Returning correct locale errors in datagen (#3578)

provider/datagen/src/transform/cldr/calendar/japanese.rs

@@ -118,16 +118,18 @@ impl crate::DatagenProvider {
 }
 
 impl DataProvider<JapaneseErasV1Marker> for crate::DatagenProvider {
-    fn load(&self, _req: DataRequest) -> Result<DataResponse<JapaneseErasV1Marker>, DataError> {
+    fn load(&self, req: DataRequest) -> Result<DataResponse<JapaneseErasV1Marker>, DataError> {
+        self.check_req::<JapaneseErasV1Marker>(req)?;
         self.load_japanese_eras(false)
     }
 }
 
 impl DataProvider<JapaneseExtendedErasV1Marker> for crate::DatagenProvider {
     fn load(
         &self,
-        _req: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<JapaneseExtendedErasV1Marker>, DataError> {
+        self.check_req::<JapaneseExtendedErasV1Marker>(req)?;
         let DataResponse { metadata, payload } = self.load_japanese_eras(true)?;
         Ok(DataResponse {
             metadata,

provider/datagen/src/transform/cldr/characters/mod.rs

@@ -22,6 +22,7 @@ macro_rules! exemplar_chars_impls {
     ($data_marker_name:ident, $cldr_serde_field_name:ident) => {
         impl DataProvider<$data_marker_name> for crate::DatagenProvider {
             fn load(&self, req: DataRequest) -> Result<DataResponse<$data_marker_name>, DataError> {
+                self.check_req::<$data_marker_name>(req)?;
                 let langid = req.locale.get_langid();
 
                 let data: &cldr_serde::exemplar_chars::Resource = self

provider/datagen/src/transform/cldr/datetime/mod.rs

@@ -37,6 +37,7 @@ macro_rules! impl_data_provider {
     ($marker:ident, $expr:expr, calendared = $calendared:expr) => {
         impl DataProvider<$marker> for crate::DatagenProvider {
             fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                self.check_req::<$marker>(req)?;
                 if $calendared == "locale" && req.locale.is_empty() {
                     return Err(DataErrorKind::NeedsLocale.into_error());
                 }

provider/datagen/src/transform/cldr/datetime/week_data.rs

@@ -38,6 +38,7 @@ impl IterableDataProvider<WeekDataV1Marker> for crate::DatagenProvider {
 
 impl DataProvider<WeekDataV1Marker> for crate::DatagenProvider {
     fn load(&self, req: DataRequest) -> Result<DataResponse<WeekDataV1Marker>, DataError> {
+        self.check_req::<WeekDataV1Marker>(req)?;
         let territory = req
             .locale
             .region()

provider/datagen/src/transform/cldr/decimal/compact.rs

@@ -14,6 +14,7 @@ impl DataProvider<ShortCompactDecimalFormatDataV1Marker> for crate::DatagenProvi
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<ShortCompactDecimalFormatDataV1Marker>, DataError> {
+        self.check_req::<ShortCompactDecimalFormatDataV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let resource: &cldr_serde::numbers::Resource = self
@@ -67,6 +68,7 @@ impl DataProvider<LongCompactDecimalFormatDataV1Marker> for crate::DatagenProvid
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<LongCompactDecimalFormatDataV1Marker>, DataError> {
+        self.check_req::<LongCompactDecimalFormatDataV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let resource: &cldr_serde::numbers::Resource = self

provider/datagen/src/transform/cldr/decimal/symbols.rs

@@ -13,6 +13,7 @@ use tinystr::TinyAsciiStr;
 
 impl DataProvider<DecimalSymbolsV1Marker> for crate::DatagenProvider {
     fn load(&self, req: DataRequest) -> Result<DataResponse<DecimalSymbolsV1Marker>, DataError> {
+        self.check_req::<DecimalSymbolsV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let resource: &cldr_serde::numbers::Resource = self

provider/datagen/src/transform/cldr/displaynames/language.rs

@@ -16,6 +16,7 @@ impl DataProvider<LanguageDisplayNamesV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<LanguageDisplayNamesV1Marker>, DataError> {
+        self.check_req::<LanguageDisplayNamesV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let data: &cldr_serde::displaynames::language::Resource = self
@@ -39,6 +40,7 @@ impl DataProvider<LocaleDisplayNamesV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<LocaleDisplayNamesV1Marker>, DataError> {
+        self.check_req::<LocaleDisplayNamesV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let data: &cldr_serde::displaynames::language::Resource = self

