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
20use crate::script::ScriptWithExt;
21use crate::Script;
22
23use core::ops::RangeInclusive;
24use core::str;
25use icu_collections::codepointinvlist::CodePointInversionList;
26use icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList;
27use icu_collections::codepointtrie::{CodePointMapRange, CodePointTrie, TrieValue};
28use icu_provider::prelude::*;
29use icu_provider::{DataKeyMetadata, FallbackPriority};
30use zerofrom::ZeroFrom;
31
32use zerovec::{VarZeroVec, ZeroSlice, ZeroVecError};
33
34#[cfg(feature = "compiled_data")]
35#[derive(Debug)]
36/// Baked data
37///
38/// <div class="stab unstable">
39/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
40/// including in SemVer minor releases. In particular, the `DataProvider` implementations are only
41/// guaranteed to match with this version's `*_unstable` providers. Use with caution.
42/// </div>
43pub struct Baked;
44
45#[cfg(feature = "compiled_data")]
46const _: () = {
47 pub mod icu {
48 pub use crate as properties;
49 pub use icu_collections as collections;
50 #[allow(unused_imports)] // baked data may or may not need this
51 pub use icu_locid_transform as locid_transform;
52 }
53 icu_properties_data::make_provider!(Baked);
54 icu_properties_data::impl_propnames_from_gcb_v1!(Baked);
55 icu_properties_data::impl_propnames_from_bc_v1!(Baked);
56 icu_properties_data::impl_propnames_from_ccc_v1!(Baked);
57 icu_properties_data::impl_propnames_from_ea_v1!(Baked);
58 icu_properties_data::impl_propnames_from_gc_v1!(Baked);
59 icu_properties_data::impl_propnames_from_gcm_v1!(Baked);
60 icu_properties_data::impl_propnames_from_hst_v1!(Baked);
61 icu_properties_data::impl_propnames_from_insc_v1!(Baked);
62 icu_properties_data::impl_propnames_from_jt_v1!(Baked);
63 icu_properties_data::impl_propnames_from_lb_v1!(Baked);
64 icu_properties_data::impl_propnames_from_sb_v1!(Baked);
65 icu_properties_data::impl_propnames_from_sc_v1!(Baked);
66 icu_properties_data::impl_propnames_from_wb_v1!(Baked);
67 icu_properties_data::impl_propnames_to_long_linear_bc_v1!(Baked);
68 icu_properties_data::impl_propnames_to_long_linear_ea_v1!(Baked);
69 icu_properties_data::impl_propnames_to_long_linear_gc_v1!(Baked);
70 icu_properties_data::impl_propnames_to_long_linear_gcb_v1!(Baked);
71 icu_properties_data::impl_propnames_to_long_linear_hst_v1!(Baked);
72 icu_properties_data::impl_propnames_to_long_linear_insc_v1!(Baked);
73 icu_properties_data::impl_propnames_to_long_linear_jt_v1!(Baked);
74 icu_properties_data::impl_propnames_to_long_linear_lb_v1!(Baked);
75 icu_properties_data::impl_propnames_to_long_linear_sb_v1!(Baked);
76 icu_properties_data::impl_propnames_to_long_linear_sc_v1!(Baked);
77 icu_properties_data::impl_propnames_to_long_linear_wb_v1!(Baked);
78 icu_properties_data::impl_propnames_to_long_sparse_ccc_v1!(Baked);
79 icu_properties_data::impl_propnames_to_short_linear_bc_v1!(Baked);
80 icu_properties_data::impl_propnames_to_short_linear_ea_v1!(Baked);
81 icu_properties_data::impl_propnames_to_short_linear_gc_v1!(Baked);
82 icu_properties_data::impl_propnames_to_short_linear_gcb_v1!(Baked);
83 icu_properties_data::impl_propnames_to_short_linear_hst_v1!(Baked);
84 icu_properties_data::impl_propnames_to_short_linear_insc_v1!(Baked);
85 icu_properties_data::impl_propnames_to_short_linear_jt_v1!(Baked);
86 icu_properties_data::impl_propnames_to_short_linear_lb_v1!(Baked);
87 icu_properties_data::impl_propnames_to_short_linear_sb_v1!(Baked);
88 icu_properties_data::impl_propnames_to_short_linear_wb_v1!(Baked);
89 icu_properties_data::impl_propnames_to_short_linear4_sc_v1!(Baked);
90 icu_properties_data::impl_propnames_to_short_sparse_ccc_v1!(Baked);
91 icu_properties_data::impl_props_ahex_v1!(Baked);
92 icu_properties_data::impl_props_alnum_v1!(Baked);
93 icu_properties_data::impl_props_alpha_v1!(Baked);
94 icu_properties_data::impl_props_basic_emoji_v1!(Baked);
95 icu_properties_data::impl_props_bc_v1!(Baked);
96 icu_properties_data::impl_props_bidi_c_v1!(Baked);
97 icu_properties_data::impl_props_bidi_m_v1!(Baked);
98 icu_properties_data::impl_props_bidiauxiliaryprops_v1!(Baked);
99 icu_properties_data::impl_props_blank_v1!(Baked);
100 icu_properties_data::impl_props_cased_v1!(Baked);
