icu_properties/
provider.rs

1// This file is part of ICU4X. For terms of use, please see the file
2// called LICENSE at the top level of the ICU4X source tree
3// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
4
5// Provider structs must be stable
6#![allow(clippy::exhaustive_structs, clippy::exhaustive_enums)]
7
8//! 🚧 \[Unstable\] Data provider struct definitions for this ICU4X component.
9//!
10//! <div class="stab unstable">
11//! 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
12//! including in SemVer minor releases. While the serde representation of data structs is guaranteed
13//! to be stable, their Rust representation might not be. Use with caution.
14//! </div>
15//!
16//! Read more about data providers: [`icu_provider`]
17
18pub mod names;
19
20#[cfg(feature = "alloc")]
21pub use names::{
22    PropertyNameLongCanonicalCombiningClassV1, PropertyNameShortCanonicalCombiningClassV1,
23};
24
25pub use names::{
26    PropertyNameLongBidiClassV1, PropertyNameLongEastAsianWidthV1,
27    PropertyNameLongGeneralCategoryV1, PropertyNameLongGraphemeClusterBreakV1,
28    PropertyNameLongHangulSyllableTypeV1, PropertyNameLongIndicConjunctBreakV1,
29    PropertyNameLongIndicSyllabicCategoryV1, PropertyNameLongJoiningGroupV1,
30    PropertyNameLongJoiningTypeV1, PropertyNameLongLineBreakV1, PropertyNameLongNumericTypeV1,
31    PropertyNameLongScriptV1, PropertyNameLongSentenceBreakV1,
32    PropertyNameLongVerticalOrientationV1, PropertyNameLongWordBreakV1,
33    PropertyNameParseBidiClassV1, PropertyNameParseCanonicalCombiningClassV1,
34    PropertyNameParseEastAsianWidthV1, PropertyNameParseGeneralCategoryMaskV1,
35    PropertyNameParseGeneralCategoryV1, PropertyNameParseGraphemeClusterBreakV1,
36    PropertyNameParseHangulSyllableTypeV1, PropertyNameParseIndicConjunctBreakV1,
37    PropertyNameParseIndicSyllabicCategoryV1, PropertyNameParseJoiningGroupV1,
38    PropertyNameParseJoiningTypeV1, PropertyNameParseLineBreakV1, PropertyNameParseNumericTypeV1,
39    PropertyNameParseScriptV1, PropertyNameParseSentenceBreakV1,
40    PropertyNameParseVerticalOrientationV1, PropertyNameParseWordBreakV1,
41    PropertyNameShortBidiClassV1, PropertyNameShortEastAsianWidthV1,
42    PropertyNameShortGeneralCategoryV1, PropertyNameShortGraphemeClusterBreakV1,
43    PropertyNameShortHangulSyllableTypeV1, PropertyNameShortIndicConjunctBreakV1,
44    PropertyNameShortIndicSyllabicCategoryV1, PropertyNameShortJoiningGroupV1,
45    PropertyNameShortJoiningTypeV1, PropertyNameShortLineBreakV1, PropertyNameShortNumericTypeV1,
46    PropertyNameShortScriptV1, PropertyNameShortSentenceBreakV1,
47    PropertyNameShortVerticalOrientationV1, PropertyNameShortWordBreakV1,
48};
49
50pub use crate::props::gc::GeneralCategoryULE;
51use crate::props::*;
52use crate::script::ScriptWithExt;
53use core::ops::RangeInclusive;
54use icu_collections::codepointinvlist::CodePointInversionList;
55use icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList;
56use icu_collections::codepointtrie::{CodePointMapRange, CodePointTrie, TrieValue};
57use icu_provider::prelude::*;
58use zerofrom::ZeroFrom;
59use zerovec::{VarZeroVec, ZeroSlice};
60
61#[cfg(feature = "compiled_data")]
62#[derive(Debug)]
63/// Baked data
64///
65/// <div class="stab unstable">
66/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
67/// including in SemVer minor releases. In particular, the `DataProvider` implementations are only
68/// guaranteed to match with this version's `*_unstable` providers. Use with caution.
69/// </div>
70pub struct Baked;
71
72#[cfg(feature = "compiled_data")]
73#[allow(unused_imports)]
74const _: () = {
75    use icu_properties_data::*;
76    pub mod icu {
77        pub use crate as properties;
78        pub use icu_collections as collections;
79    }
80    make_provider!(Baked);
81    impl_property_binary_alnum_v1!(Baked);
82    impl_property_binary_alphabetic_v1!(Baked);
83    impl_property_binary_ascii_hex_digit_v1!(Baked);
84    impl_property_binary_basic_emoji_v1!(Baked);
85    impl_property_binary_bidi_control_v1!(Baked);
86    impl_property_binary_bidi_mirrored_v1!(Baked);
87    impl_property_binary_blank_v1!(Baked);
88    impl_property_binary_case_ignorable_v1!(Baked);
89    impl_property_binary_case_sensitive_v1!(Baked);
90    impl_property_binary_cased_v1!(Baked);
91    impl_property_binary_changes_when_casefolded_v1!(Baked);
92    impl_property_binary_changes_when_casemapped_v1!(Baked);
93    impl_property_binary_changes_when_lowercased_v1!(Baked);
94    impl_property_binary_changes_when_nfkc_casefolded_v1!(Baked);
95    impl_property_binary_changes_when_titlecased_v1!(Baked);
96    impl_property_binary_changes_when_uppercased_v1!(Baked);
97    impl_property_binary_dash_v1!(Baked);
98    impl_property_binary_default_ignorable_code_point_v1!(Baked);
99    impl_property_binary_deprecated_v1!(Baked);
100    impl_property_binary_diacritic_v1!(Baked);
101    impl_property_binary_emoji_component_v1!(Baked);
102    impl_property_binary_emoji_modifier_base_v1!(Baked);
103    impl_property_binary_emoji_modifier_v1!(Baked);
104    impl_property_binary_emoji_presentation_v1!(Baked);
105    impl_property_binary_emoji_v1!(Baked);
106    impl_property_binary_extended_pictographic_v1!(Baked);
107    impl_property_binary_extender_v1!(Baked);
108    impl_property_binary_full_composition_exclusion_v1!(Baked);
109    impl_property_binary_graph_v1!(Baked);
110    impl_property_binary_grapheme_base_v1!(Baked);
111    impl_property_binary_grapheme_extend_v1!(Baked);
112    impl_property_binary_grapheme_link_v1!(Baked);
113    impl_property_binary_hex_digit_v1!(Baked);
114    impl_property_binary_hyphen_v1!(Baked);