provider/datagen/src/transform/cldr/displaynames/region.rs

@@ -16,6 +16,7 @@ impl DataProvider<RegionDisplayNamesV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<RegionDisplayNamesV1Marker>, DataError> {
+        self.check_req::<RegionDisplayNamesV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let data: &cldr_serde::displaynames::region::Resource = self

provider/datagen/src/transform/cldr/displaynames/script.rs

@@ -16,6 +16,7 @@ impl DataProvider<ScriptDisplayNamesV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<ScriptDisplayNamesV1Marker>, DataError> {
+        self.check_req::<ScriptDisplayNamesV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let data: &cldr_serde::displaynames::script::Resource = self

provider/datagen/src/transform/cldr/displaynames/variant.rs

@@ -16,6 +16,7 @@ impl DataProvider<VariantDisplayNamesV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<VariantDisplayNamesV1Marker>, DataError> {
+        self.check_req::<VariantDisplayNamesV1Marker>(req)?;
         let langid = req.locale.get_langid();
 
         let data: &cldr_serde::displaynames::variant::Resource = self

provider/datagen/src/transform/cldr/fallback/mod.rs

@@ -21,8 +21,9 @@ use zerovec::{maps::ZeroMap2d, ule::UnvalidatedStr};
 impl DataProvider<LocaleFallbackLikelySubtagsV1Marker> for crate::DatagenProvider {
     fn load(
         &self,
-        _: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<LocaleFallbackLikelySubtagsV1Marker>, DataError> {
+        self.check_req::<LocaleFallbackLikelySubtagsV1Marker>(req)?;
         let resources = LikelySubtagsResources::try_from_source_data(&self.source)?;
 
         let metadata = DataResponseMetadata::default();
@@ -36,8 +37,9 @@ impl DataProvider<LocaleFallbackLikelySubtagsV1Marker> for crate::DatagenProvide
 impl DataProvider<LocaleFallbackParentsV1Marker> for crate::DatagenProvider {
     fn load(
         &self,
-        _: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<LocaleFallbackParentsV1Marker>, DataError> {
+        self.check_req::<LocaleFallbackParentsV1Marker>(req)?;
         let parents_data: &cldr_serde::parent_locales::Resource = self
             .source
             .cldr()?
@@ -55,8 +57,9 @@ impl DataProvider<LocaleFallbackParentsV1Marker> for crate::DatagenProvider {
 impl DataProvider<CollationFallbackSupplementV1Marker> for crate::DatagenProvider {
     fn load(
         &self,
-        _req: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<CollationFallbackSupplementV1Marker>, DataError> {
+        self.check_req::<CollationFallbackSupplementV1Marker>(req)?;
         // TODO(#1964): Load this data from its proper sources. For now, it is copied from:
         // https://github.com/unicode-org/icu/blob/main/tools/cldr/cldr-to-icu/build-icu-data.xml
         // as well as from CLDR XML.

provider/datagen/src/transform/cldr/list/mod.rs

@@ -122,6 +122,7 @@ macro_rules! implement {
     ($marker:ident) => {
         impl DataProvider<$marker> for crate::DatagenProvider {
             fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                self.check_req::<$marker>(req)?;
                 load(self, req)
             }
         }

provider/datagen/src/transform/cldr/locale_canonicalizer/aliases.rs

@@ -12,7 +12,8 @@ use tinystr::TinyAsciiStr;
 use zerovec::ZeroSlice;
 
 impl DataProvider<AliasesV1Marker> for crate::DatagenProvider {
-    fn load(&self, _: DataRequest) -> Result<DataResponse<AliasesV1Marker>, DataError> {
+    fn load(&self, req: DataRequest) -> Result<DataResponse<AliasesV1Marker>, DataError> {
+        self.check_req::<AliasesV1Marker>(req)?;
         let data: &cldr_serde::aliases::Resource = self
             .source
             .cldr()?