101 icu_properties_data::impl_props_ccc_v1!(Baked);
102 icu_properties_data::impl_props_ci_v1!(Baked);
103 icu_properties_data::impl_props_comp_ex_v1!(Baked);
104 icu_properties_data::impl_props_cwcf_v1!(Baked);
105 icu_properties_data::impl_props_cwcm_v1!(Baked);
106 icu_properties_data::impl_props_cwkcf_v1!(Baked);
107 icu_properties_data::impl_props_cwl_v1!(Baked);
108 icu_properties_data::impl_props_cwt_v1!(Baked);
109 icu_properties_data::impl_props_cwu_v1!(Baked);
110 icu_properties_data::impl_props_dash_v1!(Baked);
111 icu_properties_data::impl_props_dep_v1!(Baked);
112 icu_properties_data::impl_props_di_v1!(Baked);
113 icu_properties_data::impl_props_dia_v1!(Baked);
114 icu_properties_data::impl_props_ea_v1!(Baked);
115 icu_properties_data::impl_props_ebase_v1!(Baked);
116 icu_properties_data::impl_props_ecomp_v1!(Baked);
117 icu_properties_data::impl_props_emod_v1!(Baked);
118 icu_properties_data::impl_props_emoji_v1!(Baked);
119 icu_properties_data::impl_props_epres_v1!(Baked);
120 icu_properties_data::impl_props_exemplarchars_auxiliary_v1!(Baked);
121 icu_properties_data::impl_props_exemplarchars_index_v1!(Baked);
122 icu_properties_data::impl_props_exemplarchars_main_v1!(Baked);
123 icu_properties_data::impl_props_exemplarchars_numbers_v1!(Baked);
124 icu_properties_data::impl_props_exemplarchars_punctuation_v1!(Baked);
125 icu_properties_data::impl_props_ext_v1!(Baked);
126 icu_properties_data::impl_props_extpict_v1!(Baked);
127 icu_properties_data::impl_props_gc_v1!(Baked);
128 icu_properties_data::impl_props_gcb_v1!(Baked);
129 icu_properties_data::impl_props_gr_base_v1!(Baked);
130 icu_properties_data::impl_props_gr_ext_v1!(Baked);
131 icu_properties_data::impl_props_gr_link_v1!(Baked);
132 icu_properties_data::impl_props_graph_v1!(Baked);
133 icu_properties_data::impl_props_hex_v1!(Baked);
134 icu_properties_data::impl_props_hst_v1!(Baked);
135 icu_properties_data::impl_props_hyphen_v1!(Baked);
136 icu_properties_data::impl_props_idc_v1!(Baked);
137 icu_properties_data::impl_props_ideo_v1!(Baked);
138 icu_properties_data::impl_props_ids_v1!(Baked);
139 icu_properties_data::impl_props_idsb_v1!(Baked);
140 icu_properties_data::impl_props_idst_v1!(Baked);
141 icu_properties_data::impl_props_insc_v1!(Baked);
142 icu_properties_data::impl_props_join_c_v1!(Baked);
143 icu_properties_data::impl_props_jt_v1!(Baked);
144 icu_properties_data::impl_props_lb_v1!(Baked);
145 icu_properties_data::impl_props_loe_v1!(Baked);
146 icu_properties_data::impl_props_lower_v1!(Baked);
147 icu_properties_data::impl_props_math_v1!(Baked);
148 icu_properties_data::impl_props_nchar_v1!(Baked);
149 icu_properties_data::impl_props_nfcinert_v1!(Baked);
150 icu_properties_data::impl_props_nfdinert_v1!(Baked);
151 icu_properties_data::impl_props_nfkcinert_v1!(Baked);
152 icu_properties_data::impl_props_nfkdinert_v1!(Baked);
153 icu_properties_data::impl_props_pat_syn_v1!(Baked);
154 icu_properties_data::impl_props_pat_ws_v1!(Baked);
155 icu_properties_data::impl_props_pcm_v1!(Baked);
156 icu_properties_data::impl_props_print_v1!(Baked);
157 icu_properties_data::impl_props_qmark_v1!(Baked);
158 icu_properties_data::impl_props_radical_v1!(Baked);
159 icu_properties_data::impl_props_ri_v1!(Baked);
160 icu_properties_data::impl_props_sb_v1!(Baked);
161 icu_properties_data::impl_props_sc_v1!(Baked);
162 icu_properties_data::impl_props_scx_v1!(Baked);
163 icu_properties_data::impl_props_sd_v1!(Baked);
164 icu_properties_data::impl_props_segstart_v1!(Baked);
165 icu_properties_data::impl_props_sensitive_v1!(Baked);
166 icu_properties_data::impl_props_sterm_v1!(Baked);
167 icu_properties_data::impl_props_term_v1!(Baked);
168 icu_properties_data::impl_props_uideo_v1!(Baked);
169 icu_properties_data::impl_props_upper_v1!(Baked);
170 icu_properties_data::impl_props_vs_v1!(Baked);
171 icu_properties_data::impl_props_wb_v1!(Baked);
172 icu_properties_data::impl_props_wspace_v1!(Baked);
173 icu_properties_data::impl_props_xdigit_v1!(Baked);
174 icu_properties_data::impl_props_xidc_v1!(Baked);
175 icu_properties_data::impl_props_xids_v1!(Baked);
176};
177
178// include the specialized structs for the compact representation of Bidi property data
179pub mod bidi_data;
180
181/// A set of characters which share a particular property value.
