add enumerated property for Numeric_Type

Author: m4rch3n1ng

components/properties/src/names.rs

@@ -807,6 +807,14 @@ impl_value_getter! {
     }
 }
 
+impl_value_getter! {
+    impl NumericType {
+        PropertyNameParseNumericTypeV1 / SINGLETON_PROPERTY_NAME_PARSE_NUMERIC_TYPE_V1;
+        PropertyEnumToValueNameLinearMap / PropertyNameShortNumericTypeV1 / SINGLETON_PROPERTY_NAME_SHORT_NUMERIC_TYPE_V1;
+        PropertyEnumToValueNameLinearMap / PropertyNameLongNumericTypeV1 / SINGLETON_PROPERTY_NAME_LONG_NUMERIC_TYPE_V1;
+    }
+}
+
 impl_value_getter! {
     impl GeneralCategory {
         PropertyNameParseGeneralCategoryV1 / SINGLETON_PROPERTY_NAME_PARSE_GENERAL_CATEGORY_V1;

components/properties/src/props.rs

@@ -112,6 +112,71 @@ macro_rules! make_enumerated_property {
     };
 }
 
+/// Enumerated property Numeric_Type.
+///
+/// See Section 4.6, Numeric Value in The Unicode Standard for the summary of
+/// each property value.
+///
+/// # Example
+///
+/// ```
+/// use icu::properties::{props::NumericType, CodePointMapData};
+///
+/// assert_eq!(
+///     CodePointMapData::<NumericType>::new().get('0'),
+///     NumericType::Decimal,
+/// ); // U+0030
+/// assert_eq!(
+///     CodePointMapData::<NumericType>::new().get('½'),
+///     NumericType::Numeric,
+/// ); // U+00BD
+/// ```
+#[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd, Hash)]
+#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
+#[allow(clippy::exhaustive_structs)] // newtype
+#[repr(transparent)]
+pub struct NumericType(pub(crate) u8);
+
+impl NumericType {
+    /// Returns an ICU4C `UNumericType` value.
+    pub const fn to_icu4c_value(self) -> u8 {
+        self.0
+    }
+    /// Constructor from an ICU4C `UNumericType` value.
+    pub const fn from_icu4c_value(value: u8) -> Self {
+        Self(value)
+    }
+}
+
+create_const_array! {
+#[allow(non_upper_case_globals)]
+impl NumericType {
+    /// Characters without numeric value
+    pub const None: NumericType = NumericType(0);
+    /// (`De`) Characters of positional decimal systems
+    ///
+    /// These are coextensive with [`GeneralCategory::DecimalNumber`].
+    pub const Decimal: NumericType = NumericType(1);
+    /// (`Di`) Variants of positional or sequences thereof.
+    ///
+    /// The distinction between [`NumericType::Digit`] and [`NumericType::Numeric`]
+    /// has not proven to be useful, so no further characters will be added to
+    /// this type.
+    pub const Digit: NumericType = NumericType(2);
+    /// (`Nu`) Other characters with numeric value
+    pub const Numeric: NumericType = NumericType(3);
+}
+}
+
+make_enumerated_property! {
+    name: "Numeric_Type";
+    short_name: "nt";
+    ident: NumericType;
+    data_marker: crate::provider::PropertyEnumNumericTypeV1;
+    singleton: SINGLETON_PROPERTY_ENUM_NUMERIC_TYPE_V1;
+    ule_ty: u8;
+}
+
 /// Enumerated property Bidi_Class
 ///
 /// These are the categories required by the Unicode Bidirectional Algorithm.
@@ -3303,6 +3368,14 @@ mod test_enumerated_property_completeness {
         );
     }
 
+    #[test]
+    fn test_nt() {
+        check_enum(
+            crate::provider::Baked::SINGLETON_PROPERTY_NAME_PARSE_NUMERIC_TYPE_V1,
+            NumericType::ALL_VALUES,
+        );
+    }
+
     #[test]
     fn test_hst() {
         check_enum(

