model BidiAuxiliaryProperties as an EnumeratedProperty

Author: robertbastian

components/properties/src/bidi.rs

@@ -2,218 +2,68 @@
 // called LICENSE at the top level of the ICU4X source tree
 // (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
 
-//! Data and APIs for supporting specific Bidi properties data in an efficient structure.
-//!
-//! Supported properties are:
-//! - `Bidi_Paired_Bracket`
-//! - `Bidi_Paired_Bracket_Type`
-//! - `Bidi_Mirrored`
-//! - `Bidi_Mirroring_Glyph`
-
-use crate::provider::bidi::{
-    BidiAuxiliaryPropertiesV1, BidiAuxiliaryPropertiesV1Marker, CheckedBidiPairedBracketType,
-};
-
-use icu_provider::prelude::*;
-
-/// A wrapper around certain Bidi properties data.
-///
-/// Most useful methods are on [`BidiAuxiliaryPropertiesBorrowed`] obtained by calling [`BidiAuxiliaryProperties::as_borrowed()`]
-#[derive(Debug)]
-pub struct BidiAuxiliaryProperties {
-    data: DataPayload<BidiAuxiliaryPropertiesV1Marker>,
-}
-
-impl BidiAuxiliaryProperties {
-    /// Creates a [`BidiAuxiliaryPropertiesV1`] struct that represents the data for certain
-    /// Bidi properties.
-    ///
-    /// ✨ *Enabled with the `compiled_data` Cargo feature.*
-    ///
-    /// [📚 Help choosing a constructor](icu_provider::constructors)
-    ///
-    /// # Examples
-    /// ```
-    /// use icu::properties::bidi::BidiAuxiliaryProperties;
-    ///
-    /// let bidi_data = BidiAuxiliaryProperties::new();
-    ///
-    /// let open_paren = bidi_data.get32_mirroring_props('(' as u32);
-    /// assert_eq!(open_paren.mirroring_glyph, Some(')'));
-    /// assert_eq!(open_paren.mirrored, true);
-    /// ```
-    #[cfg(feature = "compiled_data")]
-    #[allow(clippy::new_ret_no_self)]
-    pub const fn new() -> BidiAuxiliaryPropertiesBorrowed<'static> {
-        BidiAuxiliaryPropertiesBorrowed {
-            data: crate::provider::Baked::SINGLETON_BIDI_AUXILIARY_PROPERTIES_V1_MARKER,
-        }
-    }
-
-    icu_provider::gen_any_buffer_data_constructors!(
-        () -> result: Result<BidiAuxiliaryProperties, DataError>,
-        functions: [
-            new: skip,
-            try_new_with_any_provider,
-            try_new_with_buffer_provider,
-            try_new_unstable,
-            Self
-        ]
-    );
-
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
-    pub fn try_new_unstable(
-        provider: &(impl DataProvider<BidiAuxiliaryPropertiesV1Marker> + ?Sized),
-    ) -> Result<BidiAuxiliaryProperties, DataError> {
-        Ok(BidiAuxiliaryProperties::from_data(
-            provider.load(Default::default())?.payload,
-        ))
-    }
-
-    /// Construct a borrowed version of this type that can be queried.
-    ///
-    /// This avoids a potential small underlying cost per API call by consolidating it
-    /// up front.
-    #[inline]
-    pub fn as_borrowed(&self) -> BidiAuxiliaryPropertiesBorrowed<'_> {
-        BidiAuxiliaryPropertiesBorrowed {
-            data: self.data.get(),
-        }
-    }
-
-    /// Construct a new one from loaded data
-    ///
-    /// Typically it is preferable to use getters like [`bidi_auxiliary_properties()`] instead
-    pub(crate) fn from_data(data: DataPayload<BidiAuxiliaryPropertiesV1Marker>) -> Self {
-        Self { data }
-    }
-}
-
-/// This struct represents the properties Bidi_Mirrored and Bidi_Mirroring_Glyph.
-/// If Bidi_Mirroring_Glyph is not defined for a code point, then the value in the
-/// struct is `None`.
-#[derive(Debug, Eq, PartialEq)]
+use crate::{props::EnumeratedProperty, provider::BidiMirroringGlyphV1Marker};
+use icu_collections::codepointtrie::TrieValue;
+use zerovec::ule::{AsULE, RawBytesULE};
+
+/// This is a bitpacked combination of the `Bidi_Mirroring_Glyph`,
+/// `Bidi_Mirrored`, and `Bidi_Paired_Bracket_Type` properties.
+#[derive(Debug, Eq, PartialEq, Clone, Copy, Default)]
+#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
+#[cfg_attr(feature = "datagen", databake(path = icu_properties::bidi))]
+#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
 #[non_exhaustive]
-pub struct BidiMirroringProperties {
-    /// Represents the Bidi_Mirroring_Glyph property value
+pub struct BidiMirroringGlyph {
+    /// The mirroring glyph
     pub mirroring_glyph: Option<char>,
-    /// Represents the Bidi_Mirrored property value
+    /// Whether the glyph is mirrored
     pub mirrored: bool,
+    /// The paired bracket type
+    pub paired_bracket_type: BidiPairedBracketType,
 }
 
