CLDR 46 Alpha0 (#5479)

Author: robertbastian

components/datetime/src/format/neo.rs

@@ -2700,7 +2700,7 @@ mod tests {
                 pattern: "<EEEE>",
                 field_symbol: fields::Weekday::Format,
                 field_length: FieldLength::Wide,
-                expected: "<пʼятницю>",
+                expected: "<пʼятниця>",
             },
             TestCase {
                 pattern: "<EEEEE>",
@@ -2725,7 +2725,7 @@ mod tests {
                 pattern: "<eeee>",
                 field_symbol: fields::Weekday::Format,
                 field_length: FieldLength::Wide,
-                expected: "<пʼятницю>",
+                expected: "<пʼятниця>",
             },
             TestCase {
                 pattern: "<eeeee>",

components/datetime/tests/fixtures/tests/components_hour_cycle.json

@@ -118,7 +118,7 @@
         },
         "output": {
             "values": {
-                "ar": "٠:٢٥ ص"
+                "ar-EG": "٠:٢٥ ص"
             }
         }
     },
@@ -140,7 +140,7 @@
         },
         "output": {
             "values": {
-                "ar": "١٢:٢٥ ص"
+                "ar-EG": "١٢:٢٥ ص"
             }
         }
     },
@@ -162,7 +162,7 @@
         },
         "output": {
             "values": {
-                "ar": "٠٠:٢٥"
+                "ar-EG": "٠٠:٢٥"
             }
         }
     },
@@ -184,7 +184,7 @@
         },
         "output": {
             "values": {
-                "ar": "٢٤:٢٥"
+                "ar-EG": "٢٤:٢٥"
             }
         }
     },

components/experimental/tests/relativetime/tests.rs

@@ -29,7 +29,7 @@ macro_rules! generate_test {
             )+
 
             let relative_time_formatter = RelativeTimeFormatter::$constructor(
-                &locale!("ar").into(),
+                &locale!("ar-EG").into(),
                 $options
             )
             .expect("locale should be present");

components/timezone/src/ids.rs

@@ -10,7 +10,8 @@ use zerotrie::cursor::ZeroAsciiIgnoreCaseTrieCursor;
 
 use crate::{
     provider::names::{
-        Bcp47ToIanaMapV1, Bcp47ToIanaMapV1Marker, IanaToBcp47MapV2, IanaToBcp47MapV2Marker,
+        Bcp47ToIanaMapV1, Bcp47ToIanaMapV1Marker, IanaToBcp47MapV3, IanaToBcp47MapV3Marker,
+        NON_REGION_CITY_PREFIX,
     },
     TimeZoneBcp47Id,
 };
@@ -89,7 +90,7 @@ use crate::{
 /// ```
 #[derive(Debug, Clone)]
 pub struct TimeZoneIdMapper {
-    data: DataPayload<IanaToBcp47MapV2Marker>,
+    data: DataPayload<IanaToBcp47MapV3Marker>,
 }
 
 #[cfg(feature = "compiled_data")]
@@ -111,7 +112,7 @@ impl TimeZoneIdMapper {
     pub fn new() -> Self {
         Self {
             data: DataPayload::from_static_ref(
-                crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V2_MARKER,
+                crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V3_MARKER,
             ),
         }
     }
@@ -129,7 +130,7 @@ impl TimeZoneIdMapper {
     #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
     pub fn try_new_unstable<P>(provider: &P) -> Result<Self, DataError>
     where
-        P: DataProvider<IanaToBcp47MapV2Marker> + ?Sized,
+        P: DataProvider<IanaToBcp47MapV3Marker> + ?Sized,
     {
         let data = provider.load(Default::default())?.payload;
         Ok(Self { data })
@@ -156,7 +157,7 @@ impl AsRef<TimeZoneIdMapper> for TimeZoneIdMapper {
 /// [`TimeZoneIdMapper::as_borrowed()`]. More efficient to query.
 #[derive(Debug, Copy, Clone)]
 pub struct TimeZoneIdMapperBorrowed<'a> {
-    data: &'a IanaToBcp47MapV2<'a>,
+    data: &'a IanaToBcp47MapV3<'a>,
 }
 