182///
183/// This data enum is extensible, more backends may be added in the future.
184/// Old data can be used with newer code but not vice versa.
185///
186/// <div class="stab unstable">
187/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
188/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
189/// to be stable, their Rust representation might not be. Use with caution.
190/// </div>
191#[derive(Debug, Eq, PartialEq, Clone, yoke::Yokeable, zerofrom::ZeroFrom)]
192#[cfg_attr(
193 feature = "datagen",
194 derive(serde::Serialize, databake::Bake),
195 databake(path = icu_properties::provider),
196)]
197#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
198#[non_exhaustive]
199pub enum PropertyCodePointSetV1<'data> {
200 /// The set of characters, represented as an inversion list
201 InversionList(#[cfg_attr(feature = "serde", serde(borrow))] CodePointInversionList<'data>),
202 // new variants should go BELOW existing ones
203 // Serde serializes based on variant name and index in the enum
204 // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
205}
206
207/// A map efficiently storing data about individual characters.
208///
209/// This data enum is extensible, more backends may be added in the future.
210/// Old data can be used with newer code but not vice versa.
211///
212/// <div class="stab unstable">
213/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
214/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
215/// to be stable, their Rust representation might not be. Use with caution.
216/// </div>
217#[derive(Clone, Debug, Eq, PartialEq, yoke::Yokeable, zerofrom::ZeroFrom)]
218#[cfg_attr(
219 feature = "datagen",
220 derive(serde::Serialize, databake::Bake),
221 databake(path = icu_properties::provider),
222)]
223#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
224#[non_exhaustive]
225pub enum PropertyCodePointMapV1<'data, T: TrieValue> {
226 /// A codepoint trie storing the data
227 CodePointTrie(#[cfg_attr(feature = "serde", serde(borrow))] CodePointTrie<'data, T>),
228 // new variants should go BELOW existing ones
229 // Serde serializes based on variant name and index in the enum
230 // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
231}
232
233/// A set of characters and strings which share a particular property value.
234///
235/// <div class="stab unstable">
236/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
237/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
238/// to be stable, their Rust representation might not be. Use with caution.
239/// </div>
240#[derive(Debug, Eq, PartialEq, Clone, yoke::Yokeable, zerofrom::ZeroFrom)]
241#[cfg_attr(
242 feature = "datagen",
243 derive(serde::Serialize, databake::Bake),
244 databake(path = icu_properties::provider),
245)]
246#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
247#[non_exhaustive]
248pub enum PropertyUnicodeSetV1<'data> {
249 /// A set representing characters in an inversion list, and the strings in a list.
250 CPInversionListStrList(
251 #[cfg_attr(feature = "serde", serde(borrow))] CodePointInversionListAndStringList<'data>,
252 ),
253 // new variants should go BELOW existing ones
254 // Serde serializes based on variant name and index in the enum
255 // https://docs.rs/serde/latest/serde/trait.Serializer.html#tymethod.serialize_unit_variant
256}
257
258impl<'data> PropertyUnicodeSetV1<'data> {
259 #[inline]
260 pub(crate) fn contains(&self, s: &str) -> bool {
261 match *self {
262 Self::CPInversionListStrList(ref l) => l.contains(s),
263 }
264 }
265
266 #[inline]
267 pub(crate) fn contains32(&self, cp: u32) -> bool {
268 match *self {
269 Self::CPInversionListStrList(ref l) => l.contains32(cp),
270 }
271 }
272
273 #[inline]
274 pub(crate) fn contains_char(&self, ch: char) -> bool {
275 match *self {
276 Self::CPInversionListStrList(ref l) => l.contains_char(ch),
277 }
278 }
279
280 #[inline]
281 pub(crate) fn from_code_point_inversion_list_string_list(
282 l: CodePointInversionListAndStringList<'static>,
283 ) -> Self {
284 Self::CPInversionListStrList(l)
285 }
286
287 #[inline]
288 pub(crate) fn as_code_point_inversion_list_string_list(
289 &'_ self,
290 ) -> Option<&'_ CodePointInversionListAndStringList<'data>> {
291 match *self {
292 Self::CPInversionListStrList(ref l) => Some(l),
293 // any other backing data structure that cannot return a CPInversionListStrList in O(1) time should return None
294 }
295 }
296
297 #[inline]
298 pub(crate) fn to_code_point_inversion_list_string_list(
299 &self,
300 ) -> CodePointInversionListAndStringList<'_> {
301 match *self {
302 Self::CPInversionListStrList(ref t) => ZeroFrom::zero_from(t),
303 }
304 }
305}
306
307/// A struct that efficiently stores `Script` and `Script_Extensions` property data.