-/// The enum represents Bidi_Paired_Bracket_Type, the char represents Bidi_Paired_Bracket.
-/// Bidi_Paired_Bracket has a value of `None` when Bidi_Paired_Bracket_Type is `None`.
-#[derive(Debug, Eq, PartialEq)]
-#[non_exhaustive]
-pub enum BidiPairingProperties {
-    /// Represents Bidi_Paired_Bracket_Type=Open, and the Bidi_Paired_Bracket value for that code point.
-    Open(char),
-    /// Represents Bidi_Paired_Bracket_Type=Close, and the Bidi_Paired_Bracket value for that code point.
-    Close(char),
-    /// Represents Bidi_Paired_Bracket_Type=None, which cooccurs with Bidi_Paired_Bracket
-    /// being undefined for that code point.
-    None,
+impl EnumeratedProperty for BidiMirroringGlyph {
+    type DataMarker = BidiMirroringGlyphV1Marker;
+    #[cfg(feature = "compiled_data")]
+    const SINGLETON: &'static crate::provider::PropertyCodePointMapV1<'static, Self> =
+        crate::provider::Baked::SINGLETON_BIDI_MIRRORING_GLYPH_V1_MARKER;
 }
 
-/// A borrowed wrapper around Bidi properties data, returned by
-/// [`BidiAuxiliaryProperties::as_borrowed()`]. More efficient to query.
-#[derive(Debug)]
-pub struct BidiAuxiliaryPropertiesBorrowed<'a> {
-    data: &'a BidiAuxiliaryPropertiesV1<'a>,
-}
+impl crate::private::Sealed for BidiMirroringGlyph {}
 
-impl<'a> BidiAuxiliaryPropertiesBorrowed<'a> {
-    // The source data coming from icuexportdata will use 0 to represent the
-    // property value in cases for which the Bidi_Mirroring_Glyph property value
-    // of a code point is undefined. Since Rust types can be more expressive, we
-    // should represent these cases as None.
-    fn convert_mirroring_glyph_data(trie_data_char: char) -> Option<char> {
-        if trie_data_char as u32 == 0 {
-            None
-        } else {
-            Some(trie_data_char)
-        }
-    }
+impl AsULE for BidiMirroringGlyph {
+    type ULE = zerovec::ule::RawBytesULE<3>;
 