 impl<'a> TimeZoneIdMapperBorrowed<'a> {
@@ -269,7 +270,7 @@ impl<'a> TimeZoneIdMapperBorrowed<'a> {
         // nearby the input IANA identifier. This should improve lookup time since
         // most renames share the same prefix like "Asia" or "Europe".
         let mut stack = Vec::with_capacity(iana_id.len());
-        let (trie_value, string) = self.iana_lookup_with_normalization(iana_id, |cursor| {
+        let (trie_value, mut string) = self.iana_lookup_with_normalization(iana_id, |cursor| {
             stack.push((cursor.clone(), 0, 1));
         })?;
         let Some(bcp47_id) = self.data.bcp47_ids.get(trie_value.index()) else {
@@ -281,6 +282,9 @@ impl<'a> TimeZoneIdMapperBorrowed<'a> {
         }
         // If we get here, we need to walk the trie to find the canonical IANA ID.
         let needle = trie_value.to_canonical();
+        if !string.contains('/') {
+            string.to_mut().insert(0, '_');
+        }
         let Some(string) = self.iana_search(needle, string.into_owned(), stack) else {
             debug_assert!(false, "every time zone should have a canonical IANA ID");
             return None;
@@ -332,7 +336,15 @@ impl<'a> TimeZoneIdMapperBorrowed<'a> {
     /// Queries the data for `iana_id` without recording the normalized string.
     /// This is a fast, no-alloc lookup.
     fn iana_lookup_quick(&self, iana_id: impl AsRef<[u8]>) -> Option<IanaTrieValue> {
-        self.data.map.get(iana_id).map(IanaTrieValue)
+        let mut cursor = self.data.map.cursor();
+        let iana_id = iana_id.as_ref();
+        if !iana_id.contains(&b'/') {
+            cursor.step(NON_REGION_CITY_PREFIX);
+        }
+        for &b in iana_id {
+            cursor.step(b);
+        }
+        cursor.take_value().map(IanaTrieValue)
     }
 
     /// Queries the data for `iana_id` while keeping track of the normalized string.
@@ -343,6 +355,10 @@ impl<'a> TimeZoneIdMapperBorrowed<'a> {
         mut cursor_fn: impl FnMut(&ZeroAsciiIgnoreCaseTrieCursor<'l>),
     ) -> Option<(IanaTrieValue, Cow<'s, str>)> {
         let mut cursor = self.data.map.cursor();
+        if !iana_id.contains('/') {
+            cursor_fn(&cursor);
+            cursor.step(NON_REGION_CITY_PREFIX);
+        }
         let mut string = Cow::Borrowed(iana_id);
         let mut i = 0;
         let trie_value = loop {
@@ -467,13 +483,13 @@ impl TimeZoneIdMapperWithFastCanonicalization<TimeZoneIdMapper> {
     #[cfg(feature = "compiled_data")]
     pub fn new() -> Self {
         const _: () = assert!(
-            crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V2_MARKER.bcp47_ids_checksum
+            crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V3_MARKER.bcp47_ids_checksum
                 == crate::provider::Baked::SINGLETON_BCP47_TO_IANA_MAP_V1_MARKER.bcp47_ids_checksum,
         );
         Self {
             inner: TimeZoneIdMapper {
                 data: DataPayload::from_static_ref(
-                    crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V2_MARKER,
+                    crate::provider::Baked::SINGLETON_IANA_TO_BCP47_MAP_V3_MARKER,
                 ),
             },
             data: DataPayload::from_static_ref(
@@ -495,7 +511,7 @@ impl TimeZoneIdMapperWithFastCanonicalization<TimeZoneIdMapper> {
     #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
     pub fn try_new_unstable<P>(provider: &P) -> Result<Self, DataError>
     where
-        P: DataProvider<IanaToBcp47MapV2Marker> + DataProvider<Bcp47ToIanaMapV1Marker> + ?Sized,
+        P: DataProvider<IanaToBcp47MapV3Marker> + DataProvider<Bcp47ToIanaMapV1Marker> + ?Sized,
     {
         let mapper = TimeZoneIdMapper::try_new_unstable(provider)?;
         Self::try_new_with_mapper_unstable(provider, mapper)
@@ -538,7 +554,7 @@ where
     #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
     pub fn try_new_with_mapper_unstable<P>(provider: &P, mapper: I) -> Result<Self, DataError>
     where
-        P: DataProvider<IanaToBcp47MapV2Marker> + DataProvider<Bcp47ToIanaMapV1Marker> + ?Sized,
+        P: DataProvider<IanaToBcp47MapV3Marker> + DataProvider<Bcp47ToIanaMapV1Marker> + ?Sized,
     {
         let data = provider.load(Default::default())?.payload;
         Self {
@@ -551,7 +567,7 @@ where
     fn validated(self) -> Result<Self, DataError> {
         if self.inner.as_ref().data.get().bcp47_ids_checksum != self.data.get().bcp47_ids_checksum {
             return Err(
-                DataErrorKind::InconsistentData(IanaToBcp47MapV2Marker::INFO)
+                DataErrorKind::InconsistentData(IanaToBcp47MapV3Marker::INFO)
                     .with_marker(Bcp47ToIanaMapV1Marker::INFO),
             );
         }

