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//! Data and APIs for supporting both Script and Script_Extensions property
6//! values in an efficient structure.
7
8use crate::error::PropertiesError;
9use crate::props::Script;
10use crate::props::ScriptULE;
11use crate::provider::*;
12
13use core::iter::FromIterator;
14use core::ops::RangeInclusive;
15use icu_collections::codepointinvlist::CodePointInversionList;
16use icu_provider::prelude::*;
17use zerovec::{ule::AsULE, ZeroSlice};
18
19/// The number of bits at the low-end of a `ScriptWithExt` value used for
20/// storing the `Script` value (or `extensions` index).
21const SCRIPT_VAL_LENGTH: u16 = 10;
22
23/// The bit mask necessary to retrieve the `Script` value (or `extensions` index)
24/// from a `ScriptWithExt` value.
25const SCRIPT_X_SCRIPT_VAL: u16 = (1 << SCRIPT_VAL_LENGTH) - 1;
26
27/// An internal-use only pseudo-property that represents the values stored in
28/// the trie of the special data structure [`ScriptWithExtensionsPropertyV1`].
29///
30/// Note: The will assume a 12-bit layout. The 2 higher order bits in positions
31/// 11..10 will indicate how to deduce the Script value and Script_Extensions,
32/// and the lower 10 bits 9..0 indicate either the Script value or the index
33/// into the `extensions` structure.
34#[derive(Copy, Clone, Debug, Eq, PartialEq)]
35#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
36#[cfg_attr(feature = "datagen", derive(databake::Bake))]
37#[cfg_attr(feature = "datagen", databake(path = icu_properties::script))]
38#[repr(transparent)]
39#[doc(hidden)]
40// `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsPropertyV1` constructor
41#[allow(clippy::exhaustive_structs)] // this type is stable
42pub struct ScriptWithExt(pub u16);
43
44#[allow(missing_docs)] // These constants don't need individual documentation.
45#[allow(non_upper_case_globals)]
46#[doc(hidden)] // `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsPropertyV1` constructor
47impl ScriptWithExt {
48 pub const Unknown: ScriptWithExt = ScriptWithExt(0);
49}
50
51impl AsULE for ScriptWithExt {
52 type ULE = ScriptULE;
53
54 #[inline]
55 fn to_unaligned(self) -> Self::ULE {
56 Script(self.0).to_unaligned()
57 }
58
59 #[inline]
60 fn from_unaligned(unaligned: Self::ULE) -> Self {
61 ScriptWithExt(Script::from_unaligned(unaligned).0)
62 }
63}
64
65#[doc(hidden)] // `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensionsPropertyV1` constructor
66impl ScriptWithExt {
67 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
68 /// also indicates a Script value of [`Script::Common`].
69 ///
70 /// # Examples
71 ///
72 /// ```
73 /// use icu::properties::script::ScriptWithExt;
74 ///
75 /// assert!(ScriptWithExt(0x04FF).is_common());
76 /// assert!(ScriptWithExt(0x0400).is_common());
77 ///
78 /// assert!(!ScriptWithExt(0x08FF).is_common());
79 /// assert!(!ScriptWithExt(0x0800).is_common());
80 ///
81 /// assert!(!ScriptWithExt(0x0CFF).is_common());
82 /// assert!(!ScriptWithExt(0x0C00).is_common());
83 ///
84 /// assert!(!ScriptWithExt(0xFF).is_common());
85 /// assert!(!ScriptWithExt(0x0).is_common());
86 /// ```
87 pub fn is_common(&self) -> bool {
88 self.0 >> SCRIPT_VAL_LENGTH == 1
89 }
90
91 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
92 /// also indicates a Script value of [`Script::Inherited`].
93 ///
94 /// # Examples
95 ///
96 /// ```
97 /// use icu::properties::script::ScriptWithExt;
98 ///
99 /// assert!(!ScriptWithExt(0x04FF).is_inherited());
100 /// assert!(!ScriptWithExt(0x0400).is_inherited());
101 ///
102 /// assert!(ScriptWithExt(0x08FF).is_inherited());
103 /// assert!(ScriptWithExt(0x0800).is_inherited());
104 ///
105 /// assert!(!ScriptWithExt(0x0CFF).is_inherited());
106 /// assert!(!ScriptWithExt(0x0C00).is_inherited());
107 ///
108 /// assert!(!ScriptWithExt(0xFF).is_inherited());
109 /// assert!(!ScriptWithExt(0x0).is_inherited());
110 /// ```
111 pub fn is_inherited(&self) -> bool {
112 self.0 >> SCRIPT_VAL_LENGTH == 2
113 }
114
115 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions and
116 /// also indicates that the Script value is neither [`Script::Common`] nor
117 /// [`Script::Inherited`].
