|
| 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 | +//! This module exposes tooling for running the [unicode bidi algorithm](https://unicode.org/reports/tr9/) using ICU4X data. |
| 6 | +//! |
| 7 | +//! `BidiClassAdapter` enables ICU4X to provide data to [`unicode-bidi`]. |
| 8 | +//! |
| 9 | +//! # Examples |
| 10 | +//! |
| 11 | +//!``` |
| 12 | +//! use icu_properties::bidi::BidiClassAdapter; |
| 13 | +//! use icu_properties::{maps, BidiClass}; |
| 14 | +//! use icu_codepointtrie::CodePointTrie; |
| 15 | +//! use unicode_bidi::BidiClass as DataSourceBidiClass; |
| 16 | +//! use unicode_bidi::BidiDataSource; |
| 17 | +//! use unicode_bidi::BidiInfo; |
| 18 | +//! // This example text is defined using `concat!` because some browsers |
| 19 | +//! // and text editors have trouble displaying bidi strings. |
| 20 | +//! let text = concat![ |
| 21 | +//! "א", |
| 22 | +//! "ב", |
| 23 | +//! "ג", |
| 24 | +//! "a", |
| 25 | +//! "b", |
| 26 | +//! "c", |
| 27 | +//! ]; |
| 28 | +//! |
| 29 | +//! // Create an adapter to provide the data to `BidiInfo`. |
| 30 | +//! let provider = icu_testdata::get_provider(); |
| 31 | +//! |
| 32 | +//! let payload = |
| 33 | +//! maps::get_bidi_class(&provider) |
| 34 | +//! .expect("The data should be valid"); |
| 35 | +//! let data_struct = payload.get(); |
| 36 | +//! let bc = &data_struct.code_point_trie; |
| 37 | +//! |
| 38 | +//! let adapter = BidiClassAdapter::new(&bc); |
| 39 | +//! // Resolve embedding levels within the text. Pass `None` to detect the |
| 40 | +//! // paragraph level automatically. |
| 41 | +//! |
| 42 | +//! let bidi_info = BidiInfo::new_with_data_source(&adapter, &text, None); |
| 43 | +//! |
| 44 | +//! // This paragraph has embedding level 1 because its first strong character is RTL. |
| 45 | +//! assert_eq!(bidi_info.paragraphs.len(), 1); |
| 46 | +//! let para = &bidi_info.paragraphs[0]; |
| 47 | +//! assert_eq!(para.level.number(), 1); |
| 48 | +//! assert_eq!(para.level.is_rtl(), true); |
| 49 | +//! |
| 50 | +//! // Re-ordering is done after wrapping each paragraph into a sequence of |
| 51 | +//! // lines. For this example, I'll just use a single line that spans the |
| 52 | +//! // entire paragraph. |
| 53 | +//! let line = para.range.clone(); |
| 54 | +//! |
| 55 | +//! let display = bidi_info.reorder_line(para, line); |
| 56 | +//! assert_eq!(display, concat![ |
| 57 | +//! "a", |
| 58 | +//! "b", |
| 59 | +//! "c", |
| 60 | +//! "ג", |
| 61 | +//! "ב", |
| 62 | +//! "א", |
| 63 | +//! ]); |
| 64 | +//! ``` |
| 65 | +
|
| 66 | +use crate::props::BidiClass; |
| 67 | +use icu_codepointtrie::CodePointTrie; |
| 68 | +use unicode_bidi::data_source::BidiDataSource; |
| 69 | +use unicode_bidi::BidiClass as DataSourceBidiClass; |
| 70 | + |
| 71 | +/// An adapter to convert from icu4x `BidiClass` to `unicode_bidi::BidiClass`. |
| 72 | +/// |
| 73 | +/// # Example |
| 74 | +/// |
| 75 | +/// ``` |
| 76 | +/// use icu_properties::bidi::BidiClassAdapter; |
| 77 | +/// use icu_properties::{maps, BidiClass}; |
| 78 | +/// use icu_codepointtrie::CodePointTrie; |
| 79 | +/// use unicode_bidi::BidiClass as DataSourceBidiClass; |
| 80 | +/// use unicode_bidi::BidiDataSource; |
| 81 | +/// |
| 82 | +/// let provider = icu_testdata::get_provider(); |
| 83 | +/// |
| 84 | +/// let payload = |
| 85 | +/// maps::get_bidi_class(&provider) |
| 86 | +/// .expect("The data should be valid"); |
| 87 | +/// let data_struct = payload.get(); |
| 88 | +/// let bc = &data_struct.code_point_trie; |
| 89 | +/// |
| 90 | +/// let adapter = BidiClassAdapter::new(&bc); |