308///
309/// <div class="stab unstable">
310/// 🚧 This code is considered unstable; it may change at any time, in breaking or non-breaking ways,
311/// including in SemVer minor releases. While the serde representation of data structs is guaranteed
312/// to be stable, their Rust representation might not be. Use with caution.
313/// </div>
314#[icu_provider::data_struct(marker(
315 ScriptWithExtensionsPropertyV1Marker,
316 "props/scx@1",
317 singleton
318))]
319#[derive(Debug, Eq, PartialEq, Clone)]
320#[cfg_attr(
321 feature = "datagen",
322 derive(serde::Serialize, databake::Bake),
323 databake(path = icu_properties::provider),
324)]
325#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
326pub struct ScriptWithExtensionsPropertyV1<'data> {
327 /// Note: The `ScriptWithExt` values in this array will assume a 12-bit layout. The 2
328 /// higher order bits 11..10 will indicate how to deduce the Script value and
329 /// Script_Extensions value, nearly matching the representation
330 /// [in ICU](https://github.com/unicode-org/icu/blob/main/icu4c/source/common/uprops.h):
331 ///
332 /// | High order 2 bits value | Script | Script_Extensions |
333 /// |-------------------------|--------------------------------------------------------|----------------------------------------------------------------|
334 /// | 3 | First value in sub-array, index given by lower 10 bits | Sub-array excluding first value, index given by lower 10 bits |
335 /// | 2 | Script=Inherited | Entire sub-array, index given by lower 10 bits |
336 /// | 1 | Script=Common | Entire sub-array, index given by lower 10 bits |
337 /// | 0 | Value in lower 10 bits | `[ Script value ]` single-element array |
338 ///
339 /// When the lower 10 bits of the value are used as an index, that index is
340 /// used for the outer-level vector of the nested `extensions` structure.
341 #[cfg_attr(feature = "serde", serde(borrow))]
342 pub trie: CodePointTrie<'data, ScriptWithExt>,
343
344 /// This companion structure stores Script_Extensions values, which are
345 /// themselves arrays / vectors. This structure only stores the values for
346 /// cases in which `scx(cp) != [ sc(cp) ]`. Each sub-vector is distinct. The
347 /// sub-vector represents the Script_Extensions array value for a code point,
348 /// and may also indicate Script value, as described for the `trie` field.
349 #[cfg_attr(feature = "serde", serde(borrow))]
350 pub extensions: VarZeroVec<'data, ZeroSlice<Script>>,
351}
352
353impl<'data> ScriptWithExtensionsPropertyV1<'data> {
354 // This method is intended to be used by constructors of deserialized data
355 // in a data provider.
356 #[doc(hidden)]
357 pub fn new(
358 trie: CodePointTrie<'data, ScriptWithExt>,
359 extensions: VarZeroVec<'data, ZeroSlice<Script>>,
360 ) -> ScriptWithExtensionsPropertyV1<'data> {
361 ScriptWithExtensionsPropertyV1 { trie, extensions }
362 }
363}
364
365// See CodePointSetData for documentation of these functions
366impl<'data> PropertyCodePointSetV1<'data> {
367 #[inline]
368 pub(crate) fn contains(&self, ch: char) -> bool {
369 match *self {
370 Self::InversionList(ref l) => l.contains(ch),
371 }
372 }
373
374 #[inline]
375 pub(crate) fn contains32(&self, ch: u32) -> bool {
376 match *self {
377 Self::InversionList(ref l) => l.contains32(ch),
378 }
379 }
380
381 #[inline]
382 pub(crate) fn iter_ranges(&self) -> impl Iterator<Item = RangeInclusive<u32>> + '_ {
383 match *self {
384 Self::InversionList(ref l) => l.iter_ranges(),
385 }
386 }
387
388 #[inline]
389 pub(crate) fn iter_ranges_complemented(
390 &self,
391 ) -> impl Iterator<Item = RangeInclusive<u32>> + '_ {
392 match *self {
393 Self::InversionList(ref l) => l.iter_ranges_complemented(),
394 }
395 }
396
397 #[inline]
398 pub(crate) fn from_code_point_inversion_list(l: CodePointInversionList<'static>) -> Self {
399 Self::InversionList(l)
400 }
401
402 #[inline]
403 pub(crate) fn as_code_point_inversion_list(
404 &'_ self,
405 ) -> Option<&'_ CodePointInversionList<'data>> {
406 match *self {
407 Self::InversionList(ref l) => Some(l),
408 // any other backing data structure that cannot return a CPInvList in O(1) time should return None
409 }
410 }
411
412 #[inline]
