Add Hangul_Syllable_Type (#4885)

Fixes #4884

components/properties/src/lib.rs

@@ -92,8 +92,8 @@ mod trievalue;
 
 pub use props::{
     BidiClass, CanonicalCombiningClass, EastAsianWidth, GeneralCategory, GeneralCategoryGroup,
-    GraphemeClusterBreak, IndicSyllabicCategory, JoiningType, LineBreak, Script, SentenceBreak,
-    WordBreak,
+    GraphemeClusterBreak, HangulSyllableType, IndicSyllabicCategory, JoiningType, LineBreak,
+    Script, SentenceBreak, WordBreak,
 };
 
 /// Module for working with the names of property values

components/properties/src/maps.rs

@@ -416,6 +416,32 @@ make_map_property! {
     pub fn load_script();
 }
 
+make_map_property! {
+    property: "Hangul_Syllable_Type";
+    marker: HangulSyllableTypeProperty;
+    value: crate::HangulSyllableType;
+    keyed_data_marker: HangulSyllableTypeV1Marker;
+    func:
+    /// Returns a [`CodePointMapDataBorrowed`] for the Hangul_Syllable_Type
+    /// Unicode enumerated property. See [`HangulSyllableType`].
+    ///
+    /// ✨ *Enabled with the `compiled_data` Cargo feature.*
+    ///
+    /// [📚 Help choosing a constructor](icu_provider::constructors)
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// use icu::properties::{maps, HangulSyllableType};
+    ///
+    /// assert_eq!(maps::hangul_syllable_type().get('ᄀ'), HangulSyllableType::LeadingJamo);  // U+1100
+    /// assert_eq!(maps::hangul_syllable_type().get('가'), HangulSyllableType::LeadingVowelSyllable);  // U+AC00
+    /// ```
+
+    pub const hangul_syllable_type => SINGLETON_PROPS_HST_V1;
+    pub fn load_hangul_syllable_type();
+}
+
 make_map_property! {
     property: "East_Asian_Width";
     marker: EastAsianWidthProperty;