118 ///
119 /// # Examples
120 ///
121 /// ```
122 /// use icu::properties::script::ScriptWithExt;
123 ///
124 /// assert!(!ScriptWithExt(0x04FF).is_other());
125 /// assert!(!ScriptWithExt(0x0400).is_other());
126 ///
127 /// assert!(!ScriptWithExt(0x08FF).is_other());
128 /// assert!(!ScriptWithExt(0x0800).is_other());
129 ///
130 /// assert!(ScriptWithExt(0x0CFF).is_other());
131 /// assert!(ScriptWithExt(0x0C00).is_other());
132 ///
133 /// assert!(!ScriptWithExt(0xFF).is_other());
134 /// assert!(!ScriptWithExt(0x0).is_other());
135 /// ```
136 pub fn is_other(&self) -> bool {
137 self.0 >> SCRIPT_VAL_LENGTH == 3
138 }
139
140 /// Returns whether the [`ScriptWithExt`] value has Script_Extensions.
141 ///
142 /// # Examples
143 ///
144 /// ```
145 /// use icu::properties::script::ScriptWithExt;
146 ///
147 /// assert!(ScriptWithExt(0x04FF).has_extensions());
148 /// assert!(ScriptWithExt(0x0400).has_extensions());
149 ///
150 /// assert!(ScriptWithExt(0x08FF).has_extensions());
151 /// assert!(ScriptWithExt(0x0800).has_extensions());
152 ///
153 /// assert!(ScriptWithExt(0x0CFF).has_extensions());
154 /// assert!(ScriptWithExt(0x0C00).has_extensions());
155 ///
156 /// assert!(!ScriptWithExt(0xFF).has_extensions());
157 /// assert!(!ScriptWithExt(0x0).has_extensions());
158 /// ```
159 pub fn has_extensions(&self) -> bool {
160 let high_order_bits = self.0 >> SCRIPT_VAL_LENGTH;
161 high_order_bits > 0
162 }
163}
164
165impl From<ScriptWithExt> for u32 {
166 fn from(swe: ScriptWithExt) -> Self {
167 swe.0 as u32
168 }
169}
170
171impl From<ScriptWithExt> for Script {
172 fn from(swe: ScriptWithExt) -> Self {
173 Script(swe.0)
174 }
175}
176
177/// A struct that wraps a [`Script`] array, such as in the return value for
178/// [`get_script_extensions_val()`](ScriptWithExtensionsBorrowed::get_script_extensions_val).
179#[derive(Copy, Clone, Debug, Eq, PartialEq)]
180pub struct ScriptExtensionsSet<'a> {
181 values: &'a ZeroSlice<Script>,
182}
183
184impl<'a> ScriptExtensionsSet<'a> {
185 /// Returns whether this set contains the given script.
186 ///
187 /// # Example
188 ///
189 /// ```
190 /// use icu::properties::{script, Script};
191 /// let swe = script::script_with_extensions();
192 ///
193 /// assert!(swe
194 /// .get_script_extensions_val(0x11303) // GRANTHA SIGN VISARGA
195 /// .contains(&Script::Grantha));
196 /// ```
197 pub fn contains(&self, x: &Script) -> bool {
198 ZeroSlice::binary_search(self.values, x).is_ok()
199 }
200
201 /// Gets an iterator over the elements.