115    impl_property_binary_id_compat_math_continue_v1!(Baked);
116    impl_property_binary_id_compat_math_start_v1!(Baked);
117    impl_property_binary_id_continue_v1!(Baked);
118    impl_property_binary_id_start_v1!(Baked);
119    impl_property_binary_ideographic_v1!(Baked);
120    impl_property_binary_ids_binary_operator_v1!(Baked);
121    impl_property_binary_ids_trinary_operator_v1!(Baked);
122    impl_property_binary_ids_unary_operator_v1!(Baked);
123    impl_property_binary_join_control_v1!(Baked);
124    impl_property_binary_logical_order_exception_v1!(Baked);
125    impl_property_binary_lowercase_v1!(Baked);
126    impl_property_binary_math_v1!(Baked);
127    impl_property_binary_modifier_combining_mark_v1!(Baked);
128    impl_property_binary_nfc_inert_v1!(Baked);
129    impl_property_binary_nfd_inert_v1!(Baked);
130    impl_property_binary_nfkc_inert_v1!(Baked);
131    impl_property_binary_nfkd_inert_v1!(Baked);
132    impl_property_binary_noncharacter_code_point_v1!(Baked);
133    impl_property_binary_pattern_syntax_v1!(Baked);
134    impl_property_binary_pattern_white_space_v1!(Baked);
135    impl_property_binary_prepended_concatenation_mark_v1!(Baked);
136    impl_property_binary_print_v1!(Baked);
137    impl_property_binary_quotation_mark_v1!(Baked);
138    impl_property_binary_radical_v1!(Baked);
139    impl_property_binary_regional_indicator_v1!(Baked);
140    impl_property_binary_segment_starter_v1!(Baked);
141    impl_property_binary_sentence_terminal_v1!(Baked);
142    impl_property_binary_soft_dotted_v1!(Baked);
143    impl_property_binary_terminal_punctuation_v1!(Baked);
144    impl_property_binary_unified_ideograph_v1!(Baked);
145    impl_property_binary_uppercase_v1!(Baked);
146    impl_property_binary_variation_selector_v1!(Baked);
147    impl_property_binary_white_space_v1!(Baked);
148    impl_property_binary_xdigit_v1!(Baked);
149    impl_property_binary_xid_continue_v1!(Baked);
150    impl_property_binary_xid_start_v1!(Baked);
151    impl_property_enum_bidi_class_v1!(Baked);
152    impl_property_enum_bidi_mirroring_glyph_v1!(Baked);
153    impl_property_enum_canonical_combining_class_v1!(Baked);
154    impl_property_enum_east_asian_width_v1!(Baked);
155    impl_property_enum_general_category_v1!(Baked);
156    impl_property_enum_grapheme_cluster_break_v1!(Baked);
157    impl_property_enum_hangul_syllable_type_v1!(Baked);
158    impl_property_enum_indic_conjunct_break_v1!(Baked);
159    impl_property_enum_indic_syllabic_category_v1!(Baked);
160    impl_property_enum_joining_group_v1!(Baked);
161    impl_property_enum_joining_type_v1!(Baked);
162    impl_property_enum_line_break_v1!(Baked);
163    impl_property_enum_numeric_type_v1!(Baked);
164    impl_property_enum_script_v1!(Baked);
165    impl_property_enum_sentence_break_v1!(Baked);
166    impl_property_enum_vertical_orientation_v1!(Baked);
167    impl_property_enum_word_break_v1!(Baked);
168    impl_property_name_long_bidi_class_v1!(Baked);
169    #[cfg(feature = "alloc")]
170    impl_property_name_long_canonical_combining_class_v1!(Baked);
171    impl_property_name_long_east_asian_width_v1!(Baked);
172    impl_property_name_long_general_category_v1!(Baked);
173    impl_property_name_long_grapheme_cluster_break_v1!(Baked);
174    impl_property_name_long_hangul_syllable_type_v1!(Baked);
175    impl_property_name_long_indic_syllabic_category_v1!(Baked);
176    impl_property_name_long_indic_conjunct_break_v1!(Baked);
177    impl_property_name_long_joining_group_v1!(Baked);
178    impl_property_name_long_joining_type_v1!(Baked);
179    impl_property_name_long_line_break_v1!(Baked);
180    impl_property_name_long_numeric_type_v1!(Baked);
181    impl_property_name_long_script_v1!(Baked);
182    impl_property_name_long_sentence_break_v1!(Baked);
183    impl_property_name_long_vertical_orientation_v1!(Baked);
184    impl_property_name_long_word_break_v1!(Baked);
185    impl_property_name_parse_bidi_class_v1!(Baked);
186    impl_property_name_parse_canonical_combining_class_v1!(Baked);
187    impl_property_name_parse_east_asian_width_v1!(Baked);
188    impl_property_name_parse_general_category_mask_v1!(Baked);
189    impl_property_name_parse_general_category_v1!(Baked);
190    impl_property_name_parse_grapheme_cluster_break_v1!(Baked);
191    impl_property_name_parse_hangul_syllable_type_v1!(Baked);
192    impl_property_name_parse_indic_syllabic_category_v1!(Baked);
193    impl_property_name_parse_indic_conjunct_break_v1!(Baked);
194    impl_property_name_parse_joining_group_v1!(Baked);
195    impl_property_name_parse_joining_type_v1!(Baked);
196    impl_property_name_parse_line_break_v1!(Baked);
197    impl_property_name_parse_numeric_type_v1!(Baked);
198    impl_property_name_parse_script_v1!(Baked);
199    impl_property_name_parse_sentence_break_v1!(Baked);
200    impl_property_name_parse_vertical_orientation_v1!(Baked);
201    impl_property_name_parse_word_break_v1!(Baked);
202    impl_property_name_short_bidi_class_v1!(Baked);
203    #[cfg(feature = "alloc")]
204    impl_property_name_short_canonical_combining_class_v1!(Baked);
205    impl_property_name_short_east_asian_width_v1!(Baked);
206    impl_property_name_short_general_category_v1!(Baked);
207    impl_property_name_short_grapheme_cluster_break_v1!(Baked);
208    impl_property_name_short_hangul_syllable_type_v1!(Baked);
209    impl_property_name_short_indic_syllabic_category_v1!(Baked);
210    impl_property_name_short_indic_conjunct_break_v1!(Baked);
211    impl_property_name_short_joining_group_v1!(Baked);
212    impl_property_name_short_joining_type_v1!(Baked);
213    impl_property_name_short_line_break_v1!(Baked);
214    impl_property_name_short_numeric_type_v1!(Baked);