413 pub(crate) fn to_code_point_inversion_list(&self) -> CodePointInversionList<'_> {
414 match *self {
415 Self::InversionList(ref t) => ZeroFrom::zero_from(t),
416 }
417 }
418}
419
420// See CodePointMapData for documentation of these functions
421impl<'data, T: TrieValue> PropertyCodePointMapV1<'data, T> {
422 #[inline]
423 pub(crate) fn get32(&self, ch: u32) -> T {
424 match *self {
425 Self::CodePointTrie(ref t) => t.get32(ch),
426 }
427 }
428
429 #[inline]
430 pub(crate) fn try_into_converted<P>(
431 self,
432 ) -> Result<PropertyCodePointMapV1<'data, P>, ZeroVecError>
433 where
434 P: TrieValue,
435 {
436 match self {
437 Self::CodePointTrie(t) => t
438 .try_into_converted()
439 .map(PropertyCodePointMapV1::CodePointTrie),
440 }
441 }
442
443 #[inline]
444 pub(crate) fn get_set_for_value(&self, value: T) -> CodePointInversionList<'static> {
445 match *self {
446 Self::CodePointTrie(ref t) => t.get_set_for_value(value),
447 }
448 }
449
450 #[inline]
451 pub(crate) fn iter_ranges(&self) -> impl Iterator<Item = CodePointMapRange<T>> + '_ {
452 match *self {
453 Self::CodePointTrie(ref t) => t.iter_ranges(),
454 }
455 }
456 #[inline]
457 pub(crate) fn iter_ranges_mapped<'a, U: Eq + 'a>(
458 &'a self,
459 map: impl FnMut(T) -> U + Copy + 'a,
460 ) -> impl Iterator<Item = CodePointMapRange<U>> + 'a {
461 match *self {
462 Self::CodePointTrie(ref t) => t.iter_ranges_mapped(map),
463 }
464 }
465
466 #[inline]
467 pub(crate) fn from_code_point_trie(trie: CodePointTrie<'static, T>) -> Self {
468 Self::CodePointTrie(trie)
469 }
470
471 #[inline]
472 pub(crate) fn as_code_point_trie(&self) -> Option<&CodePointTrie<'data, T>> {
473 match *self {
474 Self::CodePointTrie(ref t) => Some(t),
475 // any other backing data structure that cannot return a CPT in O(1) time should return None
476 }
477 }
478
479 #[inline]
480 pub(crate) fn to_code_point_trie(&self) -> CodePointTrie<'_, T> {
481 match *self {
482 Self::CodePointTrie(ref t) => ZeroFrom::zero_from(t),
483 }
484 }
485}
486
487macro_rules! expand {
488 (
489 ($(($code_point_set_marker:ident, $bin_cp_s:literal),)+),
490 ($(($unicode_set_marker:ident, $bin_us_s:literal, $us_singleton:literal),)+),
491 ($(($code_point_map_marker:ident,
492 $name_value_marker:ident,
493
494 $((sparse: $value_short_name_marker_sparse:ident, $value_long_name_marker_sparse:ident),)?
495 $((linear: $value_short_name_marker_linear:ident, $value_long_name_marker_linear:ident ),)?
496 $((linear4: $value_short_name_marker_linear4:ident, $value_long_name_marker_linear4:ident ),)?
497 $enum_s:literal, $value_ty:ident),)+)
498 ) => {
499
500 // Data keys that return code point sets (represented as CodePointSetData).
501 // For now, synonymous with binary properties of code points only.
502 $(
503 #[doc = core::concat!("Data marker for the '", $bin_cp_s, "' Unicode property")]
504 #[derive(Debug, Default)]
505 #[cfg_attr(
506 feature = "datagen",
507 derive(databake::Bake),
508 databake(path = icu_properties::provider),
509 )]
510 pub struct $code_point_set_marker;
511
512 impl DataMarker for $code_point_set_marker {
513 type Yokeable = PropertyCodePointSetV1<'static>;
514 }
515 impl KeyedDataMarker for $code_point_set_marker {
516 const KEY: DataKey = data_key!(concat!("props/", $bin_cp_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
517 }
518
519 )+
520
521 // Data keys that return sets of strings + code points (represented as UnicodeSetData).
522 // Includes:
523 // - binary properties of strings + code points
524 // - exemplar characters
525 $(
526 #[doc = core::concat!("Data marker for the '", $bin_us_s, "' Unicode property")]
527 #[derive(Debug, Default)]
528 #[cfg_attr(
529 feature = "datagen",
530 derive(databake::Bake),
531 databake(path = icu_properties::provider),
532 )]
533 pub struct $unicode_set_marker;
534
535 impl DataMarker for $unicode_set_marker {
536 type Yokeable = PropertyUnicodeSetV1<'static>;
537 }
538 impl KeyedDataMarker for $unicode_set_marker {
539 const KEY: DataKey = data_key!(concat!("props/", $bin_us_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, $us_singleton));
540 }
541 )+
542
543 // Data keys that return code point map (represented as CodePointMapData).
544 // For now, synonymous with enumerated properties [of code points only].