| 91 | +/// assert_eq!(adapter.bidi_class('a'), DataSourceBidiClass::L); |
| 92 | +/// assert_eq!(adapter.bidi_class('ع'), DataSourceBidiClass::AL); |
| 93 | +/// ``` |
| 94 | +pub struct BidiClassAdapter<'a> { |
| 95 | + bidi_trie: &'a CodePointTrie<'a, BidiClass>, |
| 96 | +} |
| 97 | + |
| 98 | +impl<'a> BidiClassAdapter<'a> { |
| 99 | + /// Creates new instance of `BidiClassAdapter`. |
| 100 | + pub fn new(bidi_trie: &'a CodePointTrie<'a, BidiClass>) -> BidiClassAdapter<'a> { |
| 101 | + BidiClassAdapter { bidi_trie } |
| 102 | + } |
| 103 | +} |
| 104 | + |
| 105 | +impl<'a> BidiDataSource for BidiClassAdapter<'a> { |
| 106 | + /// Returns a [`DataSourceBidiClass`] given a unicode character. |
| 107 | + /// |
| 108 | + /// # Example |
| 109 | + /// |
| 110 | + /// ``` |
| 111 | + /// use icu_properties::bidi::BidiClassAdapter; |
| 112 | + /// use icu_properties::{maps, BidiClass}; |
| 113 | + /// use icu_codepointtrie::CodePointTrie; |
| 114 | + /// use unicode_bidi::BidiClass as DataSourceBidiClass; |
| 115 | + /// use unicode_bidi::BidiDataSource; |
| 116 | + /// |
| 117 | + /// let provider = icu_testdata::get_provider(); |
| 118 | + /// |
| 119 | + /// let payload = |
| 120 | + /// maps::get_bidi_class(&provider) |
| 121 | + /// .expect("The data should be valid"); |
| 122 | + /// let data_struct = payload.get(); |
| 123 | + /// let bc = &data_struct.code_point_trie; |
| 124 | + /// |
| 125 | + /// let adapter = BidiClassAdapter::new(&bc); |
| 126 | + /// assert_eq!(adapter.bidi_class('a'), DataSourceBidiClass::L); |
| 127 | + /// ``` |
| 128 | + /// |
| 129 | + /// [`CodePointTrie`]: icu_codepointtrie::CodePointTrie |
| 130 | + fn bidi_class(&self, c: char) -> DataSourceBidiClass { |
| 131 | + let bidi_class = self.bidi_trie.get(c as u32); |
| 132 | + match bidi_class { |
| 133 | + BidiClass::LeftToRight => DataSourceBidiClass::L, |
| 134 | + BidiClass::RightToLeft => DataSourceBidiClass::R, |
| 135 | + BidiClass::EuropeanNumber => DataSourceBidiClass::EN, |
| 136 | + BidiClass::EuropeanSeparator => DataSourceBidiClass::ES, |
| 137 | + BidiClass::EuropeanTerminator => DataSourceBidiClass::ET, |
| 138 | + BidiClass::ArabicNumber => DataSourceBidiClass::AN, |
| 139 | + BidiClass::CommonSeparator => DataSourceBidiClass::CS, |
| 140 | + BidiClass::ParagraphSeparator => DataSourceBidiClass::B, |
| 141 | + BidiClass::SegmentSeparator => DataSourceBidiClass::S, |
| 142 | + BidiClass::WhiteSpace => DataSourceBidiClass::WS, |
| 143 | + BidiClass::OtherNeutral => DataSourceBidiClass::ON, |
| 144 | + BidiClass::LeftToRightEmbedding => DataSourceBidiClass::LRE, |
| 145 | + BidiClass::LeftToRightOverride => DataSourceBidiClass::LRO, |
| 146 | + BidiClass::ArabicLetter => DataSourceBidiClass::AL, |
| 147 | + BidiClass::RightToLeftEmbedding => DataSourceBidiClass::RLE, |
| 148 | + BidiClass::RightToLeftOverride => DataSourceBidiClass::RLO, |
| 149 | + BidiClass::PopDirectionalFormat => DataSourceBidiClass::PDF, |
| 150 | + BidiClass::NonspacingMark => DataSourceBidiClass::NSM, |
| 151 | + BidiClass::BoundaryNeutral => DataSourceBidiClass::BN, |
| 152 | + BidiClass::FirstStrongIsolate => DataSourceBidiClass::FSI, |
| 153 | + BidiClass::LeftToRightIsolate => DataSourceBidiClass::LRI, |
| 154 | + BidiClass::RightToLeftIsolate => DataSourceBidiClass::RLI, |
| 155 | + BidiClass::PopDirectionalIsolate => DataSourceBidiClass::PDI, |
| 156 | + _ => |
| 157 | + // This must not happen. |
| 158 | + { |
| 159 | + DataSourceBidiClass::ON |
| 160 | + } |
| 161 | + } |
| 162 | + } |
| 163 | +} |
0 commit comments