provider/datagen/src/transform/cldr/locale_canonicalizer/directionality.rs

@@ -9,7 +9,8 @@ use icu_provider::datagen::IterableDataProvider;
 use icu_provider::prelude::*;
 
 impl DataProvider<ScriptDirectionV1Marker> for crate::DatagenProvider {
-    fn load(&self, _: DataRequest) -> Result<DataResponse<ScriptDirectionV1Marker>, DataError> {
+    fn load(&self, req: DataRequest) -> Result<DataResponse<ScriptDirectionV1Marker>, DataError> {
+        self.check_req::<ScriptDirectionV1Marker>(req)?;
         let data: &cldr_serde::directionality::Resource = self
             .source
             .cldr()?

provider/datagen/src/transform/cldr/locale_canonicalizer/likely_subtags.rs

@@ -12,7 +12,8 @@ use icu_provider::prelude::*;
 use std::collections::{BTreeMap, HashSet};
 
 impl DataProvider<LikelySubtagsV1Marker> for crate::DatagenProvider {
-    fn load(&self, _: DataRequest) -> Result<DataResponse<LikelySubtagsV1Marker>, DataError> {
+    fn load(&self, req: DataRequest) -> Result<DataResponse<LikelySubtagsV1Marker>, DataError> {
+        self.check_req::<LikelySubtagsV1Marker>(req)?;
         let resources = LikelySubtagsResources::try_from_source_data(&self.source)?;
 
         Ok(DataResponse {
@@ -31,8 +32,9 @@ impl IterableDataProvider<LikelySubtagsV1Marker> for crate::DatagenProvider {
 impl DataProvider<LikelySubtagsExtendedV1Marker> for crate::DatagenProvider {
     fn load(
         &self,
-        _: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<LikelySubtagsExtendedV1Marker>, DataError> {
+        self.check_req::<LikelySubtagsExtendedV1Marker>(req)?;
         let resources = LikelySubtagsResources::try_from_source_data(&self.source)?;
 
         Ok(DataResponse {
@@ -55,6 +57,7 @@ impl DataProvider<LikelySubtagsForLanguageV1Marker> for crate::DatagenProvider {
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<LikelySubtagsForLanguageV1Marker>, DataError> {
+        self.check_req::<LikelySubtagsForLanguageV1Marker>(req)?;
         let response = DataProvider::<LikelySubtagsV1Marker>::load(self, req)?;
         Ok(DataResponse {
             metadata: response.metadata,
@@ -74,6 +77,7 @@ impl DataProvider<LikelySubtagsForScriptRegionV1Marker> for crate::DatagenProvid
         &self,
         req: DataRequest,
     ) -> Result<DataResponse<LikelySubtagsForScriptRegionV1Marker>, DataError> {
+        self.check_req::<LikelySubtagsForScriptRegionV1Marker>(req)?;
         let response = DataProvider::<LikelySubtagsV1Marker>::load(self, req)?;
         Ok(DataResponse {
             metadata: response.metadata,

provider/datagen/src/transform/cldr/plurals/mod.rs

@@ -37,6 +37,7 @@ macro_rules! implement {
     ($marker:ident) => {
         impl DataProvider<$marker> for crate::DatagenProvider {
             fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                self.check_req::<$marker>(req)?;
                 Ok(DataResponse {
                     metadata: Default::default(),
                     payload: Some(DataPayload::from_owned(PluralRulesV1::from(

provider/datagen/src/transform/cldr/relativetime/mod.rs

@@ -73,6 +73,7 @@ macro_rules! make_data_provider {
                     &self,
                     req: DataRequest,
                 ) -> Result<DataResponse<$marker>, DataError> {
+                    self.check_req::<$marker>(req)?;
                     let langid = req.locale.get_langid();
                     let resource: &cldr_serde::date_fields::Resource = self
                         .source

provider/datagen/src/transform/cldr/time_zones/mod.rs

@@ -29,6 +29,7 @@ macro_rules! impl_data_provider {
         $(
             impl DataProvider<$marker> for crate::DatagenProvider {
                 fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                    self.check_req::<$marker>(req)?;
                     let langid = req.locale.get_langid();
 