545 $(
546 #[doc = core::concat!("Data marker for the '", $enum_s, "' Unicode property")]
547 #[derive(Debug, Default)]
548 #[cfg_attr(
549 feature = "datagen",
550 derive(databake::Bake),
551 databake(path = icu_properties::provider),
552 )]
553 pub struct $code_point_map_marker;
554
555 impl DataMarker for $code_point_map_marker {
556 type Yokeable = PropertyCodePointMapV1<'static, crate::$value_ty>;
557 }
558
559 impl KeyedDataMarker for $code_point_map_marker {
560 const KEY: DataKey = data_key!(concat!("props/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
561 }
562
563
564 #[doc = core::concat!("Data marker for parsing the names of the values of the '", $enum_s, "' Unicode property")]
565 #[derive(Debug, Default)]
566 #[cfg_attr(
567 feature = "datagen",
568 derive(databake::Bake),
569 databake(path = icu_properties::provider),
570 )]
571 pub struct $name_value_marker;
572
573 impl DataMarker for $name_value_marker {
574 type Yokeable = names::PropertyValueNameToEnumMapV1<'static>;
575 }
576
577 impl KeyedDataMarker for $name_value_marker {
578 const KEY: DataKey = data_key!(concat!("propnames/from/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
579 }
580
581 $(
582 #[doc = core::concat!("Data marker for producing short names of the values of the '", $enum_s, "' Unicode property")]
583 #[derive(Debug, Default)]
584 #[cfg_attr(
585 feature = "datagen",
586 derive(databake::Bake),
587 databake(path = icu_properties::provider),
588 )]
589 pub struct $value_short_name_marker_sparse;
590
591 impl DataMarker for $value_short_name_marker_sparse {
592 type Yokeable = names::PropertyEnumToValueNameSparseMapV1<'static>;
593 }
594
595 impl KeyedDataMarker for $value_short_name_marker_sparse {
596 const KEY: DataKey = data_key!(concat!("propnames/to/short/sparse/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
597 }
598
599 #[doc = core::concat!("Data marker for producing long names of the values of the '", $enum_s, "' Unicode property")]
600 #[derive(Debug, Default)]
601 #[cfg_attr(
602 feature = "datagen",
603 derive(databake::Bake),
604 databake(path = icu_properties::provider),
605 )]
606 pub struct $value_long_name_marker_sparse;
607
608 impl DataMarker for $value_long_name_marker_sparse {
609 type Yokeable = names::PropertyEnumToValueNameSparseMapV1<'static>;
610 }
611
612 impl KeyedDataMarker for $value_long_name_marker_sparse {
613 const KEY: DataKey = data_key!(concat!("propnames/to/long/sparse/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
614 }
615 )?
616
617 $(
618 #[doc = core::concat!("Data marker for producing short names of the values of the '", $enum_s, "' Unicode property")]
619 #[derive(Debug, Default)]
620 #[cfg_attr(
621 feature = "datagen",
622 derive(databake::Bake),
623 databake(path = icu_properties::provider),
624 )]
625 pub struct $value_short_name_marker_linear;
626
627 impl DataMarker for $value_short_name_marker_linear {
628 type Yokeable = names::PropertyEnumToValueNameLinearMapV1<'static>;
629 }
630
631 impl KeyedDataMarker for $value_short_name_marker_linear {
632 const KEY: DataKey = data_key!(concat!("propnames/to/short/linear/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
633 }
634
635 #[doc = core::concat!("Data marker for producing long names of the values of the '", $enum_s, "' Unicode property")]
636 #[derive(Debug, Default)]
637 #[cfg_attr(
638 feature = "datagen",
639 derive(databake::Bake),
640 databake(path = icu_properties::provider),
641 )]
642 pub struct $value_long_name_marker_linear;
643
644 impl DataMarker for $value_long_name_marker_linear {
645 type Yokeable = names::PropertyEnumToValueNameLinearMapV1<'static>;
646 }
647
648 impl KeyedDataMarker for $value_long_name_marker_linear {
649 const KEY: DataKey = data_key!(concat!("propnames/to/long/linear/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
650 }
651 )?
652
653 $(
654 #[doc = core::concat!("Data marker for producing short names of the values of the '", $enum_s, "' Unicode property")]
655 #[derive(Debug, Default)]
656 #[cfg_attr(
657 feature = "datagen",
658 derive(databake::Bake),
659 databake(path = icu_properties::provider),
660 )]
661 pub struct $value_short_name_marker_linear4;
662
663 impl DataMarker for $value_short_name_marker_linear4 {
664 type Yokeable = names::PropertyEnumToValueNameLinearTiny4MapV1<'static>;
665 }
666
667 impl KeyedDataMarker for $value_short_name_marker_linear4 {
668 const KEY: DataKey = data_key!(concat!("propnames/to/short/linear4/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
669 }
670
671 #[doc = core::concat!("Data marker for producing long names of the values of the '", $enum_s, "' Unicode property")]
672 #[derive(Debug, Default)]
673 #[cfg_attr(
674 feature = "datagen",
675 derive(databake::Bake),
676 databake(path = icu_properties::provider),
677 )]
678 pub struct $value_long_name_marker_linear4;
679
680 impl DataMarker for $value_long_name_marker_linear4 {
681 // Tiny4 is only for short names
682 type Yokeable = names::PropertyEnumToValueNameLinearMapV1<'static>;
683 }
684
685 impl KeyedDataMarker for $value_long_name_marker_linear4 {
686 const KEY: DataKey = data_key!(concat!("propnames/to/long/linear/", $enum_s, "@1"), DataKeyMetadata::construct_internal(FallbackPriority::Language, None, None, true));
687 }
688 )?