215    impl_property_name_short_script_v1!(Baked);
216    impl_property_name_short_sentence_break_v1!(Baked);
217    impl_property_name_short_vertical_orientation_v1!(Baked);
218    impl_property_name_short_word_break_v1!(Baked);
219    impl_property_script_with_extensions_v1!(Baked);
220};
221
222icu_provider::data_marker!(
223    /// `PropertyBinaryAlnumV1`
224    PropertyBinaryAlnumV1,
225    PropertyCodePointSet<'static>,
226    is_singleton = true
227);
228icu_provider::data_marker!(
229    /// `PropertyBinaryAlphabeticV1`
230    PropertyBinaryAlphabeticV1,
231    PropertyCodePointSet<'static>,
232    is_singleton = true
233);
234icu_provider::data_marker!(
235    /// `PropertyBinaryAsciiHexDigitV1`
236    PropertyBinaryAsciiHexDigitV1,
237    PropertyCodePointSet<'static>,
238    is_singleton = true
239);
240icu_provider::data_marker!(
241    /// `PropertyBinaryBidiControlV1`
242    PropertyBinaryBidiControlV1,
243    PropertyCodePointSet<'static>,
244    is_singleton = true
245);
246icu_provider::data_marker!(
247    /// `PropertyBinaryBidiMirroredV1`
248    PropertyBinaryBidiMirroredV1,
249    PropertyCodePointSet<'static>,
250    is_singleton = true
251);
252icu_provider::data_marker!(
253    /// `PropertyBinaryBlankV1`
254    PropertyBinaryBlankV1,
255    PropertyCodePointSet<'static>,
256    is_singleton = true
257);
258icu_provider::data_marker!(
259    /// `PropertyBinaryCasedV1`
260    PropertyBinaryCasedV1,
261    PropertyCodePointSet<'static>,
262    is_singleton = true
263);
264icu_provider::data_marker!(
265    /// `PropertyBinaryCaseIgnorableV1`
266    PropertyBinaryCaseIgnorableV1,
267    PropertyCodePointSet<'static>,
268    is_singleton = true
269);
270icu_provider::data_marker!(
271    /// `PropertyBinaryCaseSensitiveV1`
272    PropertyBinaryCaseSensitiveV1,
273    PropertyCodePointSet<'static>,
274    is_singleton = true
275);
276icu_provider::data_marker!(
277    /// `PropertyBinaryChangesWhenCasefoldedV1`
278    PropertyBinaryChangesWhenCasefoldedV1,
279    PropertyCodePointSet<'static>,
280    is_singleton = true
281);
282icu_provider::data_marker!(
283    /// `PropertyBinaryChangesWhenCasemappedV1`
284    PropertyBinaryChangesWhenCasemappedV1,
285    PropertyCodePointSet<'static>,
286    is_singleton = true
287);
288icu_provider::data_marker!(
289    /// `PropertyBinaryChangesWhenLowercasedV1`
290    PropertyBinaryChangesWhenLowercasedV1,
291    PropertyCodePointSet<'static>,
292    is_singleton = true
293);
294icu_provider::data_marker!(
295    /// `PropertyBinaryChangesWhenNfkcCasefoldedV1`
296    PropertyBinaryChangesWhenNfkcCasefoldedV1,
297    PropertyCodePointSet<'static>,
298    is_singleton = true
299);
300icu_provider::data_marker!(
301    /// `PropertyBinaryChangesWhenTitlecasedV1`
302    PropertyBinaryChangesWhenTitlecasedV1,
303    PropertyCodePointSet<'static>,
304    is_singleton = true
305);
306icu_provider::data_marker!(
307    /// `PropertyBinaryChangesWhenUppercasedV1`
308    PropertyBinaryChangesWhenUppercasedV1,
309    PropertyCodePointSet<'static>,
310    is_singleton = true
311);
312icu_provider::data_marker!(
313    /// `PropertyBinaryDashV1`
314    PropertyBinaryDashV1,
315    PropertyCodePointSet<'static>,
316    is_singleton = true
317);
318icu_provider::data_marker!(
319    /// `PropertyBinaryDefaultIgnorableCodePointV1`
320    PropertyBinaryDefaultIgnorableCodePointV1,
321    PropertyCodePointSet<'static>,
322    is_singleton = true
323);
324icu_provider::data_marker!(
325    /// `PropertyBinaryDeprecatedV1`
326    PropertyBinaryDeprecatedV1,
327    PropertyCodePointSet<'static>,
328    is_singleton = true
329);
330icu_provider::data_marker!(
331    /// `PropertyBinaryDiacriticV1`
332    PropertyBinaryDiacriticV1,
333    PropertyCodePointSet<'static>,
334    is_singleton = true
335);
336icu_provider::data_marker!(
337    /// `PropertyBinaryEmojiComponentV1`
338    PropertyBinaryEmojiComponentV1,
339    PropertyCodePointSet<'static>,
340    is_singleton = true
341);
342icu_provider::data_marker!(
343    /// `PropertyBinaryEmojiModifierBaseV1`
344    PropertyBinaryEmojiModifierBaseV1,
345    PropertyCodePointSet<'static>,
346    is_singleton = true
347);
348icu_provider::data_marker!(
349    /// `PropertyBinaryEmojiModifierV1`
350    PropertyBinaryEmojiModifierV1,
351    PropertyCodePointSet<'static>,
352    is_singleton = true
353);
354icu_provider::data_marker!(
355    /// `PropertyBinaryEmojiPresentationV1`
356    PropertyBinaryEmojiPresentationV1,
357    PropertyCodePointSet<'static>,
358    is_singleton = true
359);
360icu_provider::data_marker!(
361    /// `PropertyBinaryEmojiV1`
362    PropertyBinaryEmojiV1,
363    PropertyCodePointSet<'static>,
364    is_singleton = true
365);
366icu_provider::data_marker!(
367    /// `PropertyBinaryExtendedPictographicV1`
368    PropertyBinaryExtendedPictographicV1,
369    PropertyCodePointSet<'static>,
370    is_singleton = true
371);
372icu_provider::data_marker!(
373    /// `PropertyBinaryExtenderV1`
374    PropertyBinaryExtenderV1,
375    PropertyCodePointSet<'static>,
376    is_singleton = true
377);
378icu_provider::data_marker!(
379    /// `PropertyBinaryFullCompositionExclusionV1`
380    PropertyBinaryFullCompositionExclusionV1,
381    PropertyCodePointSet<'static>,
382    is_singleton = true
383);
384icu_provider::data_marker!(
385    /// `PropertyBinaryGraphemeBaseV1`
386    PropertyBinaryGraphemeBaseV1,
387    PropertyCodePointSet<'static>,
388    is_singleton = true
389);
390icu_provider::data_marker!(
391    /// `PropertyBinaryGraphemeExtendV1`
392    PropertyBinaryGraphemeExtendV1,
393    PropertyCodePointSet<'static>,
394    is_singleton = true
395);
396icu_provider::data_marker!(
397    /// `PropertyBinaryGraphemeLinkV1`
