icu_properties/
bidi.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
5use crate::{props::EnumeratedProperty, provider::PropertyEnumBidiMirroringGlyphV1};
6use icu_collections::codepointtrie::TrieValue;
7use zerovec::ule::{AsULE, RawBytesULE};
8
9/// This is a bitpacked combination of the `Bidi_Mirroring_Glyph`,
10/// `Bidi_Mirrored`, and `Bidi_Paired_Bracket_Type` properties.
11#[derive(Debug, Eq, PartialEq, Clone, Copy, Default)]
12#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
13#[cfg_attr(feature = "datagen", databake(path = icu_properties::props))]
14#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
15#[allow(clippy::exhaustive_structs)] // needed for baked construction
16pub struct BidiMirroringGlyph {
17    /// The mirroring glyph
18    pub mirroring_glyph: Option<char>,
19    /// Whether the glyph is mirrored
20    pub mirrored: bool,
21    /// The paired bracket type
22    pub paired_bracket_type: BidiPairedBracketType,
23}
24
25impl EnumeratedProperty for BidiMirroringGlyph {
26    type DataMarker = PropertyEnumBidiMirroringGlyphV1;
27    #[cfg(feature = "compiled_data")]
28    const SINGLETON: &'static crate::provider::PropertyCodePointMap<'static, Self> =
29        crate::provider::Baked::SINGLETON_PROPERTY_ENUM_BIDI_MIRRORING_GLYPH_V1;
30    const NAME: &'static [u8] = b"Bidi_Mirroring_Glyph";
31    const SHORT_NAME: &'static [u8] = b"bmg";
32}
33
34impl crate::private::Sealed for BidiMirroringGlyph {}
35
36impl AsULE for BidiMirroringGlyph {
37    type ULE = zerovec::ule::RawBytesULE<3>;
38
39    fn to_unaligned(self) -> Self::ULE {
40        let [a, b, c, _] = TrieValue::to_u32(self).to_le_bytes();
41        RawBytesULE([a, b, c])
42    }
43    fn from_unaligned(unaligned: Self::ULE) -> Self {
44        let [a, b, c] = unaligned.0;
45        TrieValue::try_from_u32(u32::from_le_bytes([a, b, c, 0])).unwrap_or_default()
46    }
47}
48
49/// The enum represents Bidi_Paired_Bracket_Type.
50///
51/// It does not implement [`EnumeratedProperty`], instead it can be obtained
52/// through the bitpacked [`BidiMirroringGlyph`] property.
53///
54/// If you have a use case this property without also needing the [`BidiMirroringGlyph`]
55/// property, and need to optimize data size, please file an issue.
56#[derive(Debug, Eq, PartialEq, Copy, Clone, Default)]
57#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
58#[cfg_attr(feature = "datagen", databake(path = icu_properties::props))]
59#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
60#[non_exhaustive]
61pub enum BidiPairedBracketType {
62    /// Represents Bidi_Paired_Bracket_Type=Open.
63    Open,
64    /// Represents Bidi_Paired_Bracket_Type=Close.
65    Close,
66    /// Represents Bidi_Paired_Bracket_Type=None.
67    #[default]
68    None,
69}
70
71#[cfg(feature = "unicode_bidi")]
72use crate::props::BidiClass;
73
74/// ✨ *Enabled with the `unicode_bidi` Cargo feature.*
75///
76/// # Examples
77///
78///```
79/// use icu::properties::CodePointMapData;
80/// use icu::properties::props::BidiClass;
81/// use unicode_bidi::BidiInfo;
82///
83/// // This example text is defined using `concat!` because some browsers
84/// // and text editors have trouble displaying bidi strings.
85/// let text =  concat!["א", // RTL#1
86///                     "ב", // RTL#2
87///                     "ג", // RTL#3
88///                     "a", // LTR#1
89///                     "b", // LTR#2
90///                     "c", // LTR#3
91///                     ]; //
92///
93///
94/// let bidi_map = CodePointMapData::<BidiClass>::new();
95///
96/// // Resolve embedding levels within the text.  Pass `None` to detect the
97/// // paragraph level automatically.
98/// let bidi_info = BidiInfo::new_with_data_source(&bidi_map, text, None);
99///
100/// // This paragraph has embedding level 1 because its first strong character is RTL.
101/// assert_eq!(bidi_info.paragraphs.len(), 1);
102/// let para = &bidi_info.paragraphs[0];
103/// assert_eq!(para.level.number(), 1);
104/// assert!(para.level.is_rtl());
105///
106/// // Re-ordering is done after wrapping each paragraph into a sequence of
107/// // lines. For this example, I'll just use a single line that spans the
108/// // entire paragraph.