-    /// Return a struct for the given code point representing Bidi mirroring-related
-    /// property values. See [`BidiMirroringProperties`].
-    ///
-    /// # Examples
-    /// ```
-    /// use icu::properties::bidi::BidiAuxiliaryProperties;
-    ///
-    /// let bidi_data = BidiAuxiliaryProperties::new();
-    ///
-    /// let open_paren = bidi_data.get32_mirroring_props('(' as u32);
-    /// assert_eq!(open_paren.mirroring_glyph, Some(')'));
-    /// assert_eq!(open_paren.mirrored, true);
-    /// let close_paren = bidi_data.get32_mirroring_props(')' as u32);
-    /// assert_eq!(close_paren.mirroring_glyph, Some('('));
-    /// assert_eq!(close_paren.mirrored, true);
-    /// let open_angle_bracket = bidi_data.get32_mirroring_props('<' as u32);
-    /// assert_eq!(open_angle_bracket.mirroring_glyph, Some('>'));
-    /// assert_eq!(open_angle_bracket.mirrored, true);
-    /// let close_angle_bracket = bidi_data.get32_mirroring_props('>' as u32);
-    /// assert_eq!(close_angle_bracket.mirroring_glyph, Some('<'));
-    /// assert_eq!(close_angle_bracket.mirrored, true);
-    /// let three = bidi_data.get32_mirroring_props('3' as u32);
-    /// assert_eq!(three.mirroring_glyph, None);
-    /// assert_eq!(three.mirrored, false);
-    /// ```
-    pub fn get32_mirroring_props(&self, code_point: u32) -> BidiMirroringProperties {
-        let BidiAuxiliaryPropertiesV1::CodePointTrie(trie) = self.data;
-        let bidi_aux_props = trie.get32(code_point);
-        let mirroring_glyph_opt =
-            Self::convert_mirroring_glyph_data(bidi_aux_props.mirroring_glyph);
-        BidiMirroringProperties {
-            mirroring_glyph: mirroring_glyph_opt,
-            mirrored: bidi_aux_props.mirrored,
-        }
+    fn to_unaligned(self) -> Self::ULE {
+        let [a, b, c, _] = TrieValue::to_u32(self).to_le_bytes();
+        RawBytesULE([a, b, c])
     }
-
-    /// Return a struct for the given code point representing Bidi bracket
-    /// pairing-related property values. See [`BidiPairingProperties`]
-    ///
-    /// # Examples
-    /// ```
-    /// use icu::properties::bidi::{BidiAuxiliaryProperties, BidiPairingProperties};
-    ///
-    /// let bidi_data = BidiAuxiliaryProperties::new();
-    ///
-    /// let open_paren = bidi_data.get32_pairing_props('(' as u32);
-    /// assert_eq!(open_paren, BidiPairingProperties::Open(')'));
-    /// let close_paren = bidi_data.get32_pairing_props(')' as u32);
-    /// assert_eq!(close_paren, BidiPairingProperties::Close('('));
-    /// let open_angle_bracket = bidi_data.get32_pairing_props('<' as u32);
-    /// assert_eq!(open_angle_bracket, BidiPairingProperties::None);
-    /// let close_angle_bracket = bidi_data.get32_pairing_props('>' as u32);
-    /// assert_eq!(close_angle_bracket, BidiPairingProperties::None);
-    /// let three = bidi_data.get32_pairing_props('3' as u32);
-    /// assert_eq!(three, BidiPairingProperties::None);
-    /// ```
-    pub fn get32_pairing_props(&self, code_point: u32) -> BidiPairingProperties {
-        let BidiAuxiliaryPropertiesV1::CodePointTrie(trie) = self.data;
-        let bidi_aux_props = trie.get32(code_point);
-        let mirroring_glyph = bidi_aux_props.mirroring_glyph;
-        let paired_bracket_type = bidi_aux_props.paired_bracket_type;
-        match paired_bracket_type {
-            CheckedBidiPairedBracketType::Open => BidiPairingProperties::Open(mirroring_glyph),
-            CheckedBidiPairedBracketType::Close => BidiPairingProperties::Close(mirroring_glyph),
-            _ => BidiPairingProperties::None,
-        }
+    fn from_unaligned(unaligned: Self::ULE) -> Self {
+        let [a, b, c] = unaligned.0;
+        TrieValue::try_from_u32(u32::from_le_bytes([a, b, c, 0])).unwrap_or_default()
     }
 }
 
-impl BidiAuxiliaryPropertiesBorrowed<'static> {
-    /// Cheaply converts a [`BidiAuxiliaryPropertiesBorrowed<'static>`] into a [`BidiAuxiliaryProperties`].
-    ///
-    /// Note: Due to branching and indirection, using [`BidiAuxiliaryProperties`] might inhibit some
-    /// compile-time optimizations that are possible with [`BidiAuxiliaryPropertiesBorrowed`].
-    pub const fn static_to_owned(self) -> BidiAuxiliaryProperties {
-        BidiAuxiliaryProperties {
-            data: DataPayload::from_static_ref(self.data),
-        }
-    }
+/// The enum represents Bidi_Paired_Bracket_Type.
+///
+/// It does not implement [`EnumeratedProperty`], instead it can be obtained
+/// through the bitpacked [`BidiMirroringGlyph`] property.
+///
+/// If you have a use case this property without also needing the [`BidiMirroringGlyph`]
+/// property, and need to optimize data size, please file an issue.
+#[derive(Debug, Eq, PartialEq, Copy, Clone, Default)]
+#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
+#[cfg_attr(feature = "datagen", databake(path = icu_properties::bidi))]
+#[cfg_attr(feature = "serde", derive(serde::Deserialize))]
+#[non_exhaustive]
+pub enum BidiPairedBracketType {
+    /// Represents Bidi_Paired_Bracket_Type=Open.
+    Open,
+    /// Represents Bidi_Paired_Bracket_Type=Close.
+    Close,
+    /// Represents Bidi_Paired_Bracket_Type=None.
+    #[default]
+    None,
 }
 
