Moving root collation data into its own singleton marker (unicode-org…

…#5498)

components/collator/src/comparison.rs

@@ -14,12 +14,13 @@ use crate::elements::{
     NO_CE_SECONDARY, NO_CE_TERTIARY, OPTIMIZED_DIACRITICS_MAX_COUNT, QUATERNARY_MASK,
 };
 use crate::options::CollatorOptionsBitField;
-use crate::provider::CollationDataV1Marker;
 use crate::provider::CollationDiacriticsV1Marker;
 use crate::provider::CollationJamoV1Marker;
 use crate::provider::CollationMetadataV1Marker;
 use crate::provider::CollationReorderingV1Marker;
+use crate::provider::CollationRootV1Marker;
 use crate::provider::CollationSpecialPrimariesV1Marker;
+use crate::provider::CollationTailoringV1Marker;
 use crate::{AlternateHandling, CollatorOptions, MaxVariable, ResolvedCollatorOptions, Strength};
 use core::cmp::Ordering;
 use core::convert::TryFrom;
@@ -55,8 +56,8 @@ impl AnyQuaternaryAccumulator {
 #[derive(Debug)]
 pub struct Collator {
     special_primaries: Option<DataPayload<CollationSpecialPrimariesV1Marker>>,
-    root: DataPayload<CollationDataV1Marker>,
-    tailoring: Option<DataPayload<CollationDataV1Marker>>,
+    root: DataPayload<CollationRootV1Marker>,
+    tailoring: Option<DataPayload<CollationTailoringV1Marker>>,
     jamo: DataPayload<CollationJamoV1Marker>,
     diacritics: DataPayload<CollationDiacriticsV1Marker>,
     options: CollatorOptionsBitField,
@@ -72,6 +73,9 @@ impl Collator {
     pub fn try_new(locale: &DataLocale, options: CollatorOptions) -> Result<Self, DataError> {
         Self::try_new_unstable_internal(
             &crate::provider::Baked,
+            DataPayload::from_static_ref(
+                crate::provider::Baked::SINGLETON_COLLATION_ROOT_V1_MARKER,
+            ),
             DataPayload::from_static_ref(
                 icu_normalizer::provider::Baked::SINGLETON_CANONICAL_DECOMPOSITION_DATA_V1_MARKER,
             ),
@@ -110,7 +114,8 @@ impl Collator {
     ) -> Result<Self, DataError>
     where
         D: DataProvider<CollationSpecialPrimariesV1Marker>
-            + DataProvider<CollationDataV1Marker>
+            + DataProvider<CollationRootV1Marker>
+            + DataProvider<CollationTailoringV1Marker>
             + DataProvider<CollationDiacriticsV1Marker>
             + DataProvider<CollationJamoV1Marker>
             + DataProvider<CollationMetadataV1Marker>
@@ -124,14 +129,17 @@ impl Collator {
             provider.load(Default::default())?.payload,
             provider.load(Default::default())?.payload,
             provider.load(Default::default())?.payload,
+            provider.load(Default::default())?.payload,
             || provider.load(Default::default()).map(|r| r.payload),
             locale,
             options,
         )
     }
 
+    #[allow(clippy::too_many_arguments)]
     fn try_new_unstable_internal<D>(
         provider: &D,
+        root: DataPayload<CollationRootV1Marker>,
         decompositions: DataPayload<CanonicalDecompositionDataV1Marker>,
         tables: DataPayload<CanonicalDecompositionTablesV1Marker>,
         jamo: DataPayload<CollationJamoV1Marker>,
@@ -143,7 +151,8 @@ impl Collator {
         options: CollatorOptions,
     ) -> Result<Self, DataError>
     where
-        D: DataProvider<CollationDataV1Marker>
+        D: DataProvider<CollationRootV1Marker>
+            + DataProvider<CollationTailoringV1Marker>
             + DataProvider<CollationDiacriticsV1Marker>
             + DataProvider<CollationMetadataV1Marker>
             + DataProvider<CollationReorderingV1Marker>
@@ -177,7 +186,7 @@ impl Collator {
 
         let metadata = metadata_payload.get();
 
-        let tailoring: Option<DataPayload<crate::provider::CollationDataV1Marker>> =
+        let tailoring: Option<DataPayload<crate::provider::CollationTailoringV1Marker>> =
             if metadata.tailored() {
                 Some(
                     provider
@@ -209,8 +218,6 @@ impl Collator {
             }
         }
 