398    PropertyBinaryGraphemeLinkV1,
399    PropertyCodePointSet<'static>,
400    is_singleton = true
401);
402icu_provider::data_marker!(
403    /// `PropertyBinaryGraphV1`
404    PropertyBinaryGraphV1,
405    PropertyCodePointSet<'static>,
406    is_singleton = true
407);
408icu_provider::data_marker!(
409    /// `PropertyBinaryHexDigitV1`
410    PropertyBinaryHexDigitV1,
411    PropertyCodePointSet<'static>,
412    is_singleton = true
413);
414icu_provider::data_marker!(
415    /// `PropertyBinaryHyphenV1`
416    PropertyBinaryHyphenV1,
417    PropertyCodePointSet<'static>,
418    is_singleton = true
419);
420icu_provider::data_marker!(
421    /// `PropertyBinaryIdCompatMathContinueV1`
422    PropertyBinaryIdCompatMathContinueV1,
423    PropertyCodePointSet<'static>,
424    is_singleton = true
425);
426icu_provider::data_marker!(
427    /// `PropertyBinaryIdCompatMathStartV1`
428    PropertyBinaryIdCompatMathStartV1,
429    PropertyCodePointSet<'static>,
430    is_singleton = true
431);
432icu_provider::data_marker!(
433    /// `PropertyBinaryIdContinueV1`
434    PropertyBinaryIdContinueV1,
435    PropertyCodePointSet<'static>,
436    is_singleton = true
437);
438icu_provider::data_marker!(
439    /// `PropertyBinaryIdeographicV1`
440    PropertyBinaryIdeographicV1,
441    PropertyCodePointSet<'static>,
442    is_singleton = true
443);
444icu_provider::data_marker!(
445    /// `PropertyBinaryIdsBinaryOperatorV1`
446    PropertyBinaryIdsBinaryOperatorV1,
447    PropertyCodePointSet<'static>,
448    is_singleton = true
449);
450icu_provider::data_marker!(
451    /// `PropertyBinaryIdStartV1`
452    PropertyBinaryIdStartV1,
453    PropertyCodePointSet<'static>,
454    is_singleton = true
455);
456icu_provider::data_marker!(
457    /// `PropertyBinaryIdsTrinaryOperatorV1`
458    PropertyBinaryIdsTrinaryOperatorV1,
459    PropertyCodePointSet<'static>,
460    is_singleton = true
461);
462icu_provider::data_marker!(
463    /// `PropertyBinaryIdsUnaryOperatorV1`
464    PropertyBinaryIdsUnaryOperatorV1,
465    PropertyCodePointSet<'static>,
466    is_singleton = true
467);
468icu_provider::data_marker!(
469    /// `PropertyBinaryJoinControlV1`
470    PropertyBinaryJoinControlV1,
471    PropertyCodePointSet<'static>,
472    is_singleton = true
473);
474icu_provider::data_marker!(
475    /// `PropertyBinaryLogicalOrderExceptionV1`
476    PropertyBinaryLogicalOrderExceptionV1,
477    PropertyCodePointSet<'static>,
478    is_singleton = true
479);
480icu_provider::data_marker!(
481    /// `PropertyBinaryLowercaseV1`
482    PropertyBinaryLowercaseV1,
483    PropertyCodePointSet<'static>,
484    is_singleton = true
485);
486icu_provider::data_marker!(
487    /// `PropertyBinaryMathV1`
488    PropertyBinaryMathV1,
489    PropertyCodePointSet<'static>,
490    is_singleton = true
491);
492icu_provider::data_marker!(
493    /// `PropertyBinaryModifierCombiningMarkV1`
494    PropertyBinaryModifierCombiningMarkV1,
495    PropertyCodePointSet<'static>,
496    is_singleton = true
497);
498icu_provider::data_marker!(
499    /// `PropertyBinaryNfcInertV1`
500    PropertyBinaryNfcInertV1,
501    PropertyCodePointSet<'static>,
502    is_singleton = true
503);
504icu_provider::data_marker!(
505    /// `PropertyBinaryNfdInertV1`
506    PropertyBinaryNfdInertV1,
507    PropertyCodePointSet<'static>,
508    is_singleton = true
509);
510icu_provider::data_marker!(
511    /// `PropertyBinaryNfkcInertV1`
512    PropertyBinaryNfkcInertV1,
513    PropertyCodePointSet<'static>,
514    is_singleton = true
515);
516icu_provider::data_marker!(
517    /// `PropertyBinaryNfkdInertV1`
518    PropertyBinaryNfkdInertV1,
519    PropertyCodePointSet<'static>,
520    is_singleton = true
521);
522icu_provider::data_marker!(
523    /// `PropertyBinaryNoncharacterCodePointV1`
524    PropertyBinaryNoncharacterCodePointV1,
525    PropertyCodePointSet<'static>,
526    is_singleton = true
527);
528icu_provider::data_marker!(
529    /// `PropertyBinaryPatternSyntaxV1`
530    PropertyBinaryPatternSyntaxV1,
531    PropertyCodePointSet<'static>,
532    is_singleton = true
533);
534icu_provider::data_marker!(
535    /// `PropertyBinaryPatternWhiteSpaceV1`
536    PropertyBinaryPatternWhiteSpaceV1,
537    PropertyCodePointSet<'static>,
538    is_singleton = true
539);
540icu_provider::data_marker!(
541    /// `PropertyBinaryPrependedConcatenationMarkV1`
542    PropertyBinaryPrependedConcatenationMarkV1,
543    PropertyCodePointSet<'static>,
544    is_singleton = true
545);
546icu_provider::data_marker!(
547    /// `PropertyBinaryPrintV1`
548    PropertyBinaryPrintV1,
549    PropertyCodePointSet<'static>,
550    is_singleton = true
551);
552icu_provider::data_marker!(
553    /// `PropertyBinaryQuotationMarkV1`
554    PropertyBinaryQuotationMarkV1,
555    PropertyCodePointSet<'static>,
556    is_singleton = true
557);
558icu_provider::data_marker!(
559    /// `PropertyBinaryRadicalV1`
560    PropertyBinaryRadicalV1,
561    PropertyCodePointSet<'static>,
562    is_singleton = true
563);
564icu_provider::data_marker!(
565    /// `PropertyBinaryRegionalIndicatorV1`
566    PropertyBinaryRegionalIndicatorV1,
567    PropertyCodePointSet<'static>,
568    is_singleton = true
569);
570icu_provider::data_marker!(
571    /// `PropertyBinarySegmentStarterV1`
572    PropertyBinarySegmentStarterV1,
573    PropertyCodePointSet<'static>,
574    is_singleton = true
575);
576icu_provider::data_marker!(
577    /// `PropertyBinarySentenceTerminalV1`
578    PropertyBinarySentenceTerminalV1,
579    PropertyCodePointSet<'static>,
580    is_singleton = true
581);
582icu_provider::data_marker!(
583    /// `PropertyBinarySoftDottedV1`
584    PropertyBinarySoftDottedV1,