components/properties/src/provider.rs

@@ -26,21 +26,22 @@ pub use names::{
     PropertyNameLongBidiClassV1, PropertyNameLongEastAsianWidthV1,
     PropertyNameLongGeneralCategoryV1, PropertyNameLongGraphemeClusterBreakV1,
     PropertyNameLongHangulSyllableTypeV1, PropertyNameLongIndicSyllabicCategoryV1,
-    PropertyNameLongJoiningTypeV1, PropertyNameLongLineBreakV1, PropertyNameLongScriptV1,
-    PropertyNameLongSentenceBreakV1, PropertyNameLongVerticalOrientationV1,
-    PropertyNameLongWordBreakV1, PropertyNameParseBidiClassV1,
-    PropertyNameParseCanonicalCombiningClassV1, PropertyNameParseEastAsianWidthV1,
-    PropertyNameParseGeneralCategoryMaskV1, PropertyNameParseGeneralCategoryV1,
-    PropertyNameParseGraphemeClusterBreakV1, PropertyNameParseHangulSyllableTypeV1,
-    PropertyNameParseIndicSyllabicCategoryV1, PropertyNameParseJoiningTypeV1,
-    PropertyNameParseLineBreakV1, PropertyNameParseScriptV1, PropertyNameParseSentenceBreakV1,
+    PropertyNameLongJoiningTypeV1, PropertyNameLongLineBreakV1, PropertyNameLongNumericTypeV1,
+    PropertyNameLongScriptV1, PropertyNameLongSentenceBreakV1,
+    PropertyNameLongVerticalOrientationV1, PropertyNameLongWordBreakV1,
+    PropertyNameParseBidiClassV1, PropertyNameParseCanonicalCombiningClassV1,
+    PropertyNameParseEastAsianWidthV1, PropertyNameParseGeneralCategoryMaskV1,
+    PropertyNameParseGeneralCategoryV1, PropertyNameParseGraphemeClusterBreakV1,
+    PropertyNameParseHangulSyllableTypeV1, PropertyNameParseIndicSyllabicCategoryV1,
+    PropertyNameParseJoiningTypeV1, PropertyNameParseLineBreakV1, PropertyNameParseNumericTypeV1,
+    PropertyNameParseScriptV1, PropertyNameParseSentenceBreakV1,
     PropertyNameParseVerticalOrientationV1, PropertyNameParseWordBreakV1,
     PropertyNameShortBidiClassV1, PropertyNameShortEastAsianWidthV1,
     PropertyNameShortGeneralCategoryV1, PropertyNameShortGraphemeClusterBreakV1,
     PropertyNameShortHangulSyllableTypeV1, PropertyNameShortIndicSyllabicCategoryV1,
-    PropertyNameShortJoiningTypeV1, PropertyNameShortLineBreakV1, PropertyNameShortScriptV1,
-    PropertyNameShortSentenceBreakV1, PropertyNameShortVerticalOrientationV1,
-    PropertyNameShortWordBreakV1,
+    PropertyNameShortJoiningTypeV1, PropertyNameShortLineBreakV1, PropertyNameShortNumericTypeV1,
+    PropertyNameShortScriptV1, PropertyNameShortSentenceBreakV1,
+    PropertyNameShortVerticalOrientationV1, PropertyNameShortWordBreakV1,
 };
 
 pub use crate::props::gc::GeneralCategoryULE;
@@ -145,6 +146,7 @@ const _: () = {
     impl_property_binary_xid_continue_v1!(Baked);
     impl_property_binary_xid_start_v1!(Baked);
     impl_property_enum_bidi_class_v1!(Baked);
+    impl_property_enum_numeric_type_v1!(Baked);
     impl_property_enum_bidi_mirroring_glyph_v1!(Baked);
     impl_property_enum_canonical_combining_class_v1!(Baked);
     impl_property_enum_east_asian_width_v1!(Baked);
@@ -160,6 +162,7 @@ const _: () = {
     impl_property_enum_vertical_orientation_v1!(Baked);
     impl_property_enum_word_break_v1!(Baked);
     impl_property_name_long_bidi_class_v1!(Baked);
+    impl_property_name_long_numeric_type_v1!(Baked);
     #[cfg(feature = "alloc")]
     impl_property_name_long_canonical_combining_class_v1!(Baked);
     impl_property_name_long_east_asian_width_v1!(Baked);
@@ -174,6 +177,7 @@ const _: () = {
     impl_property_name_long_vertical_orientation_v1!(Baked);
     impl_property_name_long_word_break_v1!(Baked);
     impl_property_name_parse_bidi_class_v1!(Baked);
+    impl_property_name_parse_numeric_type_v1!(Baked);
     impl_property_name_parse_canonical_combining_class_v1!(Baked);
     impl_property_name_parse_east_asian_width_v1!(Baked);
     impl_property_name_parse_general_category_mask_v1!(Baked);
@@ -188,6 +192,7 @@ const _: () = {
     impl_property_name_parse_vertical_orientation_v1!(Baked);
     impl_property_name_parse_word_break_v1!(Baked);
     impl_property_name_short_bidi_class_v1!(Baked);
+    impl_property_name_short_numeric_type_v1!(Baked);
     #[cfg(feature = "alloc")]
     impl_property_name_short_canonical_combining_class_v1!(Baked);
     impl_property_name_short_east_asian_width_v1!(Baked);
@@ -624,6 +629,12 @@ icu_provider::data_marker!(
     PropertyCodePointMap<'static, crate::props::BidiClass>,
     is_singleton = true,
 );
+icu_provider::data_marker!(
+    /// Data marker for the #NumericValue' Unicode property
+    PropertyEnumNumericTypeV1,
+    PropertyCodePointMap<'static, crate::props::NumericType>,
+    is_singleton = true,
+);
 icu_provider::data_marker!(
     /// Data marker for the 'CanonicalCombiningClass' Unicode property
     PropertyEnumCanonicalCombiningClassV1,
@@ -724,6 +735,7 @@ icu_provider::data_marker!(
 /// All data keys in this module.
 pub const MARKERS: &[DataMarkerInfo] = &[
     PropertyNameLongBidiClassV1::INFO,
+    PropertyNameLongNumericTypeV1::INFO,
     #[cfg(feature = "alloc")]
     PropertyNameLongCanonicalCombiningClassV1::INFO,
     PropertyNameLongEastAsianWidthV1::INFO,
@@ -738,6 +750,7 @@ pub const MARKERS: &[DataMarkerInfo] = &[
     PropertyNameLongVerticalOrientationV1::INFO,
     PropertyNameLongWordBreakV1::INFO,
     PropertyNameParseBidiClassV1::INFO,
+    PropertyNameParseNumericTypeV1::INFO,
     PropertyNameParseCanonicalCombiningClassV1::INFO,
     PropertyNameParseEastAsianWidthV1::INFO,
     PropertyNameParseGeneralCategoryMaskV1::INFO,
@@ -752,6 +765,7 @@ pub const MARKERS: &[DataMarkerInfo] = &[
     PropertyNameParseVerticalOrientationV1::INFO,
     PropertyNameParseWordBreakV1::INFO,
     PropertyNameShortBidiClassV1::INFO,
+    PropertyNameShortNumericTypeV1::INFO,
     #[cfg(feature = "alloc")]
     PropertyNameShortCanonicalCombiningClassV1::INFO,
     PropertyNameShortEastAsianWidthV1::INFO,
@@ -835,6 +849,7 @@ pub const MARKERS: &[DataMarkerInfo] = &[
     PropertyBinaryXidContinueV1::INFO,
     PropertyBinaryXidStartV1::INFO,
     PropertyEnumBidiClassV1::INFO,
+    PropertyEnumNumericTypeV1::INFO,
     PropertyEnumCanonicalCombiningClassV1::INFO,
     PropertyEnumEastAsianWidthV1::INFO,
     PropertyEnumGeneralCategoryV1::INFO,