689 )+
690
691 /// All data keys in this module.
692 pub const KEYS: &[DataKey] = &[
693 $($code_point_set_marker::KEY,)+
694 $($unicode_set_marker::KEY,)+
695 $(
696 $code_point_map_marker::KEY,
697 $name_value_marker::KEY,
698 $($value_short_name_marker_sparse::KEY, $value_long_name_marker_sparse::KEY,)?
699 $($value_short_name_marker_linear::KEY, $value_long_name_marker_linear::KEY,)?
700 $($value_short_name_marker_linear4::KEY, $value_long_name_marker_linear4::KEY,)?
701 )+
702 bidi_data::BidiAuxiliaryPropertiesV1Marker::KEY,
703 GeneralCategoryMaskNameToValueV1Marker::KEY,
704 ScriptWithExtensionsPropertyV1Marker::KEY,
705 ];
706 };
707}
708
709pub use self::names::GeneralCategoryMaskNameToValueV1Marker;
710
711expand!(
712 (
713 // code point sets
714 (AsciiHexDigitV1Marker, "AHex"),
715 (AlnumV1Marker, "alnum"),
716 (AlphabeticV1Marker, "Alpha"),
717 (BidiControlV1Marker, "Bidi_C"),
718 (BidiMirroredV1Marker, "Bidi_M"),
719 (BlankV1Marker, "blank"),
720 (CasedV1Marker, "Cased"),
721 (CaseIgnorableV1Marker, "CI"),
722 (FullCompositionExclusionV1Marker, "Comp_Ex"),
723 (ChangesWhenCasefoldedV1Marker, "CWCF"),
724 (ChangesWhenCasemappedV1Marker, "CWCM"),
725 (ChangesWhenNfkcCasefoldedV1Marker, "CWKCF"),
726 (ChangesWhenLowercasedV1Marker, "CWL"),
727 (ChangesWhenTitlecasedV1Marker, "CWT"),
728 (ChangesWhenUppercasedV1Marker, "CWU"),
729 (DashV1Marker, "Dash"),
730 (DeprecatedV1Marker, "Dep"),
731 (DefaultIgnorableCodePointV1Marker, "DI"),
732 (DiacriticV1Marker, "Dia"),
733 (EmojiModifierBaseV1Marker, "EBase"),
734 (EmojiComponentV1Marker, "EComp"),
735 (EmojiModifierV1Marker, "EMod"),
736 (EmojiV1Marker, "Emoji"),
737 (EmojiPresentationV1Marker, "EPres"),
738 (ExtenderV1Marker, "Ext"),
739 (ExtendedPictographicV1Marker, "ExtPict"),
740 (GraphV1Marker, "graph"),
741 (GraphemeBaseV1Marker, "Gr_Base"),
742 (GraphemeExtendV1Marker, "Gr_Ext"),
743 (GraphemeLinkV1Marker, "Gr_Link"),
744 (HexDigitV1Marker, "Hex"),
745 (HyphenV1Marker, "Hyphen"),
746 (IdContinueV1Marker, "IDC"),
747 (IdeographicV1Marker, "Ideo"),
748 (IdStartV1Marker, "IDS"),
749 (IdsBinaryOperatorV1Marker, "IDSB"),
750 (IdsTrinaryOperatorV1Marker, "IDST"),
751 (JoinControlV1Marker, "Join_C"),
752 (LogicalOrderExceptionV1Marker, "LOE"),
753 (LowercaseV1Marker, "Lower"),
754 (MathV1Marker, "Math"),
755 (NoncharacterCodePointV1Marker, "NChar"),
756 (NfcInertV1Marker, "nfcinert"),
757 (NfdInertV1Marker, "nfdinert"),
758 (NfkcInertV1Marker, "nfkcinert"),
759 (NfkdInertV1Marker, "nfkdinert"),
760 (PatternSyntaxV1Marker, "Pat_Syn"),
761 (PatternWhiteSpaceV1Marker, "Pat_WS"),
762 (PrependedConcatenationMarkV1Marker, "PCM"),
763 (PrintV1Marker, "print"),
764 (QuotationMarkV1Marker, "QMark"),
765 (RadicalV1Marker, "Radical"),
766 (RegionalIndicatorV1Marker, "RI"),
767 (SoftDottedV1Marker, "SD"),
768 (SegmentStarterV1Marker, "segstart"),
769 (CaseSensitiveV1Marker, "Sensitive"),
770 (SentenceTerminalV1Marker, "STerm"),
771 (TerminalPunctuationV1Marker, "Term"),
772 (UnifiedIdeographV1Marker, "UIdeo"),
773 (UppercaseV1Marker, "Upper"),
774 (VariationSelectorV1Marker, "VS"),
775 (WhiteSpaceV1Marker, "WSpace"),
776 (XdigitV1Marker, "xdigit"),
777 (XidContinueV1Marker, "XIDC"),
778 (XidStartV1Marker, "XIDS"),
779 ),
780 (