585    PropertyCodePointSet<'static>,
586    is_singleton = true
587);
588icu_provider::data_marker!(
589    /// `PropertyBinaryTerminalPunctuationV1`
590    PropertyBinaryTerminalPunctuationV1,
591    PropertyCodePointSet<'static>,
592    is_singleton = true
593);
594icu_provider::data_marker!(
595    /// `PropertyBinaryUnifiedIdeographV1`
596    PropertyBinaryUnifiedIdeographV1,
597    PropertyCodePointSet<'static>,
598    is_singleton = true
599);
600icu_provider::data_marker!(
601    /// `PropertyBinaryUppercaseV1`
602    PropertyBinaryUppercaseV1,
603    PropertyCodePointSet<'static>,
604    is_singleton = true
605);
606icu_provider::data_marker!(
607    /// `PropertyBinaryVariationSelectorV1`
608    PropertyBinaryVariationSelectorV1,
609    PropertyCodePointSet<'static>,
610    is_singleton = true
611);
612icu_provider::data_marker!(
613    /// `PropertyBinaryWhiteSpaceV1`
614    PropertyBinaryWhiteSpaceV1,
615    PropertyCodePointSet<'static>,
616    is_singleton = true
617);
618icu_provider::data_marker!(
619    /// `PropertyBinaryXdigitV1`
620    PropertyBinaryXdigitV1,
621    PropertyCodePointSet<'static>,
622    is_singleton = true
623);
624icu_provider::data_marker!(
625    /// `PropertyBinaryXidContinueV1`
626    PropertyBinaryXidContinueV1,
627    PropertyCodePointSet<'static>,
628    is_singleton = true
629);
630icu_provider::data_marker!(
631    /// `PropertyBinaryXidStartV1`
632    PropertyBinaryXidStartV1,
633    PropertyCodePointSet<'static>,
634    is_singleton = true
635);
636icu_provider::data_marker!(
637    /// Data marker for the 'BidiClass' Unicode property
638    PropertyEnumBidiClassV1,
639    PropertyCodePointMap<'static, BidiClass>,
640    is_singleton = true,
641);
642icu_provider::data_marker!(
643    /// Data marker for the 'CanonicalCombiningClass' Unicode property
644    PropertyEnumCanonicalCombiningClassV1,
645    PropertyCodePointMap<'static, CanonicalCombiningClass>,
646    is_singleton = true,
647);
648icu_provider::data_marker!(
649    /// Data marker for the 'EastAsianWidth' Unicode property
650    PropertyEnumEastAsianWidthV1,
651    PropertyCodePointMap<'static, EastAsianWidth>,
652    is_singleton = true,
653);
654icu_provider::data_marker!(
655    /// Data marker for the 'GeneralCategory' Unicode property
656    PropertyEnumGeneralCategoryV1,
657    PropertyCodePointMap<'static, GeneralCategory>,
658    is_singleton = true,
659);
660icu_provider::data_marker!(
661    /// Data marker for the 'GraphemeClusterBreak' Unicode property
662    PropertyEnumGraphemeClusterBreakV1,
663    PropertyCodePointMap<'static, GraphemeClusterBreak>,
664    is_singleton = true,
665);
666icu_provider::data_marker!(
667    /// Data marker for the 'HangulSyllableType' Unicode property
668    PropertyEnumHangulSyllableTypeV1,
669    PropertyCodePointMap<'static, HangulSyllableType>,
670    is_singleton = true,
671);
672icu_provider::data_marker!(
673    /// Data marker for the 'IndicConjunctBreak' Unicode property
674    PropertyEnumIndicConjunctBreakV1,
675    PropertyCodePointMap<'static, IndicConjunctBreak>,
676    is_singleton = true,
677);
678icu_provider::data_marker!(
679    /// Data marker for the 'IndicSyllabicCategory' Unicode property
680    PropertyEnumIndicSyllabicCategoryV1,
681    PropertyCodePointMap<'static, IndicSyllabicCategory>,
682    is_singleton = true,
683);
684icu_provider::data_marker!(
685    /// Data marker for the 'JoiningGroup' Unicode property
686    PropertyEnumJoiningGroupV1,
687    PropertyCodePointMap<'static, JoiningGroup>,
688    is_singleton = true,
689);
690icu_provider::data_marker!(
691    /// Data marker for the 'JoiningType' Unicode property
692    PropertyEnumJoiningTypeV1,
693    PropertyCodePointMap<'static, JoiningType>,
694    is_singleton = true,
695);
696icu_provider::data_marker!(
697    /// Data marker for the 'LineBreak' Unicode property
698    PropertyEnumLineBreakV1,
699    PropertyCodePointMap<'static, LineBreak>,
700    is_singleton = true,
701);
702icu_provider::data_marker!(
703    /// Data marker for the 'NumericValue' Unicode property
704    PropertyEnumNumericTypeV1,
705    PropertyCodePointMap<'static, NumericType>,
706    is_singleton = true,
707);
708icu_provider::data_marker!(
709    /// Data marker for the 'Script' Unicode property
710    PropertyEnumScriptV1,
711    PropertyCodePointMap<'static, Script>,
712    is_singleton = true,
713);
714icu_provider::data_marker!(
715    /// Data marker for the 'SentenceBreak' Unicode property
716    PropertyEnumSentenceBreakV1,
717    PropertyCodePointMap<'static, SentenceBreak>,
718    is_singleton = true,
719);
720icu_provider::data_marker!(
721    /// Data marker for the 'Vertical_Orientation' Unicode property
722    PropertyEnumVerticalOrientationV1,
723    PropertyCodePointMap<'static, VerticalOrientation>,
724    is_singleton = true,
725);
726icu_provider::data_marker!(
727    /// Data marker for the 'WordBreak' Unicode property
728    PropertyEnumWordBreakV1,
729    PropertyCodePointMap<'static, WordBreak>,
730    is_singleton = true,
731);
732icu_provider::data_marker!(
733    /// Data marker for the 'BidiMirroringGlyph' Unicode property
734    PropertyEnumBidiMirroringGlyphV1,
735    PropertyCodePointMap<'static, BidiMirroringGlyph>,
736    is_singleton = true,
737);
738icu_provider::data_marker!(
739    /// `PropertyBinaryBasicEmojiV1`
740    PropertyBinaryBasicEmojiV1,
741    PropertyUnicodeSet<'static>,
742    is_singleton = true
743);
744icu_provider::data_marker!(
745    /// `PropertyScriptWithExtensionsV1`
746    PropertyScriptWithExtensionsV1,
747    ScriptWithExtensionsProperty<'static>,
748    is_singleton = true
749);
750
751/// All data keys in this module.