202 ///
203 /// # Example
204 ///
205 /// ```
206 /// use icu::properties::{script, Script};
207 /// let swe = script::script_with_extensions();
208 ///
209 /// assert_eq!(
210 /// swe.get_script_extensions_val('௫' as u32) // U+0BEB TAMIL DIGIT FIVE
211 /// .iter()
212 /// .collect::<Vec<Script>>(),
213 /// vec![Script::Tamil, Script::Grantha]
214 /// );
215 /// ```
216 pub fn iter(&self) -> impl DoubleEndedIterator<Item = Script> + 'a {
217 ZeroSlice::iter(self.values)
218 }
219
220 /// For accessing this set as an array instead of an iterator
221 /// only needed for the FFI bindings; shouldn't be used directly from Rust
222 #[doc(hidden)]
223 pub fn array_len(&self) -> usize {
224 self.values.len()
225 }
226 /// For accessing this set as an array instead of an iterator
227 /// only needed for the FFI bindings; shouldn't be used directly from Rust
228 #[doc(hidden)]
229 pub fn array_get(&self, index: usize) -> Option<Script> {
230 self.values.get(index)
231 }
232}
233
234/// A wrapper around script extensions data. Can be obtained via [`load_script_with_extensions_unstable()`] and
235/// related getters.
236///
237/// Most useful methods are on [`ScriptWithExtensionsBorrowed`] obtained by calling [`ScriptWithExtensions::as_borrowed()`]
238#[derive(Debug)]
239pub struct ScriptWithExtensions {
240 data: DataPayload<ScriptWithExtensionsPropertyV1Marker>,
241}
242
243/// A borrowed wrapper around script extension data, returned by
244/// [`ScriptWithExtensions::as_borrowed()`]. More efficient to query.
245#[derive(Clone, Copy, Debug)]
246pub struct ScriptWithExtensionsBorrowed<'a> {
247 data: &'a ScriptWithExtensionsPropertyV1<'a>,
248}
249
250impl ScriptWithExtensions {
251 /// Construct a borrowed version of this type that can be queried.
252 ///
253 /// This avoids a potential small underlying cost per API call (ex: `contains()`) by consolidating it
254 /// up front.
255 #[inline]
256 pub fn as_borrowed(&self) -> ScriptWithExtensionsBorrowed<'_> {
257 ScriptWithExtensionsBorrowed {
258 data: self.data.get(),
259 }
260 }
261
262 /// Construct a new one from loaded data
263 ///
264 /// Typically it is preferable to use getters like [`load_script_with_extensions_unstable()`] instead
265 pub fn from_data(data: DataPayload<ScriptWithExtensionsPropertyV1Marker>) -> Self {
266 Self { data }
267 }
268}
269
270impl<'a> ScriptWithExtensionsBorrowed<'a> {
271 /// Returns the `Script` property value for this code point.
272 ///
273 /// # Examples
274 ///
275 /// ```
276 /// use icu::properties::{script, Script};
277 ///
278 /// let swe = script::script_with_extensions();
279 ///
280 /// // U+0640 ARABIC TATWEEL
281 /// assert_eq!(swe.get_script_val(0x0640), Script::Common); // main Script value
282 /// assert_ne!(swe.get_script_val(0x0640), Script::Arabic);
283 /// assert_ne!(swe.get_script_val(0x0640), Script::Syriac);
284 /// assert_ne!(swe.get_script_val(0x0640), Script::Thaana);
285 ///
286 /// // U+0650 ARABIC KASRA
287 /// assert_eq!(swe.get_script_val(0x0650), Script::Inherited); // main Script value
288 /// assert_ne!(swe.get_script_val(0x0650), Script::Arabic);
289 /// assert_ne!(swe.get_script_val(0x0650), Script::Syriac);
290 /// assert_ne!(swe.get_script_val(0x0650), Script::Thaana);
291 ///
292 /// // U+0660 ARABIC-INDIC DIGIT ZERO
293 /// assert_ne!(swe.get_script_val(0x0660), Script::Common);
294 /// assert_eq!(swe.get_script_val(0x0660), Script::Arabic); // main Script value
295 /// assert_ne!(swe.get_script_val(0x0660), Script::Syriac);
296 /// assert_ne!(swe.get_script_val(0x0660), Script::Thaana);
297 ///
298 /// // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
299 /// assert_ne!(swe.get_script_val(0xFDF2), Script::Common);
300 /// assert_eq!(swe.get_script_val(0xFDF2), Script::Arabic); // main Script value
301 /// assert_ne!(swe.get_script_val(0xFDF2), Script::Syriac);
302 /// assert_ne!(swe.get_script_val(0xFDF2), Script::Thaana);
303 /// ```
304 pub fn get_script_val(self, code_point: u32) -> Script {
305 let sc_with_ext = self.data.trie.get32(code_point);
306
307 if sc_with_ext.is_other() {
308 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
309 let scx_val = self.data.extensions.get(ext_idx as usize);
310 let scx_first_sc = scx_val.and_then(|scx| scx.get(0));
311
312 let default_sc_val = Script::Unknown;
313
314 scx_first_sc.unwrap_or(default_sc_val)
315 } else if sc_with_ext.is_common() {
316 Script::Common
317 } else if sc_with_ext.is_inherited() {
318 Script::Inherited
319 } else {
320 let script_val = sc_with_ext.0;
321 Script(script_val)
322 }
323 }
324 // Returns the Script_Extensions value for a code_point when the trie value
325 // is already known.
