a "code point". |
572 | // This makes it easier to write code agnostic of the _LIBCPP_HAS_NO_UNICODE define. |
573 | template <class _CharT> |
574 | class __code_point_view { |
575 | using _Iterator = typename basic_string_view<_CharT>::const_iterator; |
576 | |
577 | public: |
578 | _LIBCPP_HIDE_FROM_ABI constexpr explicit __code_point_view(_Iterator __first, _Iterator __last) |
579 | : __first_(__first), __last_(__last) {} |
580 | |
581 | _LIBCPP_HIDE_FROM_ABI constexpr bool __at_end() const noexcept { return __first_ == __last_; } |
582 | _LIBCPP_HIDE_FROM_ABI constexpr _Iterator __position() const noexcept { return __first_; } |
583 | |
584 | [[nodiscard]] _LIBCPP_HIDE_FROM_ABI constexpr __consume_result __consume() noexcept { |
585 | _LIBCPP_ASSERT_INTERNAL(__first_ != __last_, "can't move beyond the end of input" ); |
586 | return {static_cast<char32_t>(*__first_++)}; |
587 | } |
588 | |
589 | private: |
590 | _Iterator __first_; |
591 | _Iterator __last_; |
592 | }; |
593 | |
594 | # endif // _LIBCPP_HAS_NO_UNICODE |
595 | |
596 | } // namespace __unicode |
597 | |
598 | #endif //_LIBCPP_STD_VER >= 20 |
599 | |
600 | _LIBCPP_END_NAMESPACE_STD |
601 | |
602 | #endif // _LIBCPP___FORMAT_UNICODE_H |
603 | |