| 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 | //! Definitions of [Unicode Properties] and APIs for |
| 6 | //! retrieving property data in an appropriate data structure. |
| 7 | //! |
| 8 | //! This module is published as its own crate ([`icu_properties`](https://docs.rs/icu_properties/latest/icu_properties/)) |
| 9 | //! and as part of the [`icu`](https://docs.rs/icu/latest/icu/) crate. See the latter for more details on the ICU4X project. |
| 10 | //! |
| 11 | //! APIs that return a [`CodePointSetData`] exist for binary properties and certain enumerated |
| 12 | //! properties. See the [`sets`] module for more details. |
| 13 | //! |
| 14 | //! APIs that return a [`CodePointMapData`] exist for certain enumerated properties. See the |
| 15 | //! [`maps`] module for more details. |
| 16 | //! |
| 17 | //! # Examples |
| 18 | //! |
| 19 | //! ## Property data as `CodePointSetData`s |
| 20 | //! |
| 21 | //! ``` |
| 22 | //! use icu::properties::{maps, sets, GeneralCategory}; |
| 23 | //! |
| 24 | //! // A binary property as a `CodePointSetData` |
| 25 | //! |
| 26 | //! assert!(sets::emoji().contains('🎃' )); // U+1F383 JACK-O-LANTERN |
| 27 | //! assert!(!sets::emoji().contains('木' )); // U+6728 |
| 28 | //! |
| 29 | //! // An individual enumerated property value as a `CodePointSetData` |
| 30 | //! |
| 31 | //! let line_sep_data = maps::general_category() |
| 32 | //! .get_set_for_value(GeneralCategory::LineSeparator); |
| 33 | //! let line_sep = line_sep_data.as_borrowed(); |
| 34 | //! |
| 35 | //! assert!(line_sep.contains32(0x2028)); |
| 36 | //! assert!(!line_sep.contains32(0x2029)); |
| 37 | //! ``` |
| 38 | //! |
| 39 | //! ## Property data as `CodePointMapData`s |
| 40 | //! |
| 41 | //! ``` |
| 42 | //! use icu::properties::{maps, Script}; |
| 43 | //! |
| 44 | //! assert_eq!(maps::script().get('🎃' ), Script::Common); // U+1F383 JACK-O-LANTERN |
| 45 | //! assert_eq!(maps::script().get('木' ), Script::Han); // U+6728 |
| 46 | //! ``` |
| 47 | //! |
| 48 | //! [`ICU4X`]: ../icu/index.html |
| 49 | //! [Unicode Properties]: https://unicode-org.github.io/icu/userguide/strings/properties.html |
| 50 | //! [`CodePointSetData`]: crate::sets::CodePointSetData |
| 51 | //! [`CodePointMapData`]: crate::maps::CodePointMapData |
| 52 | //! [`sets`]: crate::sets |
| 53 | |
| 54 | // https://github.com/unicode-org/icu4x/blob/main/documents/process/boilerplate.md#library-annotations |
| 55 | #![cfg_attr (not(any(test, feature = "std" )), no_std)] |
| 56 | #![cfg_attr ( |
| 57 | not(test), |
| 58 | deny( |
| 59 | clippy::indexing_slicing, |
| 60 | clippy::unwrap_used, |
| 61 | clippy::expect_used, |
| 62 | clippy::panic, |
| 63 | clippy::exhaustive_structs, |
| 64 | clippy::exhaustive_enums, |
| 65 | missing_debug_implementations, |
| 66 | ) |
| 67 | )] |
| 68 | #![warn (missing_docs)] |
| 69 | |
| 70 | extern crate alloc; |
| 71 | |
| 72 | #[cfg (feature = "bidi" )] |
| 73 | pub mod bidi; |
| 74 | |
| 75 | mod error; |
| 76 | pub mod maps; |
| 77 | |
| 78 | // NOTE: The Pernosco debugger has special knowledge |
| 79 | // of the `CanonicalCombiningClass` struct inside the `props` |
| 80 | // module. Please do not change the crate-module-qualified |
| 81 | // name of that struct without coordination. |
| 82 | mod props; |
| 83 | |
| 84 | pub mod bidi_data; |
| 85 | pub mod exemplar_chars; |
| 86 | pub mod provider; |
| 87 | pub(crate) mod runtime; |
| 88 | #[allow (clippy::exhaustive_structs)] // TODO |
| 89 | pub mod script; |
| 90 | pub mod sets; |
| 91 | mod trievalue; |
| 92 | |
| 93 | pub use props::{ |
| 94 | BidiClass, CanonicalCombiningClass, EastAsianWidth, GeneralCategory, GeneralCategoryGroup, |
| 95 | GraphemeClusterBreak, HangulSyllableType, IndicSyllabicCategory, JoiningType, LineBreak, |
| 96 | Script, SentenceBreak, WordBreak, |
| 97 | }; |
| 98 | |
| 99 | /// Module for working with the names of property values |
| 100 | pub mod names { |
| 101 | pub use crate::props::{ |
| 102 | PropertyEnumToValueNameLinearMapper, PropertyEnumToValueNameLinearMapperBorrowed, |
| 103 | }; |
| 104 | pub use crate::props::{ |
| 105 | PropertyEnumToValueNameLinearTiny4Mapper, PropertyEnumToValueNameLinearTiny4MapperBorrowed, |
| 106 | }; |
| 107 | pub use crate::props::{ |
| 108 | PropertyEnumToValueNameSparseMapper, PropertyEnumToValueNameSparseMapperBorrowed, |
| 109 | }; |
| 110 | pub use crate::props::{PropertyValueNameToEnumMapper, PropertyValueNameToEnumMapperBorrowed}; |
| 111 | } |
| 112 | |
| 113 | pub use error::PropertiesError; |
| 114 | |
| 115 | #[doc (no_inline)] |
| 116 | pub use PropertiesError as Error; |
| 117 | |