752pub const MARKERS: &[DataMarkerInfo] = &[
753    PropertyNameLongBidiClassV1::INFO,
754    #[cfg(feature = "alloc")]
755    PropertyNameLongCanonicalCombiningClassV1::INFO,
756    PropertyNameLongEastAsianWidthV1::INFO,
757    PropertyNameLongGeneralCategoryV1::INFO,
758    PropertyNameLongGraphemeClusterBreakV1::INFO,
759    PropertyNameLongHangulSyllableTypeV1::INFO,
760    PropertyNameLongIndicSyllabicCategoryV1::INFO,
761    PropertyNameLongIndicConjunctBreakV1::INFO,
762    PropertyNameLongJoiningGroupV1::INFO,
763    PropertyNameLongJoiningTypeV1::INFO,
764    PropertyNameLongLineBreakV1::INFO,
765    PropertyNameLongNumericTypeV1::INFO,
766    PropertyNameLongScriptV1::INFO,
767    PropertyNameLongSentenceBreakV1::INFO,
768    PropertyNameLongVerticalOrientationV1::INFO,
769    PropertyNameLongWordBreakV1::INFO,
770    PropertyNameParseBidiClassV1::INFO,
771    PropertyNameParseCanonicalCombiningClassV1::INFO,
772    PropertyNameParseEastAsianWidthV1::INFO,
773    PropertyNameParseGeneralCategoryMaskV1::INFO,
774    PropertyNameParseGeneralCategoryV1::INFO,
775    PropertyNameParseGraphemeClusterBreakV1::INFO,
776    PropertyNameParseHangulSyllableTypeV1::INFO,
777    PropertyNameParseIndicSyllabicCategoryV1::INFO,
778    PropertyNameParseIndicConjunctBreakV1::INFO,
779    PropertyNameParseJoiningGroupV1::INFO,
780    PropertyNameParseJoiningTypeV1::INFO,
781    PropertyNameParseLineBreakV1::INFO,
782    PropertyNameParseNumericTypeV1::INFO,
783    PropertyNameParseScriptV1::INFO,
784    PropertyNameParseSentenceBreakV1::INFO,
785    PropertyNameParseVerticalOrientationV1::INFO,
786    PropertyNameParseWordBreakV1::INFO,
787    PropertyNameShortBidiClassV1::INFO,
788    #[cfg(feature = "alloc")]
789    PropertyNameShortCanonicalCombiningClassV1::INFO,
790    PropertyNameShortEastAsianWidthV1::INFO,
791    PropertyNameShortGeneralCategoryV1::INFO,
792    PropertyNameShortGraphemeClusterBreakV1::INFO,
793    PropertyNameShortHangulSyllableTypeV1::INFO,
794    PropertyNameShortIndicSyllabicCategoryV1::INFO,
795    PropertyNameShortIndicConjunctBreakV1::INFO,
796    PropertyNameShortJoiningGroupV1::INFO,
797    PropertyNameShortJoiningTypeV1::INFO,
798    PropertyNameShortLineBreakV1::INFO,
799    PropertyNameShortNumericTypeV1::INFO,
800    PropertyNameShortScriptV1::INFO,
801    PropertyNameShortSentenceBreakV1::INFO,
802    PropertyNameShortVerticalOrientationV1::INFO,
803    PropertyNameShortWordBreakV1::INFO,
804    PropertyBinaryAlnumV1::INFO,
805    PropertyBinaryAlphabeticV1::INFO,
806    PropertyBinaryAsciiHexDigitV1::INFO,
807    PropertyBinaryBidiControlV1::INFO,
808    PropertyBinaryBidiMirroredV1::INFO,
809    PropertyBinaryBlankV1::INFO,
810    PropertyBinaryCasedV1::INFO,
811    PropertyBinaryCaseIgnorableV1::INFO,
812    PropertyBinaryCaseSensitiveV1::INFO,
813    PropertyBinaryChangesWhenCasefoldedV1::INFO,
814    PropertyBinaryChangesWhenCasemappedV1::INFO,
815    PropertyBinaryChangesWhenLowercasedV1::INFO,
816    PropertyBinaryChangesWhenNfkcCasefoldedV1::INFO,
817    PropertyBinaryChangesWhenTitlecasedV1::INFO,
818    PropertyBinaryChangesWhenUppercasedV1::INFO,
819    PropertyBinaryDashV1::INFO,
820    PropertyBinaryDefaultIgnorableCodePointV1::INFO,
821    PropertyBinaryDeprecatedV1::INFO,
822    PropertyBinaryDiacriticV1::INFO,
823    PropertyBinaryEmojiComponentV1::INFO,
824    PropertyBinaryEmojiModifierBaseV1::INFO,
825    PropertyBinaryEmojiModifierV1::INFO,
826    PropertyBinaryEmojiPresentationV1::INFO,
827    PropertyBinaryEmojiV1::INFO,
828    PropertyBinaryExtendedPictographicV1::INFO,
829    PropertyBinaryExtenderV1::INFO,
830    PropertyBinaryFullCompositionExclusionV1::INFO,
831    PropertyBinaryGraphemeBaseV1::INFO,
832    PropertyBinaryGraphemeExtendV1::INFO,
833    PropertyBinaryGraphemeLinkV1::INFO,
834    PropertyBinaryGraphV1::INFO,
835    PropertyBinaryHexDigitV1::INFO,
836    PropertyBinaryHyphenV1::INFO,
837    PropertyBinaryIdCompatMathContinueV1::INFO,
838    PropertyBinaryIdCompatMathStartV1::INFO,
839    PropertyBinaryIdContinueV1::INFO,
840    PropertyBinaryIdeographicV1::INFO,
841    PropertyBinaryIdsBinaryOperatorV1::INFO,
842    PropertyBinaryIdStartV1::INFO,
843    PropertyBinaryIdsTrinaryOperatorV1::INFO,
844    PropertyBinaryIdsUnaryOperatorV1::INFO,
845    PropertyBinaryJoinControlV1::INFO,
846    PropertyBinaryLogicalOrderExceptionV1::INFO,
847    PropertyBinaryLowercaseV1::INFO,
848    PropertyBinaryMathV1::INFO,
849    PropertyBinaryModifierCombiningMarkV1::INFO,
850    PropertyBinaryNfcInertV1::INFO,
851    PropertyBinaryNfdInertV1::INFO,
852    PropertyBinaryNfkcInertV1::INFO,
853    PropertyBinaryNfkdInertV1::INFO,
854    PropertyBinaryNoncharacterCodePointV1::INFO,
855    PropertyBinaryPatternSyntaxV1::INFO,
856    PropertyBinaryPatternWhiteSpaceV1::INFO,
857    PropertyBinaryPrependedConcatenationMarkV1::INFO,
858    PropertyBinaryPrintV1::INFO,
859    PropertyBinaryQuotationMarkV1::INFO,
860    PropertyBinaryRadicalV1::INFO,
861    PropertyBinaryRegionalIndicatorV1::INFO,
862    PropertyBinarySegmentStarterV1::INFO,
863    PropertyBinarySentenceTerminalV1::INFO,
864    PropertyBinarySoftDottedV1::INFO,
865    PropertyBinaryTerminalPunctuationV1::INFO,
866    PropertyBinaryUnifiedIdeographV1::INFO,
867    PropertyBinaryUppercaseV1::INFO,
868    PropertyBinaryVariationSelectorV1::INFO,
869    PropertyBinaryWhiteSpaceV1::INFO,
870    PropertyBinaryXdigitV1::INFO,
871    PropertyBinaryXidContinueV1::INFO,
872    PropertyBinaryXidStartV1::INFO,
873    PropertyEnumBidiClassV1::INFO,
874    PropertyEnumCanonicalCombiningClassV1::INFO,
875    PropertyEnumEastAsianWidthV1::INFO,
876    PropertyEnumGeneralCategoryV1::INFO,
877    PropertyEnumGraphemeClusterBreakV1::INFO,
878    PropertyEnumHangulSyllableTypeV1::INFO,
879    PropertyEnumIndicConjunctBreakV1::INFO,
880    PropertyEnumIndicSyllabicCategoryV1::INFO,
881    PropertyEnumJoiningGroupV1::INFO,
882    PropertyEnumJoiningTypeV1::INFO,
883    PropertyEnumLineBreakV1::INFO,
884    PropertyEnumNumericTypeV1::INFO,
885    PropertyEnumScriptV1::INFO,
886    PropertyEnumSentenceBreakV1::INFO,
887    PropertyEnumVerticalOrientationV1::INFO,
888    PropertyEnumWordBreakV1::INFO,
889    PropertyEnumBidiMirroringGlyphV1::INFO,
890    PropertyBinaryBasicEmojiV1::INFO,
891    PropertyScriptWithExtensionsV1::INFO,
892];
893
894/// A set of characters which share a particular property value.