components/properties/src/provider/names.rs

@@ -25,6 +25,12 @@ icu_provider::data_marker!(
     PropertyValueNameToEnumMap<'static>,
     is_singleton = true
 );
+icu_provider::data_marker!(
+    /// `PropertyNameParseNumericTypeV1`
+    PropertyNameParseNumericTypeV1,
+    PropertyValueNameToEnumMap<'static>,
+    is_singleton = true
+);
 icu_provider::data_marker!(
     /// `PropertyNameParseCanonicalCombiningClassV1`
     PropertyNameParseCanonicalCombiningClassV1,
@@ -115,6 +121,18 @@ icu_provider::data_marker!(
     PropertyEnumToValueNameLinearMap<'static>,
     is_singleton = true
 );
+icu_provider::data_marker!(
+    /// `PropertyNameLongNumericTypeV1`
+    PropertyNameLongNumericTypeV1,
+    PropertyEnumToValueNameLinearMap<'static>,
+    is_singleton = true,
+);
+icu_provider::data_marker!(
+    /// `PropertyNameShortNumericTypeV1`
+    PropertyNameShortNumericTypeV1,
+    PropertyEnumToValueNameLinearMap<'static>,
+    is_singleton = true,
+);
 icu_provider::data_marker!(
     /// `PropertyNameLongEastAsianWidthV1`
     PropertyNameLongEastAsianWidthV1,

components/properties/src/trievalue.rs

@@ -6,7 +6,7 @@ use crate::bidi::BidiMirroringGlyph;
 use crate::props::{
     BidiClass, CanonicalCombiningClass, EastAsianWidth, GeneralCategory, GeneralCategoryGroup,
     GraphemeClusterBreak, HangulSyllableType, IndicConjunctBreak, IndicSyllabicCategory,
-    JoiningType, LineBreak, Script, SentenceBreak, VerticalOrientation, WordBreak,
+    JoiningType, LineBreak, NumericType, Script, SentenceBreak, VerticalOrientation, WordBreak,
 };
 use crate::script::ScriptWithExt;
 use core::convert::TryInto;
@@ -29,6 +29,18 @@ impl TrieValue for CanonicalCombiningClass {
     }
 }
 
+impl TrieValue for NumericType {
+    type TryFromU32Error = TryFromIntError;
+
+    fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
+        u8::try_from(i).map(Self)
+    }
+
+    fn to_u32(self) -> u32 {
+        u32::from(self.0)
+    }
+}
+
 impl TrieValue for BidiClass {
     type TryFromU32Error = TryFromIntError;