Allow CodePointTrie to determine error_value at runtime from data

components/properties/src/props.rs

@@ -48,6 +48,8 @@ pub enum EnumeratedProperty {
 /// For more information, see [Unicode Standard Annex #9](https://unicode.org/reports/tr41/tr41-28.html#UAX9).
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(BidiClassULE)]
@@ -113,6 +115,8 @@ impl BidiClass {
 /// It does not support grouped categories (eg `Letter`). For grouped categories, use [`GeneralCategoryGroup`].
 #[derive(Copy, Clone, PartialEq, Eq, Debug, Ord, PartialOrd)]
 #[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_enums)] // this type is stable
 #[zerovec::make_ule(GeneralCategoryULE)]
 #[repr(u8)]
@@ -388,6 +392,8 @@ impl From<GeneralCategory> for GeneralCategoryGroup {
 /// See `UScriptCode` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(ScriptULE)]
@@ -568,6 +574,8 @@ impl Script {
 /// The numeric value is compatible with `UEastAsianWidth` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(EastAsianWidthULE)]
@@ -592,6 +600,8 @@ impl EastAsianWidth {
 /// The numeric value is compatible with `ULineBreak` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(LineBreakULE)]
@@ -654,6 +664,8 @@ impl LineBreak {
 /// The numeric value is compatible with `UGraphemeClusterBreak` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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)] // this type is stable
 #[repr(transparent)]
 #[zerovec::make_ule(GraphemeClusterBreakULE)]
@@ -695,6 +707,8 @@ impl GraphemeClusterBreak {
 /// The numeric value is compatible with `UWordBreakValues` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(WordBreakULE)]
@@ -740,6 +754,8 @@ impl WordBreak {
 /// The numeric value is compatible with `USentenceBreak` in ICU4C.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(SentenceBreakULE)]
@@ -770,6 +786,8 @@ impl SentenceBreak {
 /// <https://www.unicode.org/reports/tr15/>.
 #[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 #[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(CanonicalCombiningClassULE)]

components/properties/src/script.rs

@@ -12,7 +12,7 @@ use crate::provider::*;
 use crate::props::ScriptULE;
 use core::iter::FromIterator;
 use core::ops::RangeInclusive;
-use icu_codepointtrie::{CodePointTrie, TrieValue};
+use icu_codepointtrie::CodePointTrie;
 use icu_provider::prelude::*;
 use icu_uniset::CodePointInversionList;
 #[cfg(feature = "serde")]
@@ -36,6 +36,8 @@ const SCRIPT_X_SCRIPT_VAL: u16 = (1 << SCRIPT_VAL_LENGTH) - 1;
 /// into the `extensions` structure.
 #[derive(Copy, Clone, Debug, Eq, PartialEq)]
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[cfg_attr(feature = "datagen", derive(databake::Bake))]
+#[cfg_attr(feature = "datagen", databake(path = icu_properties::script))]
 #[repr(transparent)]
 #[doc(hidden)] // `ScriptWithExt` not intended as public-facing but for `ScriptWithExtensions` constructor
 #[allow(clippy::exhaustive_structs)] // this type is stable
@@ -317,7 +319,7 @@ impl<'data> ScriptWithExtensions<'data> {
             let scx_val = self.extensions.get(ext_idx as usize);
             let scx_first_sc = scx_val.and_then(|scx| scx.get(0));
 
-            let default_sc_val = <Script as TrieValue>::DATA_GET_ERROR_VALUE;
+            let default_sc_val = Script::Unknown;
 
             scx_first_sc.unwrap_or(default_sc_val)
         } else if sc_with_ext.is_common() {

components/properties/src/trievalue.rs

@@ -14,7 +14,6 @@ use icu_codepointtrie::TrieValue;
 use core::convert::TryFrom;
 
 impl TrieValue for CanonicalCombiningClass {
-    const DATA_GET_ERROR_VALUE: CanonicalCombiningClass = CanonicalCombiningClass::NotReordered;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -23,7 +22,6 @@ impl TrieValue for CanonicalCombiningClass {
 }
 
 impl TrieValue for BidiClass {
-    const DATA_GET_ERROR_VALUE: BidiClass = BidiClass::OtherNeutral;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -32,7 +30,6 @@ impl TrieValue for BidiClass {
 }
 
 impl TrieValue for GeneralCategory {
-    const DATA_GET_ERROR_VALUE: GeneralCategory = GeneralCategory::Unassigned;
     type TryFromU32Error = &'static str;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -43,7 +40,6 @@ impl TrieValue for GeneralCategory {
 }
 
 impl TrieValue for Script {
-    const DATA_GET_ERROR_VALUE: Script = Script::Unknown;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -52,7 +48,6 @@ impl TrieValue for Script {
 }
 