895///
896/// This data enum is extensible, more backends may be added in the future.
897/// Old data can be used with newer code but not vice versa.
898///
899/// <div class="stab unstable">
900/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
901/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
902/// to be stable, their Rust representation might not be. Use with caution.
903/// </div>
904#[derive(Debug, Eq, PartialEq, Clone, yoke::Yokeable, zerofrom::ZeroFrom)]
905#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
906#[cfg_attr(feature = "datagen", databake(path = icu_properties::provider))]
907#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
908#[non_exhaustive]
909pub enum PropertyCodePointSet<'data> {
910    /// The set of characters, represented as an inversion list
911    InversionList(#[cfg_attr(feature = "serde", serde(borrow))] CodePointInversionList<'data>),
912    // new variants should go BELOW existing ones
913    // Serde serializes based on variant name and index in the enum
914    // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
915}
916
917icu_provider::data_struct!(
918    PropertyCodePointSet<'_>,
919    #[cfg(feature = "datagen")]
920);
921
922// See CodePointSetData for documentation of these functions
923impl<'data> PropertyCodePointSet<'data> {
924    #[inline]
925    pub(crate) fn contains(&self, ch: char) -> bool {
926        match *self {
927            Self::InversionList(ref l) => l.contains(ch),
928        }
929    }
930
931    #[inline]
932    pub(crate) fn contains32(&self, ch: u32) -> bool {
933        match *self {
934            Self::InversionList(ref l) => l.contains32(ch),
935        }
936    }
937
938    #[inline]
939    pub(crate) fn iter_ranges(&self) -> impl Iterator<Item = RangeInclusive<u32>> + '_ {
940        match *self {
941            Self::InversionList(ref l) => l.iter_ranges(),
942        }
943    }
944
945    #[inline]
946    pub(crate) fn iter_ranges_complemented(
947        &self,
948    ) -> impl Iterator<Item = RangeInclusive<u32>> + '_ {
949        match *self {
950            Self::InversionList(ref l) => l.iter_ranges_complemented(),
951        }
952    }
953
954    #[inline]
955    pub(crate) fn from_code_point_inversion_list(l: CodePointInversionList<'static>) -> Self {
956        Self::InversionList(l)
957    }
958
959    #[inline]
960    #[allow(clippy::unnecessary_wraps)] // extensibly designed
961    pub(crate) fn as_code_point_inversion_list(
962        &'_ self,
963    ) -> Option<&'_ CodePointInversionList<'data>> {
964        match *self {
965            Self::InversionList(ref l) => Some(l),
966            // any other backing data structure that cannot return a CPInvList in O(1) time should return None
967        }
968    }
969
970    #[inline]
971    pub(crate) fn to_code_point_inversion_list(&self) -> CodePointInversionList<'_> {
972        match *self {
973            Self::InversionList(ref t) => ZeroFrom::zero_from(t),
974        }
975    }
976}
977
978/// A map efficiently storing data about individual characters.
979///
980/// This data enum is extensible, more backends may be added in the future.
981/// Old data can be used with newer code but not vice versa.
982///
983/// <div class="stab unstable">
984/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
985/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
986/// to be stable, their Rust representation might not be. Use with caution.
987/// </div>
988#[derive(Clone, Debug, Eq, PartialEq, yoke::Yokeable, zerofrom::ZeroFrom)]
989#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
990#[cfg_attr(feature = "datagen", databake(path = icu_properties::provider))]
991#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
992#[non_exhaustive]
993pub enum PropertyCodePointMap<'data, T: TrieValue> {
994    /// A codepoint trie storing the data
995    CodePointTrie(#[cfg_attr(feature = "serde", serde(borrow))] CodePointTrie<'data, T>),
996    // new variants should go BELOW existing ones
997    // Serde serializes based on variant name and index in the enum
998    // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
999}
1000
1001icu_provider::data_struct!(
1002    <T: TrieValue> PropertyCodePointMap<'_, T>,
1003    #[cfg(feature = "datagen")]
1004);
1005
1006// See CodePointMapData for documentation of these functions
1007impl<'data, T: TrieValue> PropertyCodePointMap<'data, T> {
1008    #[inline]
1009    pub(crate) fn get32(&self, ch: u32) -> T {
1010        match *self {
1011            Self::CodePointTrie(ref t) => t.get32(ch),
1012        }
1013    }
1014
1015    #[inline]
1016    pub(crate) fn get(&self, c: char) -> T {
1017        match *self {
1018            Self::CodePointTrie(ref t) => t.get(c),
1019        }
1020    }
1021
1022    #[inline]
1023    #[cfg(feature = "alloc")]
1024    pub(crate) fn try_into_converted<P>(
1025        self,
1026    ) -> Result<PropertyCodePointMap<'data, P>, zerovec::ule::UleError>
1027    where
1028        P: TrieValue,
1029    {
1030        match self {
1031            Self::CodePointTrie(t) => t
1032                .try_into_converted()
1033                .map(PropertyCodePointMap::CodePointTrie),
1034        }
1035    }
1036
1037    #[inline]
1038    #[cfg(feature = "alloc")]
1039    pub(crate) fn get_set_for_value(&self, value: T) -> CodePointInversionList<'static> {
1040        match *self {
1041            Self::CodePointTrie(ref t) => t.get_set_for_value(value),
1042        }
1043    }
1044
1045    #[inline]
1046    pub(crate) fn iter_ranges(&self) -> impl Iterator<Item = CodePointMapRange<T>> + '_ {
1047        match *self {
1048            Self::CodePointTrie(ref t) => t.iter_ranges(),
1049        }
1050    }
1051    #[inline]
1052    pub(crate) fn iter_ranges_mapped<'a, U: Eq + 'a>(
1053        &'a self,
1054        map: impl FnMut(T) -> U + Copy + 'a,
1055    ) -> impl Iterator<Item = CodePointMapRange<U>> + 'a {
1056        match *self {
1057            Self::CodePointTrie(ref t) => t.iter_ranges_mapped(map),
1058        }
1059    }
1060
1061    #[inline]
1062    pub(crate) fn from_code_point_trie(trie: CodePointTrie<'static, T>) -> Self {
1063        Self::CodePointTrie(trie)
1064    }
1065
1066    #[inline]
1067    #[allow(clippy::unnecessary_wraps)] // extensibly designed
1068    pub(crate) fn as_code_point_trie(&self) -> Option<&CodePointTrie<'data, T>> {
1069        match *self {
1070            Self::CodePointTrie(ref t) => Some(t),
1071            // any other backing data structure that cannot return a CPT in O(1) time should return None
1072        }
1073    }
1074
1075    #[inline]
1076    pub(crate) fn to_code_point_trie(&self) -> CodePointTrie<'_, T> {
1077        match *self {
1078            Self::CodePointTrie(ref t) => ZeroFrom::zero_from(t),
1079        }
1080    }
1081}
1082
1083/// A set of characters and strings which share a particular property value.