326 // This private helper method exists to prevent code duplication in callers like
327 // `get_script_extensions_val`, `get_script_extensions_set`, and `has_script`.
328 fn get_scx_val_using_trie_val(
329 self,
330 sc_with_ext_ule: &'a <ScriptWithExt as AsULE>::ULE,
331 ) -> &'a ZeroSlice<Script> {
332 let sc_with_ext = ScriptWithExt::from_unaligned(*sc_with_ext_ule);
333 if sc_with_ext.is_other() {
334 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
335 let ext_subarray = self.data.extensions.get(ext_idx as usize);
336 // In the OTHER case, where the 2 higher-order bits of the
337 // `ScriptWithExt` value in the trie doesn't indicate the Script value,
338 // the Script value is copied/inserted into the first position of the
339 // `extensions` array. So we must remove it to return the actual scx array val.
340 let scx_slice = ext_subarray
341 .and_then(|zslice| zslice.as_ule_slice().get(1..))
342 .unwrap_or_default();
343 ZeroSlice::from_ule_slice(scx_slice)
344 } else if sc_with_ext.is_common() || sc_with_ext.is_inherited() {
345 let ext_idx = sc_with_ext.0 & SCRIPT_X_SCRIPT_VAL;
346 let scx_val = self.data.extensions.get(ext_idx as usize);
347 scx_val.unwrap_or_default()
348 } else {
349 // Note: `Script` and `ScriptWithExt` are both represented as the same
350 // u16 value when the `ScriptWithExt` has no higher-order bits set.
351 let script_ule_slice = core::slice::from_ref(sc_with_ext_ule);
352 ZeroSlice::from_ule_slice(script_ule_slice)
353 }
354 }
355 /// Return the `Script_Extensions` property value for this code point.
356 ///
357 /// If `code_point` has Script_Extensions, then return the Script codes in
358 /// the Script_Extensions. In this case, the Script property value
359 /// (normally Common or Inherited) is not included in the [`ScriptExtensionsSet`].
360 ///
361 /// If c does not have Script_Extensions, then the one Script code is put
362 /// into the [`ScriptExtensionsSet`] and also returned.
363 ///
364 /// If c is not a valid code point, then return an empty [`ScriptExtensionsSet`].
365 ///
366 /// # Examples
367 ///
368 /// ```
369 /// use icu::properties::{script, Script};
370 ///
371 /// let swe = script::script_with_extensions();
372 ///
373 /// assert_eq!(
374 /// swe.get_script_extensions_val('𐓐' as u32) // U+104D0 OSAGE CAPITAL LETTER KHA
375 /// .iter()
376 /// .collect::<Vec<Script>>(),
377 /// vec![Script::Osage]
378 /// );
379 /// assert_eq!(
380 /// swe.get_script_extensions_val('🥳' as u32) // U+1F973 FACE WITH PARTY HORN AND PARTY HAT
381 /// .iter()
382 /// .collect::<Vec<Script>>(),
383 /// vec![Script::Common]
384 /// );
385 /// assert_eq!(
386 /// swe.get_script_extensions_val(0x200D) // ZERO WIDTH JOINER
387 /// .iter()
388 /// .collect::<Vec<Script>>(),
389 /// vec![Script::Inherited]
390 /// );
391 /// assert_eq!(
392 /// swe.get_script_extensions_val('௫' as u32) // U+0BEB TAMIL DIGIT FIVE
393 /// .iter()
394 /// .collect::<Vec<Script>>(),
395 /// vec![Script::Tamil, Script::Grantha]
396 /// );
397 /// ```
398 pub fn get_script_extensions_val(self, code_point: u32) -> ScriptExtensionsSet<'a> {
399 let sc_with_ext_ule = self.data.trie.get32_ule(code_point);
400
401 ScriptExtensionsSet {
402 values: match sc_with_ext_ule {
403 Some(ule_ref) => self.get_scx_val_using_trie_val(ule_ref),
404 None => ZeroSlice::from_ule_slice(&[]),
405 },
406 }
407 }
408
409 /// Returns whether `script` is contained in the Script_Extensions
410 /// property value if the code_point has Script_Extensions, otherwise
411 /// if the code point does not have Script_Extensions then returns
412 /// whether the Script property value matches.