 impl TrieValue for ScriptWithExt {
-    const DATA_GET_ERROR_VALUE: ScriptWithExt = ScriptWithExt::Unknown;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -61,7 +56,6 @@ impl TrieValue for ScriptWithExt {
 }
 
 impl TrieValue for EastAsianWidth {
-    const DATA_GET_ERROR_VALUE: EastAsianWidth = EastAsianWidth::Neutral;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -70,7 +64,6 @@ impl TrieValue for EastAsianWidth {
 }
 
 impl TrieValue for LineBreak {
-    const DATA_GET_ERROR_VALUE: LineBreak = LineBreak::Unknown;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -79,7 +72,6 @@ impl TrieValue for LineBreak {
 }
 
 impl TrieValue for GraphemeClusterBreak {
-    const DATA_GET_ERROR_VALUE: GraphemeClusterBreak = GraphemeClusterBreak::Other;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -88,7 +80,6 @@ impl TrieValue for GraphemeClusterBreak {
 }
 
 impl TrieValue for WordBreak {
-    const DATA_GET_ERROR_VALUE: WordBreak = WordBreak::Other;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {
@@ -97,7 +88,6 @@ impl TrieValue for WordBreak {
 }
 
 impl TrieValue for SentenceBreak {
-    const DATA_GET_ERROR_VALUE: SentenceBreak = SentenceBreak::Other;
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {

experimental/casemapping/src/internals.rs

@@ -95,11 +95,12 @@ impl DotType {
     }
 }
 
-// The datatype stored in the codepoint trie for casemapping.
+/// The datatype stored in the codepoint trie for casemapping.
 #[derive(Copy, Clone, Debug, Eq, PartialEq)]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]
-#[cfg_attr(feature = "datagen", derive(serde::Serialize))]
-pub struct CaseMappingData(u16);
+#[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake))]
+#[cfg_attr(feature = "datagen", databake(path = icu_casemapping::provider))]
+pub struct CaseMappingData(pub u16);
 
 impl CaseMappingData {
     // Sequences of case-ignorable characters are skipped when determining
@@ -237,7 +238,6 @@ impl AsULE for CaseMappingData {
 }
 
 impl TrieValue for CaseMappingData {
-    const DATA_GET_ERROR_VALUE: CaseMappingData = CaseMappingData(0);
     type TryFromU32Error = TryFromIntError;
 
     fn try_from_u32(i: u32) -> Result<Self, Self::TryFromU32Error> {

experimental/casemapping/src/provider.rs

@@ -9,6 +9,7 @@
 use icu_provider::{yoke, zerofrom};
 
 pub use crate::exceptions::CaseMappingExceptions;
+pub use crate::internals::CaseMappingData;
 pub use crate::internals::CaseMappingInternals;
 pub use crate::internals::CaseMappingUnfoldData;
 

provider/testdata/data/baked/collator/data_v1.rs

@@ -115,6 +115,7 @@ static BN: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 0u8, 192u8, 0u8, 0u8, 0u8, 192u8, 0u8, 0u8, 0u8,
             ])
         },
+        192u32,
     ),
     ces: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
     ce32s: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
@@ -207,6 +208,7 @@ static ES: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 0u8, 0u8, 192u8, 0u8, 0u8, 0u8, 192u8, 0u8, 0u8, 0u8,
             ])
         },
+        192u32,
     ),
     ces: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
     ce32s: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
@@ -5754,6 +5756,7 @@ static JA: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 0u8, 192u8, 0u8, 0u8, 0u8,
             ])
         },
+        192u32,
     ),
     ces: unsafe {
         ::zerovec::ZeroVec::from_bytes_unchecked(&[
@@ -6228,6 +6231,7 @@ static TH: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 0u8, 0u8, 0u8, 192u8, 0u8, 0u8, 0u8, 192u8, 0u8, 0u8, 0u8, 192u8, 0u8, 0u8, 0u8,
             ])
         },
+        192u32,
     ),
     ces: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
     ce32s: unsafe {
@@ -6346,6 +6350,7 @@ static TR: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 0u8, 0u8,
             ])
         },
+        192u32,
     ),
     ces: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
     ce32s: unsafe { ::zerovec::ZeroVec::from_bytes_unchecked(&[]) },
@@ -13300,6 +13305,7 @@ static UND: &DataStruct = &::icu_collator::provider::CollationDataV1 {
                 255u8,
             ])
         },
+        4294967295u32,
     ),
     ces: unsafe {
         ::zerovec::ZeroVec::from_bytes_unchecked(&[