109/// let line = para.range.clone();
110///
111/// let display = bidi_info.reorder_line(para, line);
112/// assert_eq!(display, concat!["a", // LTR#1
113///                             "b", // LTR#2
114///                             "c", // LTR#3
115///                             "ג", // RTL#3
116///                             "ב", // RTL#2
117///                             "א", // RTL#1
118///                             ]);
119/// ```
120#[cfg(feature = "unicode_bidi")]
121impl unicode_bidi::data_source::BidiDataSource for crate::CodePointMapDataBorrowed<'_, BidiClass> {
122    fn bidi_class(&self, c: char) -> unicode_bidi::BidiClass {
123        self.get(c).into()
124    }
125}
126
127#[cfg(feature = "unicode_bidi")]
128impl From<BidiClass> for unicode_bidi::BidiClass {
129    fn from(value: BidiClass) -> Self {
130        match value {
131            BidiClass::LeftToRight => unicode_bidi::BidiClass::L,
132            BidiClass::RightToLeft => unicode_bidi::BidiClass::R,
133            BidiClass::EuropeanNumber => unicode_bidi::BidiClass::EN,
134            BidiClass::EuropeanSeparator => unicode_bidi::BidiClass::ES,
135            BidiClass::EuropeanTerminator => unicode_bidi::BidiClass::ET,
136            BidiClass::ArabicNumber => unicode_bidi::BidiClass::AN,
137            BidiClass::CommonSeparator => unicode_bidi::BidiClass::CS,
138            BidiClass::ParagraphSeparator => unicode_bidi::BidiClass::B,
139            BidiClass::SegmentSeparator => unicode_bidi::BidiClass::S,
140            BidiClass::WhiteSpace => unicode_bidi::BidiClass::WS,
141            BidiClass::OtherNeutral => unicode_bidi::BidiClass::ON,
142            BidiClass::LeftToRightEmbedding => unicode_bidi::BidiClass::LRE,
143            BidiClass::LeftToRightOverride => unicode_bidi::BidiClass::LRO,
144            BidiClass::ArabicLetter => unicode_bidi::BidiClass::AL,
145            BidiClass::RightToLeftEmbedding => unicode_bidi::BidiClass::RLE,
146            BidiClass::RightToLeftOverride => unicode_bidi::BidiClass::RLO,
147            BidiClass::PopDirectionalFormat => unicode_bidi::BidiClass::PDF,
148            BidiClass::NonspacingMark => unicode_bidi::BidiClass::NSM,
149            BidiClass::BoundaryNeutral => unicode_bidi::BidiClass::BN,
150            BidiClass::FirstStrongIsolate => unicode_bidi::BidiClass::FSI,
151            BidiClass::LeftToRightIsolate => unicode_bidi::BidiClass::LRI,
152            BidiClass::RightToLeftIsolate => unicode_bidi::BidiClass::RLI,
153            BidiClass::PopDirectionalIsolate => unicode_bidi::BidiClass::PDI,
154            // This must not happen.
155            _ => unicode_bidi::BidiClass::ON,
156        }
157    }
158}
159
160#[cfg(feature = "unicode_bidi")]
161impl From<unicode_bidi::BidiClass> for BidiClass {
162    fn from(value: unicode_bidi::BidiClass) -> Self {
163        match value {
164            unicode_bidi::BidiClass::L => BidiClass::LeftToRight,
165            unicode_bidi::BidiClass::R => BidiClass::RightToLeft,
166            unicode_bidi::BidiClass::EN => BidiClass::EuropeanNumber,
167            unicode_bidi::BidiClass::ES => BidiClass::EuropeanSeparator,
168            unicode_bidi::BidiClass::ET => BidiClass::EuropeanTerminator,
169            unicode_bidi::BidiClass::AN => BidiClass::ArabicNumber,
170            unicode_bidi::BidiClass::CS => BidiClass::CommonSeparator,
171            unicode_bidi::BidiClass::B => BidiClass::ParagraphSeparator,
172            unicode_bidi::BidiClass::S => BidiClass::SegmentSeparator,
173            unicode_bidi::BidiClass::WS => BidiClass::WhiteSpace,
174            unicode_bidi::BidiClass::ON => BidiClass::OtherNeutral,
175            unicode_bidi::BidiClass::LRE => BidiClass::LeftToRightEmbedding,
176            unicode_bidi::BidiClass::LRO => BidiClass::LeftToRightOverride,
177            unicode_bidi::BidiClass::AL => BidiClass::ArabicLetter,
178            unicode_bidi::BidiClass::RLE => BidiClass::RightToLeftEmbedding,
179            unicode_bidi::BidiClass::RLO => BidiClass::RightToLeftOverride,
180            unicode_bidi::BidiClass::PDF => BidiClass::PopDirectionalFormat,
181            unicode_bidi::BidiClass::NSM => BidiClass::NonspacingMark,
182            unicode_bidi::BidiClass::BN => BidiClass::BoundaryNeutral,
183            unicode_bidi::BidiClass::FSI => BidiClass::FirstStrongIsolate,
184            unicode_bidi::BidiClass::LRI => BidiClass::LeftToRightIsolate,
185            unicode_bidi::BidiClass::RLI => BidiClass::RightToLeftIsolate,
186            unicode_bidi::BidiClass::PDI => BidiClass::PopDirectionalIsolate,
187        }
188    }
189}
icu_properties/bidi.rs

icu_properties/
bidi.rs