components/properties/src/props.rs

@@ -1578,6 +1578,106 @@ impl_value_getter! {
     }
 }
 
+/// Enumerated property Hangul_Syllable_Type
+///
+/// The Unicode standard provides both precomposed Hangul syllables and conjoining Jamo to compose
+/// arbitrary Hangul syllables. This property provies that ontology of Hangul code points.
+///
+/// For more information, see the [Unicode Korean FAQ](https://www.unicode.org/faq/korean.html).
+#[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd, Hash)]
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[cfg_attr(feature = "datagen", derive(databake::Bake))]
+#[cfg_attr(feature = "datagen", databake(path = icu_properties))]
+#[allow(clippy::exhaustive_structs)] // newtype
+#[repr(transparent)]
+#[zerovec::make_ule(HangulSyllableTypeULE)]
+pub struct HangulSyllableType(pub u8);
+
+create_const_array! {
+#[allow(non_upper_case_globals)]
+impl HangulSyllableType {
+    /// (`NA`) not applicable (e.g. not a Hangul code point).
+    pub const NotApplicable: HangulSyllableType = HangulSyllableType(0);
+    /// (`L`) a conjoining leading consonant Jamo.
+    pub const LeadingJamo: HangulSyllableType = HangulSyllableType(1);
+    /// (`V`) a conjoining vowel Jamo.
+    pub const VowelJamo: HangulSyllableType = HangulSyllableType(2);
+    /// (`T`) a conjoining trailing consonent Jamo.
+    pub const TrailingJamo: HangulSyllableType = HangulSyllableType(3);
+    /// (`LV`) a precomposed syllable with a leading consonant and a vowel.
+    pub const LeadingVowelSyllable: HangulSyllableType = HangulSyllableType(4);
+    /// (`LVT`) a precomposed syllable with a leading consonant, a vowel, and a trailing consonant.
+    pub const LeadingVowelTrailingSyllable: HangulSyllableType = HangulSyllableType(5);
+}
+}
+
+impl_value_getter! {
+    markers: HangulSyllableTypeNameToValueV1Marker / SINGLETON_PROPNAMES_FROM_HST_V1, HangulSyllableTypeValueToShortNameV1Marker / SINGLETON_PROPNAMES_TO_SHORT_LINEAR_HST_V1, HangulSyllableTypeValueToLongNameV1Marker / SINGLETON_PROPNAMES_TO_LONG_LINEAR_HST_V1;
+    impl HangulSyllableType {
+        /// Return a [`PropertyValueNameToEnumMapper`], capable of looking up values
+        /// from strings for the `Bidi_Class` enumerated property
+        ///
+        /// ✨ *Enabled with the `compiled_data` Cargo feature.*
+        ///
+        /// [📚 Help choosing a constructor](icu_provider::constructors)
+        ///
+        /// # Example
+        ///
+        /// ```
+        /// use icu::properties::HangulSyllableType;
+        ///
+        /// let lookup = HangulSyllableType::name_to_enum_mapper();
+        /// // short name for value
+        /// assert_eq!(lookup.get_strict("L"), Some(HangulSyllableType::LeadingJamo));
+        /// assert_eq!(lookup.get_strict("LV"), Some(HangulSyllableType::LeadingVowelSyllable));
+        /// // long name for value
+        /// assert_eq!(lookup.get_strict("Leading_Jamo"), Some(HangulSyllableType::LeadingJamo));
+        /// assert_eq!(lookup.get_strict("LV_Syllable"), Some(HangulSyllableType::LeadingVowelSyllable));
+        /// // name has incorrect casing
+        /// assert_eq!(lookup.get_strict("lv"), None);
+        /// // loose matching of name
+        /// assert_eq!(lookup.get_loose("lv"), Some(HangulSyllableType::LeadingVowelSyllable));
+        /// // fake property
+        /// assert_eq!(lookup.get_strict("LT_Syllable"), None);
+        /// ```
+        pub fn get_name_to_enum_mapper() / name_to_enum_mapper();
+        /// Return a [`PropertyEnumToValueNameLinearMapper`], capable of looking up short names
+        /// for values of the `Bidi_Class` enumerated property
+        ///
+        /// ✨ *Enabled with the `compiled_data` Cargo feature.*
+        ///
+        /// [📚 Help choosing a constructor](icu_provider::constructors)
+        ///
+        /// # Example
+        ///
+        /// ```
+        /// use icu::properties::HangulSyllableType;
+        ///
+        /// let lookup = HangulSyllableType::enum_to_short_name_mapper();
+        /// assert_eq!(lookup.get(HangulSyllableType::LeadingJamo), Some("L"));
+        /// assert_eq!(lookup.get(HangulSyllableType::LeadingVowelSyllable), Some("LV"));
+        /// ```
+        pub fn get_enum_to_short_name_mapper() / enum_to_short_name_mapper() -> PropertyEnumToValueNameLinearMapper / PropertyEnumToValueNameLinearMapperBorrowed;
+        /// Return a [`PropertyEnumToValueNameLinearMapper`], capable of looking up long names
+        /// for values of the `Bidi_Class` enumerated property
+        ///
+        /// ✨ *Enabled with the `compiled_data` Cargo feature.*
+        ///
+        /// [📚 Help choosing a constructor](icu_provider::constructors)
+        ///
+        /// # Example
+        ///
+        /// ```
+        /// use icu::properties::HangulSyllableType;
+        ///
+        /// let lookup = HangulSyllableType::enum_to_long_name_mapper();
+        /// assert_eq!(lookup.get(HangulSyllableType::LeadingJamo), Some("Leading_Jamo"));
+        /// assert_eq!(lookup.get(HangulSyllableType::LeadingVowelSyllable), Some("LV_Syllable"));
+        /// ```
+        pub fn get_enum_to_long_name_mapper() / enum_to_long_name_mapper() -> PropertyEnumToValueNameLinearMapper / PropertyEnumToValueNameLinearMapperBorrowed;
+    }
+}
+
 /// Enumerated property East_Asian_Width.
 ///
 /// See "Definition" in UAX #11 for the summary of each property value:
@@ -2616,4 +2716,12 @@ mod test_enumerated_property_completeness {
             BidiClass::ALL_CONSTS,
         );
     }
+
+    #[test]
+    fn test_hst() {
+        check_enum(
+            crate::provider::Baked::SINGLETON_PROPNAMES_FROM_HST_V1,
+            HangulSyllableType::ALL_CONSTS,
+        );
+    }
 }