413 ///
414 /// Some characters are commonly used in multiple scripts. For more information,
415 /// see UAX #24: <http://www.unicode.org/reports/tr24/>.
416 ///
417 /// # Examples
418 ///
419 /// ```
420 /// use icu::properties::{script, Script};
421 ///
422 /// let swe = script::script_with_extensions();
423 ///
424 /// // U+0650 ARABIC KASRA
425 /// assert!(!swe.has_script(0x0650, Script::Inherited)); // main Script value
426 /// assert!(swe.has_script(0x0650, Script::Arabic));
427 /// assert!(swe.has_script(0x0650, Script::Syriac));
428 /// assert!(!swe.has_script(0x0650, Script::Thaana));
429 ///
430 /// // U+0660 ARABIC-INDIC DIGIT ZERO
431 /// assert!(!swe.has_script(0x0660, Script::Common)); // main Script value
432 /// assert!(swe.has_script(0x0660, Script::Arabic));
433 /// assert!(!swe.has_script(0x0660, Script::Syriac));
434 /// assert!(swe.has_script(0x0660, Script::Thaana));
435 ///
436 /// // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
437 /// assert!(!swe.has_script(0xFDF2, Script::Common));
438 /// assert!(swe.has_script(0xFDF2, Script::Arabic)); // main Script value
439 /// assert!(!swe.has_script(0xFDF2, Script::Syriac));
440 /// assert!(swe.has_script(0xFDF2, Script::Thaana));
441 /// ```
442 pub fn has_script(self, code_point: u32, script: Script) -> bool {
443 let sc_with_ext_ule = if let Some(scwe_ule) = self.data.trie.get32_ule(code_point) {
444 scwe_ule
445 } else {
446 return false;
447 };
448 let sc_with_ext = <ScriptWithExt as AsULE>::from_unaligned(*sc_with_ext_ule);
449
450 if !sc_with_ext.has_extensions() {
451 let script_val = sc_with_ext.0;
452 script == Script(script_val)
453 } else {
454 let scx_val = self.get_scx_val_using_trie_val(sc_with_ext_ule);
455 let script_find = scx_val.iter().find(|&sc| sc == script);
456 script_find.is_some()
457 }
458 }
459
460 /// Returns all of the matching `CodePointMapRange`s for the given [`Script`]
461 /// in which `has_script` will return true for all of the contained code points.