1084///
1085/// <div class="stab unstable">
1086/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
1087/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
1088/// to be stable, their Rust representation might not be. Use with caution.
1089/// </div>
1090#[derive(Debug, Eq, PartialEq, Clone, yoke::Yokeable, zerofrom::ZeroFrom)]
1091#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
1092#[cfg_attr(feature = "datagen", databake(path = icu_properties::provider))]
1093#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
1094#[non_exhaustive]
1095pub enum PropertyUnicodeSet<'data> {
1096    /// A set representing characters in an inversion list, and the strings in a list.
1097    CPInversionListStrList(
1098        #[cfg_attr(feature = "serde", serde(borrow))] CodePointInversionListAndStringList<'data>,
1099    ),
1100    // new variants should go BELOW existing ones
1101    // Serde serializes based on variant name and index in the enum
1102    // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
1103}
1104
1105icu_provider::data_struct!(
1106    PropertyUnicodeSet<'_>,
1107    #[cfg(feature = "datagen")]
1108);
1109
1110impl<'data> PropertyUnicodeSet<'data> {
1111    #[inline]
1112    pub(crate) fn contains_str(&self, s: &str) -> bool {
1113        match *self {
1114            Self::CPInversionListStrList(ref l) => l.contains_str(s),
1115        }
1116    }
1117
1118    #[inline]
1119    pub(crate) fn contains_utf8(&self, s: &[u8]) -> bool {
1120        match *self {
1121            Self::CPInversionListStrList(ref l) => l.contains_utf8(s),
1122        }
1123    }
1124
1125    #[inline]
1126    pub(crate) fn contains32(&self, cp: u32) -> bool {
1127        match *self {
1128            Self::CPInversionListStrList(ref l) => l.contains32(cp),
1129        }
1130    }
1131
1132    #[inline]
1133    pub(crate) fn contains(&self, ch: char) -> bool {
1134        match *self {
1135            Self::CPInversionListStrList(ref l) => l.contains(ch),
1136        }
1137    }
1138
1139    #[inline]
1140    pub(crate) fn from_code_point_inversion_list_string_list(
1141        l: CodePointInversionListAndStringList<'static>,
1142    ) -> Self {
1143        Self::CPInversionListStrList(l)
1144    }
1145
1146    #[inline]
1147    #[allow(clippy::unnecessary_wraps)] // extensibly designed
1148    pub(crate) fn as_code_point_inversion_list_string_list(
1149        &'_ self,
1150    ) -> Option<&'_ CodePointInversionListAndStringList<'data>> {
1151        match *self {
1152            Self::CPInversionListStrList(ref l) => Some(l),
1153            // any other backing data structure that cannot return a CPInversionListStrList in O(1) time should return None
1154        }
1155    }
1156
1157    #[inline]
1158    pub(crate) fn to_code_point_inversion_list_string_list(
1159        &self,
1160    ) -> CodePointInversionListAndStringList<'_> {
1161        match *self {
1162            Self::CPInversionListStrList(ref t) => ZeroFrom::zero_from(t),
1163        }
1164    }
1165}
1166
1167/// A struct that efficiently stores `Script` and `Script_Extensions` property data.
1168///
1169/// <div class="stab unstable">
1170/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
1171/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
1172/// to be stable, their Rust representation might not be. Use with caution.
1173/// </div>
1174#[derive(Debug, Eq, PartialEq, Clone, yoke::Yokeable, zerofrom::ZeroFrom)]
1175#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
1176#[cfg_attr(feature = "datagen", databake(path = icu_properties::provider))]
1177#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
1178pub struct ScriptWithExtensionsProperty<'data> {
1179    /// Note: The `ScriptWithExt` values in this array will assume a 12-bit layout. The 2
1180    /// higher order bits 11..10 will indicate how to deduce the Script value and
1181    /// `Script_Extensions` value, nearly matching the representation
1182    /// [in ICU](https://github.com/unicode-org/icu/blob/main/icu4c/source/common/uprops.h):
1183    ///
1184    /// | High order 2 bits value | Script                                                 | `Script_Extensions`                                              |
1185    /// |-------------------------|--------------------------------------------------------|----------------------------------------------------------------|
1186    /// | 3                       | First value in sub-array, index given by lower 10 bits | Sub-array excluding first value, index given by lower 10 bits  |
1187    /// | 2                       | Script=Inherited                                       | Entire sub-array, index given by lower 10 bits                 |
1188    /// | 1                       | Script=Common                                          | Entire sub-array, index given by lower 10 bits                 |
1189    /// | 0                       | Value in lower 10 bits                                 | `[ Script value ]` single-element array                        |
1190    ///
1191    /// When the lower 10 bits of the value are used as an index, that index is
1192    /// used for the outer-level vector of the nested `extensions` structure.
1193    #[cfg_attr(feature = "serde", serde(borrow))]
1194    pub trie: CodePointTrie<'data, ScriptWithExt>,
1195
1196    /// This companion structure stores `Script_Extensions` values, which are
1197    /// themselves arrays / vectors. This structure only stores the values for
1198    /// cases in which `scx(cp) != [ sc(cp) ]`. Each sub-vector is distinct. The
1199    /// sub-vector represents the `Script_Extensions` array value for a code point,
1200    /// and may also indicate Script value, as described for the `trie` field.
1201    #[cfg_attr(feature = "serde", serde(borrow))]
1202    pub extensions: VarZeroVec<'data, ZeroSlice<Script>>,
1203}
1204
1205icu_provider::data_struct!(
1206    ScriptWithExtensionsProperty<'_>,
1207    #[cfg(feature = "datagen")]
1208);
icu_properties/provider.rs

icu_properties/
provider.rs