 /// Implements [`unicode_bidi::BidiDataSource`] on [`CodePointMapDataBorrowed<BidiClass>`](crate::CodePointMapDataBorrowed).

components/properties/src/lib.rs

@@ -83,7 +83,6 @@ pub use names::{
 mod runtime;
 pub use runtime::UnicodeProperty;
 
-pub mod bidi;
 // NOTE: The Pernosco debugger has special knowledge
 // of the `CanonicalCombiningClass` struct inside the `props`
 // module. Please do not change the crate-module-qualified
@@ -92,6 +91,7 @@ pub mod props;
 pub mod provider;
 pub mod script;
 
+mod bidi;
 mod trievalue;
 
 mod private {

components/properties/src/props.rs

@@ -19,6 +19,8 @@
 
 pub use crate::names::{NamedEnumeratedProperty, ParseableEnumeratedProperty};
 
+pub use crate::bidi::{BidiMirroringGlyph, BidiPairedBracketType};
+
 /// See [`test_enumerated_property_completeness`] for usage.
 /// Example input:
 /// ```ignore

components/properties/src/provider.rs

@@ -15,7 +15,6 @@
 //!
 //! Read more about data providers: [`icu_provider`]
 
-pub mod bidi;
 pub mod names;
 
 pub use names::{
@@ -40,18 +39,15 @@ pub use names::{
     WordBreakValueToShortNameV1Marker,
 };
 
+use crate::bidi::BidiMirroringGlyph;
 pub use crate::props::gc::GeneralCategoryULE;
-pub use bidi::BidiAuxiliaryPropertiesV1Marker;
-
 use crate::script::ScriptWithExt;
-
 use core::ops::RangeInclusive;
 use icu_collections::codepointinvlist::CodePointInversionList;
 use icu_collections::codepointinvliststringlist::CodePointInversionListAndStringList;
 use icu_collections::codepointtrie::{CodePointMapRange, CodePointTrie, TrieValue};
 use icu_provider::prelude::*;
 use zerofrom::ZeroFrom;
-
 use zerovec::{ule::UleError, VarZeroVec, ZeroSlice};
 
 #[cfg(feature = "compiled_data")]
@@ -78,13 +74,13 @@ const _: () = {
     impl_alphabetic_v1_marker!(Baked);
     impl_ascii_hex_digit_v1_marker!(Baked);
     impl_basic_emoji_v1_marker!(Baked);
-    impl_bidi_auxiliary_properties_v1_marker!(Baked);
     impl_bidi_class_name_to_value_v1_marker!(Baked);
     impl_bidi_class_v1_marker!(Baked);
     impl_bidi_class_value_to_long_name_v1_marker!(Baked);
     impl_bidi_class_value_to_short_name_v1_marker!(Baked);
     impl_bidi_control_v1_marker!(Baked);
     impl_bidi_mirrored_v1_marker!(Baked);
+    impl_bidi_mirroring_glyph_v1_marker!(Baked);
     impl_blank_v1_marker!(Baked);
     impl_canonical_combining_class_name_to_value_v1_marker!(Baked);
     impl_canonical_combining_class_v1_marker!(Baked);
@@ -199,9 +195,9 @@ pub const MARKERS: &[DataMarkerInfo] = &[
     AlphabeticV1Marker::INFO,
     AsciiHexDigitV1Marker::INFO,
     BasicEmojiV1Marker::INFO,
-    bidi::BidiAuxiliaryPropertiesV1Marker::INFO,
     BidiControlV1Marker::INFO,
     BidiMirroredV1Marker::INFO,
+    BidiMirroringGlyphV1Marker::INFO,
     BlankV1Marker::INFO,
     CasedV1Marker::INFO,
     CaseIgnorableV1Marker::INFO,
@@ -536,6 +532,11 @@ data_struct_generic!(
     marker(ScriptV1Marker, Script, "props/sc@1"),
     marker(SentenceBreakV1Marker, SentenceBreak, "props/SB@1"),
     marker(WordBreakV1Marker, WordBreak, "props/WB@1"),
+    marker(
+        BidiMirroringGlyphV1Marker,
+        BidiMirroringGlyph,
+        "props/Bidi_G@1"
+    ),
 );
 
 // See CodePointMapData for documentation of these functions