provider/testdata/data/baked/normalizer/decomp_v1.rs

@@ -285,6 +285,7 @@ static UND: &DataStruct = &::icu_normalizer::provider::NonRecursiveDecomposition
                 0u8,
             ])
         },
+        0u32,
     ),
     scalars24: unsafe {
         ::zerovec::ZeroVec::from_bytes_unchecked(&[

provider/testdata/data/baked/normalizer/nfc_v1.rs

@@ -202,5 +202,6 @@ static UND: &DataStruct = &::icu_normalizer::provider::CompositionPassthroughV1
                 0u8,
             ])
         },
+        0u8,
     ),
 };

provider/testdata/data/baked/normalizer/nfd_v1.rs

@@ -1562,5 +1562,6 @@ static UND: &DataStruct = &::icu_normalizer::provider::DecompositionDataV1 {
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        0u32,
     ),
 };

provider/testdata/data/baked/normalizer/nfkc_v1.rs

@@ -253,5 +253,6 @@ static UND: &DataStruct = &::icu_normalizer::provider::CompositionPassthroughV1
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        0u8,
     ),
 };

provider/testdata/data/baked/normalizer/nfkd_v1.rs

@@ -1333,6 +1333,7 @@ static UND: &DataStruct = &::icu_normalizer::provider::DecompositionSupplementV1
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        0u32,
     ),
     flags: 1u8,
 };

provider/testdata/data/baked/normalizer/uts46_v1.rs

@@ -266,5 +266,6 @@ static UND: &DataStruct = &::icu_normalizer::provider::CompositionPassthroughV1
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        0u8,
     ),
 };

provider/testdata/data/baked/normalizer/uts46d_v1.rs

@@ -1841,6 +1841,7 @@ static UND: &DataStruct = &::icu_normalizer::provider::DecompositionSupplementV1
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        0u32,
     ),
     flags: 3u8,
 };

provider/testdata/data/baked/props/bc_v1.rs

@@ -668,5 +668,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 18u8, 18u8, 18u8, 18u8, 18u8, 0u8,
             ])
         },
+        ::icu_properties::BidiClass(0u8),
     ),
 );

provider/testdata/data/baked/props/casemap_v1.rs

@@ -1276,6 +1276,7 @@ static UND: &DataStruct = &::icu_casemapping::provider::CaseMappingV1 {
                     0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
                 ])
             },
+            ::icu_casemapping::provider::CaseMappingData(0u16),
         ),
         exceptions: ::icu_casemapping::provider::CaseMappingExceptions {
             slots: unsafe {

provider/testdata/data/baked/props/ccc_v1.rs

@@ -363,5 +363,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::CanonicalCombiningClass(0u8),
     ),
 );

provider/testdata/data/baked/props/ea_v1.rs

@@ -350,5 +350,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 5u8, 5u8, 5u8, 5u8, 0u8, 0u8, 0u8, 5u8, 5u8, 5u8, 5u8, 5u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::EastAsianWidth(0u8),
     ),
 );

provider/testdata/data/baked/props/gc_v1.rs

@@ -1166,5 +1166,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 17u8, 17u8, 17u8, 17u8, 17u8, 17u8, 17u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::GeneralCategory::Unassigned,
     ),
 );

provider/testdata/data/baked/props/gcb_v1.rs

@@ -519,5 +519,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 12u8, 12u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::GraphemeClusterBreak(0u8),
     ),
 );

provider/testdata/data/baked/props/lb_v1.rs

@@ -1028,5 +1028,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 14u8, 14u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::LineBreak(0u8),
     ),
 );

provider/testdata/data/baked/props/sb_v1.rs

@@ -927,5 +927,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::SentenceBreak(0u8),
     ),
 );

provider/testdata/data/baked/props/sc_v1.rs

@@ -1750,5 +1750,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 0u8, 103u8, 0u8, 103u8, 0u8, 103u8, 0u8,
             ])
         },
+        ::icu_properties::Script(103u16),
     ),
 );

provider/testdata/data/baked/props/scx_v1.rs

@@ -1895,6 +1895,7 @@ static UND: &DataStruct = &::icu_properties::provider::ScriptWithExtensionsPrope
                     103u8, 0u8, 103u8, 0u8, 103u8, 0u8, 103u8, 0u8, 103u8, 0u8, 103u8, 0u8,
                 ])
             },
+            ::icu_properties::script::ScriptWithExt(103u16),
         ),
         extensions: unsafe {
             ::zerovec::VarZeroVec::from_bytes_unchecked(&[

provider/testdata/data/baked/props/wb_v1.rs

@@ -727,5 +727,6 @@ static UND: &DataStruct = &::icu_properties::provider::PropertyCodePointMapV1::C
                 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8, 0u8,
             ])
         },
+        ::icu_properties::WordBreak(0u8),
     ),
 );