components/properties/src/provider.rs

@@ -57,6 +57,7 @@ const _: () = {
     icu_properties_data::impl_propnames_from_ea_v1!(Baked);
     icu_properties_data::impl_propnames_from_gc_v1!(Baked);
     icu_properties_data::impl_propnames_from_gcm_v1!(Baked);
+    icu_properties_data::impl_propnames_from_hst_v1!(Baked);
     icu_properties_data::impl_propnames_from_insc_v1!(Baked);
     icu_properties_data::impl_propnames_from_jt_v1!(Baked);
     icu_properties_data::impl_propnames_from_lb_v1!(Baked);
@@ -67,6 +68,7 @@ const _: () = {
     icu_properties_data::impl_propnames_to_long_linear_ea_v1!(Baked);
     icu_properties_data::impl_propnames_to_long_linear_gc_v1!(Baked);
     icu_properties_data::impl_propnames_to_long_linear_gcb_v1!(Baked);
+    icu_properties_data::impl_propnames_to_long_linear_hst_v1!(Baked);
     icu_properties_data::impl_propnames_to_long_linear_insc_v1!(Baked);
     icu_properties_data::impl_propnames_to_long_linear_jt_v1!(Baked);
     icu_properties_data::impl_propnames_to_long_linear_lb_v1!(Baked);
@@ -78,6 +80,7 @@ const _: () = {
     icu_properties_data::impl_propnames_to_short_linear_ea_v1!(Baked);
     icu_properties_data::impl_propnames_to_short_linear_gc_v1!(Baked);
     icu_properties_data::impl_propnames_to_short_linear_gcb_v1!(Baked);
+    icu_properties_data::impl_propnames_to_short_linear_hst_v1!(Baked);
     icu_properties_data::impl_propnames_to_short_linear_insc_v1!(Baked);
     icu_properties_data::impl_propnames_to_short_linear_jt_v1!(Baked);
     icu_properties_data::impl_propnames_to_short_linear_lb_v1!(Baked);
@@ -128,6 +131,7 @@ const _: () = {
     icu_properties_data::impl_props_gr_link_v1!(Baked);
     icu_properties_data::impl_props_graph_v1!(Baked);
     icu_properties_data::impl_props_hex_v1!(Baked);
+    icu_properties_data::impl_props_hst_v1!(Baked);
     icu_properties_data::impl_props_hyphen_v1!(Baked);
     icu_properties_data::impl_props_idc_v1!(Baked);
     icu_properties_data::impl_props_ideo_v1!(Baked);
@@ -840,6 +844,16 @@ expand!(
             "sc",
             Script
         ),
+        (
+            HangulSyllableTypeV1Marker,
+            HangulSyllableTypeNameToValueV1Marker,
+            (
+                linear: HangulSyllableTypeValueToShortNameV1Marker,
+                HangulSyllableTypeValueToLongNameV1Marker
+            ),
+            "hst",
+            HangulSyllableType
+        ),
         (
             EastAsianWidthV1Marker,
             EastAsianWidthNameToValueV1Marker,

components/properties/src/trievalue.rs

@@ -8,8 +8,8 @@ use crate::provider::bidi_data::{
 use crate::script::ScriptWithExt;
 use crate::{
     BidiClass, CanonicalCombiningClass, EastAsianWidth, GeneralCategory, GeneralCategoryGroup,
-    GraphemeClusterBreak, IndicSyllabicCategory, JoiningType, LineBreak, Script, SentenceBreak,
-    WordBreak,
+    GraphemeClusterBreak, HangulSyllableType, IndicSyllabicCategory, JoiningType, LineBreak,
+    Script, SentenceBreak, WordBreak,
 };
 use core::convert::TryInto;
 use core::num::TryFromIntError;
@@ -69,6 +69,18 @@ impl TrieValue for Script {
     }
 }
 
+impl TrieValue for HangulSyllableType {
+    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 ScriptWithExt {
     type TryFromU32Error = TryFromIntError;