462 ///
463 /// # Examples
464 ///
465 /// ```
466 /// use icu::properties::{script, Script};
467 ///
468 /// let swe = script::script_with_extensions();
469 ///
470 /// let syriac_script_extensions_ranges = swe.get_script_extensions_ranges(Script::Syriac);
471 ///
472 /// let exp_ranges = vec![
473 /// 0x060C..=0x060C, // ARABIC COMMA
474 /// 0x061B..=0x061C, // ARABIC SEMICOLON, ARABIC LETTER MARK
475 /// 0x061F..=0x061F, // ARABIC QUESTION MARK
476 /// 0x0640..=0x0640, // ARABIC TATWEEL
477 /// 0x064B..=0x0655, // ARABIC FATHATAN..ARABIC HAMZA BELOW
478 /// 0x0670..=0x0670, // ARABIC LETTER SUPERSCRIPT ALEF
479 /// 0x0700..=0x070D, // Syriac block begins at U+0700
480 /// 0x070F..=0x074A, // Syriac block
481 /// 0x074D..=0x074F, // Syriac block ends at U+074F
482 /// 0x0860..=0x086A, // Syriac Supplement block is U+0860..=U+086F
483 /// 0x1DF8..=0x1DF8, // U+1DF8 COMBINING DOT ABOVE LEFT
484 /// 0x1DFA..=0x1DFA, // U+1DFA COMBINING DOT BELOW LEFT
485 /// ];
486 /// let mut exp_ranges_iter = exp_ranges.iter();
487 ///
488 /// for act_range in syriac_script_extensions_ranges {
489 /// let exp_range = exp_ranges_iter
490 /// .next()
491 /// .expect("There are too many ranges returned by get_script_extensions_ranges()");
492 /// assert_eq!(act_range.start(), exp_range.start());
493 /// assert_eq!(act_range.end(), exp_range.end());
494 /// }
495 /// assert!(
496 /// exp_ranges_iter.next().is_none(),
497 /// "There are too few ranges returned by get_script_extensions_ranges()"
498 /// );
499 /// ```
500 pub fn get_script_extensions_ranges(
501 self,
502 script: Script,
503 ) -> impl Iterator<Item = RangeInclusive<u32>> + 'a {
504 self.data
505 .trie
506 .iter_ranges_mapped(move |value| {
507 let sc_with_ext = ScriptWithExt(value.0);
508 if sc_with_ext.has_extensions() {
509 self.get_scx_val_using_trie_val(&sc_with_ext.to_unaligned())
510 .iter()
511 .any(|sc| sc == script)
512 } else {
513 script == sc_with_ext.into()
514 }
515 })
516 .filter(|v| v.value)
517 .map(|v| v.range)
518 }
519
520 /// Returns a [`CodePointInversionList`] for the given [`Script`] which represents all
521 /// code points for which `has_script` will return true.
522 ///
523 /// # Examples
524 ///
525 /// ```
526 /// use icu::properties::{script, Script};
527 ///
528 /// let swe = script::script_with_extensions();
529 ///
530 /// let syriac = swe.get_script_extensions_set(Script::Syriac);
531 ///
532 /// assert!(!syriac.contains32(0x061E)); // ARABIC TRIPLE DOT PUNCTUATION MARK
533 /// assert!(syriac.contains32(0x061F)); // ARABIC QUESTION MARK
534 /// assert!(!syriac.contains32(0x0620)); // ARABIC LETTER KASHMIRI YEH
535 ///
536 /// assert!(syriac.contains32(0x0700)); // SYRIAC END OF PARAGRAPH
537 /// assert!(syriac.contains32(0x074A)); // SYRIAC BARREKH
538 /// assert!(!syriac.contains32(0x074B)); // unassigned
539 /// assert!(syriac.contains32(0x074F)); // SYRIAC LETTER SOGDIAN FE
540 /// assert!(!syriac.contains32(0x0750)); // ARABIC LETTER BEH WITH THREE DOTS HORIZONTALLY BELOW
541 ///
542 /// assert!(syriac.contains32(0x1DF8)); // COMBINING DOT ABOVE LEFT
543 /// assert!(!syriac.contains32(0x1DF9)); // COMBINING WIDE INVERTED BRIDGE BELOW
544 /// assert!(syriac.contains32(0x1DFA)); // COMBINING DOT BELOW LEFT
545 /// assert!(!syriac.contains32(0x1DFB)); // COMBINING DELETION MARK
546 /// ```
547 pub fn get_script_extensions_set(self, script: Script) -> CodePointInversionList<'a> {
548 CodePointInversionList::from_iter(self.get_script_extensions_ranges(script))
549 }
550}
551
552impl ScriptWithExtensionsBorrowed<'static> {
553 /// Cheaply converts a [`ScriptWithExtensionsBorrowed<'static>`] into a [`ScriptWithExtensions`].
554 ///
555 /// Note: Due to branching and indirection, using [`ScriptWithExtensions`] might inhibit some
556 /// compile-time optimizations that are possible with [`ScriptWithExtensionsBorrowed`].
557 pub const fn static_to_owned(self) -> ScriptWithExtensions {
558 ScriptWithExtensions {
559 data: DataPayload::from_static_ref(self.data),
560 }
561 }
562}
563
564/// Returns a [`ScriptWithExtensionsBorrowed`] struct that represents the data for the Script
565/// and Script_Extensions properties.