781 // UnicodeSets (code points + strings)
782 (BasicEmojiV1Marker, "Basic_Emoji", true),
783 (ExemplarCharactersMainV1Marker, "exemplarchars/main", false),
784 (
785 ExemplarCharactersAuxiliaryV1Marker,
786 "exemplarchars/auxiliary",
787 false
788 ),
789 (
790 ExemplarCharactersPunctuationV1Marker,
791 "exemplarchars/punctuation",
792 false
793 ),
794 (
795 ExemplarCharactersNumbersV1Marker,
796 "exemplarchars/numbers",
797 false
798 ),
799 (
800 ExemplarCharactersIndexV1Marker,
801 "exemplarchars/index",
802 false
803 ),
804 ),
805 (
806 // code point maps
807 (
808 CanonicalCombiningClassV1Marker,
809 CanonicalCombiningClassNameToValueV1Marker,
810 (
811 sparse: CanonicalCombiningClassValueToShortNameV1Marker,
812 CanonicalCombiningClassValueToLongNameV1Marker
813 ),
814 "ccc",
815 CanonicalCombiningClass
816 ),
817 (
818 GeneralCategoryV1Marker,
819 GeneralCategoryNameToValueV1Marker,
820 (
821 linear: GeneralCategoryValueToShortNameV1Marker,
822 GeneralCategoryValueToLongNameV1Marker
823 ),
824 "gc",
825 GeneralCategory
826 ),
827 (
828 BidiClassV1Marker,
829 BidiClassNameToValueV1Marker,
830 (
831 linear: BidiClassValueToShortNameV1Marker,
832 BidiClassValueToLongNameV1Marker
833 ),
834 "bc",
835 BidiClass
836 ),
837 (
838 ScriptV1Marker,
839 ScriptNameToValueV1Marker,
840 (
841 linear4: ScriptValueToShortNameV1Marker,
842 ScriptValueToLongNameV1Marker
843 ),
844 "sc",
845 Script
846 ),
847 (
848 HangulSyllableTypeV1Marker,
849 HangulSyllableTypeNameToValueV1Marker,
850 (
851 linear: HangulSyllableTypeValueToShortNameV1Marker,
852 HangulSyllableTypeValueToLongNameV1Marker
853 ),
854 "hst",
855 HangulSyllableType
856 ),
857 (
858 EastAsianWidthV1Marker,
859 EastAsianWidthNameToValueV1Marker,
860 (
861 linear: EastAsianWidthValueToShortNameV1Marker,
862 EastAsianWidthValueToLongNameV1Marker
863 ),
864 "ea",
865 EastAsianWidth
866 ),
867 (
868 LineBreakV1Marker,
869 LineBreakNameToValueV1Marker,
870 (
871 linear: LineBreakValueToShortNameV1Marker,
872 LineBreakValueToLongNameV1Marker
873 ),
874 "lb",
875 LineBreak
876 ),
877 (
878 GraphemeClusterBreakV1Marker,
879 GraphemeClusterBreakNameToValueV1Marker,
880 (
881 linear: GraphemeClusterBreakValueToShortNameV1Marker,
882 GraphemeClusterBreakValueToLongNameV1Marker
883 ),
884 "GCB",
885 GraphemeClusterBreak
886 ),
887 (
888 WordBreakV1Marker,
889 WordBreakNameToValueV1Marker,
890 (
891 linear: WordBreakValueToShortNameV1Marker,
892 WordBreakValueToLongNameV1Marker
893 ),
894 "WB",
895 WordBreak
896 ),
897 (
898 SentenceBreakV1Marker,
899 SentenceBreakNameToValueV1Marker,
900 (
901 linear: SentenceBreakValueToShortNameV1Marker,
902 SentenceBreakValueToLongNameV1Marker
903 ),
904 "SB",
905 SentenceBreak
906 ),
907 (
908 IndicSyllabicCategoryV1Marker,
909 IndicSyllabicCategoryNameToValueV1Marker,
910 (
911 linear: IndicSyllabicCategoryValueToShortNameV1Marker,
912 IndicSyllabicCategoryValueToLongNameV1Marker
913 ),
914 "InSC",
915 IndicSyllabicCategory
916 ),
917 (
918 JoiningTypeV1Marker,
919 JoiningTypeNameToValueV1Marker,
920 (
921 linear: JoiningTypeValueToShortNameV1Marker,
922 JoiningTypeValueToLongNameV1Marker
923 ),
924 "jt",
925 JoiningType
926 ),
927 // note: the names key for the GCM mask is handled above
928 )
929);
930