components/timezone/src/provider.rs

@@ -43,7 +43,7 @@ const _: () = {
     }
     make_provider!(Baked);
     impl_bcp47_to_iana_map_v1_marker!(Baked);
-    impl_iana_to_bcp47_map_v2_marker!(Baked);
+    impl_iana_to_bcp47_map_v3_marker!(Baked);
     impl_metazone_period_v1_marker!(Baked);
 };
 
@@ -52,7 +52,7 @@ const _: () = {
 pub const MARKERS: &[DataMarkerInfo] = &[
     MetazonePeriodV1Marker::INFO,
     names::Bcp47ToIanaMapV1Marker::INFO,
-    names::IanaToBcp47MapV2Marker::INFO,
+    names::IanaToBcp47MapV3Marker::INFO,
 ];
 
 /// TimeZone ID in BCP47 format

components/timezone/src/provider/names.rs

@@ -50,6 +50,14 @@ pub struct IanaToBcp47MapV1<'data> {
     pub bcp47_ids_checksum: u64,
 }
 
+/// [`IanaToBcp47MapV3`]'s trie cannot handle differently-cased prefixes, like `Mexico/BajaSur`` and `MET`.
+/// Therefore, any ID that is not of the shape `{region}/{city}` gets prefixed with this character
+/// inside the trie.
+///
+/// During lookup, if the input is not of the shape `{region}/{city}`, the trie cursor has to be advanced over
+/// this byte.
+pub const NON_REGION_CITY_PREFIX: u8 = b'_';
+
 /// A mapping from normal-case IANA time zone identifiers to BCP-47 time zone identifiers.
 ///
 /// Multiple IANA time zone IDs can map to the same BCP-47 time zone ID.
@@ -61,14 +69,14 @@ pub struct IanaToBcp47MapV1<'data> {
 /// </div>
 #[derive(Debug, Clone, PartialEq)]
 #[icu_provider::data_struct(marker(
-    IanaToBcp47MapV2Marker,
-    "time_zone/iana_to_bcp47@2",
+    IanaToBcp47MapV3Marker,
+    "time_zone/iana_to_bcp47@3",
     singleton
 ))]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
 #[cfg_attr(feature = "datagen", databake(path = icu_timezone::provider::names))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]
-pub struct IanaToBcp47MapV2<'data> {
+pub struct IanaToBcp47MapV3<'data> {
     /// A map from normal-case IANA time zone identifiers to indexes of BCP-47 time zone
     /// identifiers along with a canonical flag. The IANA identifiers are normal-case.
     ///