566///
567/// ✨ *Enabled with the `compiled_data` Cargo feature.*
568///
569/// [📚 Help choosing a constructor](icu_provider::constructors)
570///
571/// # Examples
572///
573/// ```
574/// use icu::properties::{script, Script};
575/// let swe = script::script_with_extensions();
576///
577/// // get the `Script` property value
578/// assert_eq!(swe.get_script_val(0x0640), Script::Common); // U+0640 ARABIC TATWEEL
579/// assert_eq!(swe.get_script_val(0x0650), Script::Inherited); // U+0650 ARABIC KASRA
580/// assert_eq!(swe.get_script_val(0x0660), Script::Arabic); // // U+0660 ARABIC-INDIC DIGIT ZERO
581/// assert_eq!(swe.get_script_val(0xFDF2), Script::Arabic); // U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
582///
583/// // get the `Script_Extensions` property value
584/// assert_eq!(
585/// swe.get_script_extensions_val(0x0640) // U+0640 ARABIC TATWEEL
586/// .iter().collect::<Vec<Script>>(),
587/// vec![Script::Arabic, Script::Syriac, Script::Mandaic, Script::Manichaean,
588/// Script::PsalterPahlavi, Script::Adlam, Script::HanifiRohingya, Script::Sogdian,
589/// Script::OldUyghur]
590/// );
591/// assert_eq!(
592/// swe.get_script_extensions_val('🥳' as u32) // U+1F973 FACE WITH PARTY HORN AND PARTY HAT
593/// .iter().collect::<Vec<Script>>(),
594/// vec![Script::Common]
595/// );
596/// assert_eq!(
597/// swe.get_script_extensions_val(0x200D) // ZERO WIDTH JOINER
598/// .iter().collect::<Vec<Script>>(),
599/// vec![Script::Inherited]
600/// );
601/// assert_eq!(
602/// swe.get_script_extensions_val('௫' as u32) // U+0BEB TAMIL DIGIT FIVE
603/// .iter().collect::<Vec<Script>>(),
604/// vec![Script::Tamil, Script::Grantha]
605/// );
606///
607/// // check containment of a `Script` value in the `Script_Extensions` value
608/// // U+0650 ARABIC KASRA
609/// assert!(!swe.has_script(0x0650, Script::Inherited)); // main Script value
610/// assert!(swe.has_script(0x0650, Script::Arabic));
611/// assert!(swe.has_script(0x0650, Script::Syriac));
612/// assert!(!swe.has_script(0x0650, Script::Thaana));
613///
614/// // get a `CodePointInversionList` for when `Script` value is contained in `Script_Extensions` value
615/// let syriac = swe.get_script_extensions_set(Script::Syriac);
616/// assert!(syriac.contains32(0x0650)); // ARABIC KASRA
617/// assert!(!syriac.contains32(0x0660)); // ARABIC-INDIC DIGIT ZERO
618/// assert!(!syriac.contains32(0xFDF2)); // ARABIC LIGATURE ALLAH ISOLATED FORM
619/// assert!(syriac.contains32(0x0700)); // SYRIAC END OF PARAGRAPH
620/// assert!(syriac.contains32(0x074A)); // SYRIAC BARREKH
621/// ```
622#[cfg(feature = "compiled_data")]
623pub const fn script_with_extensions() -> ScriptWithExtensionsBorrowed<'static> {
624 ScriptWithExtensionsBorrowed {
625 data: crate::provider::Baked::SINGLETON_PROPS_SCX_V1,
626 }
627}
628
629icu_provider::gen_any_buffer_data_constructors!(
630 locale: skip,
631 options: skip,
632 result: Result<ScriptWithExtensions, PropertiesError>,
633 #[cfg(skip)]
634 functions: [
635 script_with_extensions,
636 load_script_with_extensions_with_any_provider,
637 load_script_with_extensions_with_buffer_provider,
638 load_script_with_extensions_unstable,
639 ]
640);
641
642#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, script_with_extensions)]
643pub fn load_script_with_extensions_unstable(
644 provider: &(impl DataProvider<ScriptWithExtensionsPropertyV1Marker> + ?Sized),
645) -> Result<ScriptWithExtensions, PropertiesError> {
646 Ok(ScriptWithExtensions::from_data(
647 provider
648 .load(Default::default())
649 .and_then(op:DataResponse::take_payload)?,
650 ))
651}
652