-        let root: DataPayload<CollationDataV1Marker> = provider.load(Default::default())?.payload;
-
         let tailored_diacritics = metadata.tailored_diacritics();
         let diacritics: DataPayload<CollationDiacriticsV1Marker> = provider
             .load(if tailored_diacritics {
@@ -339,7 +346,7 @@ impl Collator {
     }
 
     fn compare_impl<I: Iterator<Item = char>>(&self, left_chars: I, right_chars: I) -> Ordering {
-        let tailoring: &DataPayload<CollationDataV1Marker> =
+        let tailoring: &DataPayload<CollationTailoringV1Marker> =
             if let Some(tailoring) = &self.tailoring {
                 tailoring
             } else {
@@ -356,7 +363,7 @@ impl Collator {
                 // should we have a no-op tailoring that contains a
                 // specially-crafted CodePointTrie that always returns
                 // a FALLBACK_CE32 after a single branch?
-                &self.root
+                self.root.cast_ref()
             };
 
         // Sadly, it looks like variable CEs and backward second level

components/collator/src/provider.rs

@@ -59,7 +59,8 @@ const _: () = {
         pub use icu_collections as collections;
     }
     make_provider!(Baked);
-    impl_collation_data_v1_marker!(Baked);
+    impl_collation_root_v1_marker!(Baked);
+    impl_collation_tailoring_v1_marker!(Baked);
     impl_collation_diacritics_v1_marker!(Baked);
     impl_collation_jamo_v1_marker!(Baked);
     impl_collation_metadata_v1_marker!(Baked);
@@ -70,7 +71,8 @@ const _: () = {
 #[cfg(feature = "datagen")]
 /// The latest minimum set of markers required by this component.
 pub const MARKERS: &[DataMarkerInfo] = &[
-    CollationDataV1Marker::INFO,
+    CollationRootV1Marker::INFO,
+    CollationTailoringV1Marker::INFO,
     CollationDiacriticsV1Marker::INFO,
     CollationJamoV1Marker::INFO,
     CollationMetadataV1Marker::INFO,
@@ -114,13 +116,20 @@ fn data_ce_to_primary(data_ce: u64, c: char) -> u32 {
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(marker(
-    CollationDataV1Marker,
-    "collator/data@1",
-    // TODO(#3867): Use script fallback
-    fallback_by = "language",
-    attributes_domain = "collator",
-))]
+#[icu_provider::data_struct(
+    marker(
+        CollationRootV1Marker,
+        "collator/root@1",
+        singleton,
+    ),
+    marker(
+        CollationTailoringV1Marker,
+        "collator/tailoring@1",
+        // TODO(#3867): Use script fallback
+        fallback_by = "language",
+        attributes_domain = "collator",
+    )
+)]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
 #[cfg_attr(feature = "datagen", databake(path = icu_collator::provider))]

components/collator/tests/tests.rs

@@ -1286,7 +1286,7 @@ fn test_nb_nn_no() {
     strs.sort_by(|a, b| collator.compare(a, b));
     assert_eq!(strs, expected);
     assert_eq!(
-        DataProvider::<CollationDataV1Marker>::load(
+        DataProvider::<CollationTailoringV1Marker>::load(
             &icu_collator::provider::Baked,
             DataRequest {
                 id: DataIdentifierBorrowed::for_locale(&locale),
@@ -1306,7 +1306,7 @@ fn test_nb_nn_no() {
     strs.sort_by(|a, b| collator.compare(a, b));
     assert_eq!(strs, expected);
     assert_eq!(
-        DataProvider::<CollationDataV1Marker>::load(
+        DataProvider::<CollationTailoringV1Marker>::load(
             &icu_collator::provider::Baked,
             DataRequest {
                 id: DataIdentifierBorrowed::for_locale(&locale),
@@ -1326,7 +1326,7 @@ fn test_nb_nn_no() {
     strs.sort_by(|a, b| collator.compare(a, b));
     assert_eq!(strs, expected);
     assert_eq!(
-        DataProvider::<CollationDataV1Marker>::load(
+        DataProvider::<CollationTailoringV1Marker>::load(
             &icu_collator::provider::Baked,
             DataRequest {
                 id: DataIdentifierBorrowed::for_locale(&locale),

provider/core/src/response.rs

@@ -730,6 +730,22 @@ where
         })
     }
 
+    /// Convert between two [`DynamicDataMarker`] types that are compatible with each other
+    /// with compile-time type checking.
+    ///
+    /// This happens if they both have the same [`DynamicDataMarker::DataStruct`] type.
+    ///
+    /// Can be used to erase the marker of a data payload in cases where multiple markers correspond
+    /// to the same data struct.
+    #[inline]
+    pub fn cast_ref<M2>(&self) -> &DataPayload<M2>
+    where
+        M2: DynamicDataMarker<DataStruct = M::DataStruct>,
+    {
+        // SAFETY: As seen in the implementation of `cast`, the struct is the same, it's just the generic that changes.
+        unsafe { core::mem::transmute(self) }
+    }
+
     /// Convert a [`DataPayload`] to one of the same type with runtime type checking.
     ///
     /// Primarily useful to convert from a generic to a concrete marker type.