                     let resource: &cldr_serde::time_zones::time_zone_names::Resource = self
@@ -256,10 +257,7 @@ mod tests {
         );
 
         let metazone_period: DataPayload<MetazonePeriodV1Marker> = provider
-            .load(DataRequest {
-                locale: &langid!("en").into(),
-                metadata: Default::default(),
-            })
+            .load(Default::default())
             .unwrap()
             .take_payload()
             .unwrap();

provider/datagen/src/transform/icuexport/collator/mod.rs

@@ -131,6 +131,7 @@ macro_rules! collation_provider {
         $(
             impl DataProvider<$marker> for crate::DatagenProvider {
                 fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                    self.check_req::<$marker>(req)?;
                     let $toml_data: &collator_serde::$serde_struct = self
                         .source
                         .icuexport()?
@@ -160,6 +161,9 @@ macro_rules! collation_provider {
 
             impl IterableDataProvider<$marker> for crate::DatagenProvider {
                 fn supported_locales(&self) -> Result<Vec<DataLocale>, DataError> {
+                    if <$marker>::KEY.metadata().singleton {
+                        return Ok(vec![Default::default()])
+                    }
                     Ok(self.source.options.locales.filter_by_langid_equality(self
                         .source
                         .icuexport()?

provider/datagen/src/transform/icuexport/collator/test.rs

@@ -209,7 +209,9 @@ fn test_nb_nn_no() {
 
 #[test]
 fn test_zh() {
-    let provider = crate::DatagenProvider::for_test();
+    let mut provider = crate::DatagenProvider::for_test();
+    // `zh-u-co-gb2312` needs to be manually enabled
+    provider.source.options.collations.insert("gb2312".into());
     let provider = LocaleFallbackProvider::try_new_unstable(provider).unwrap();
     let provider = ResolvedLocaleAdapter::new(provider);
 

provider/datagen/src/transform/icuexport/normalizer/mod.rs

@@ -20,7 +20,8 @@ macro_rules! normalization_provider {
         use icu_normalizer::provider::$marker;
 
         impl DataProvider<$marker> for crate::DatagenProvider {
-            fn load(&self, _req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+            fn load(&self, req: DataRequest) -> Result<DataResponse<$marker>, DataError> {
+                self.check_req::<$marker>(req)?;
                 let $toml_data: &normalizer_serde::$serde_struct =
                     self.source.icuexport()?.read_and_parse_toml(&format!(
                         "norm/{}/{}.toml",

provider/datagen/src/transform/icuexport/ucase/mod.rs

@@ -14,7 +14,8 @@ use std::convert::TryFrom;
 mod ucase_serde;
 
 impl DataProvider<CaseMappingV1Marker> for crate::DatagenProvider {
-    fn load(&self, _req: DataRequest) -> Result<DataResponse<CaseMappingV1Marker>, DataError> {
+    fn load(&self, req: DataRequest) -> Result<DataResponse<CaseMappingV1Marker>, DataError> {
+        self.check_req::<CaseMappingV1Marker>(req)?;
         let toml = &self
             .source
             .icuexport()?

provider/datagen/src/transform/icuexport/uprops/bidi_data.rs

@@ -31,7 +31,7 @@ impl DataProvider<BidiAuxiliaryPropertiesV1Marker> for crate::DatagenProvider {
     #[cfg(any(feature = "use_wasm", feature = "use_icu4c"))]
     fn load(
         &self,
-        _: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<BidiAuxiliaryPropertiesV1Marker>, DataError> {
         use crate::transform::icuexport::uprops::{bin_cp_set, enum_codepointtrie};
         use icu_codepointtrie_builder::{CodePointTrieBuilder, CodePointTrieBuilderData};
@@ -42,6 +42,8 @@ impl DataProvider<BidiAuxiliaryPropertiesV1Marker> for crate::DatagenProvider {
             BidiAuxiliaryPropertiesV1, MirroredPairedBracketData,
         };
 
+        self.check_req::<BidiAuxiliaryPropertiesV1Marker>(req)?;
+
         // Bidi_M / Bidi_Mirrored
         let bidi_m_data = bin_cp_set::get_binary_prop_for_code_point_set(&self.source, "Bidi_M")?;
         let mut bidi_m_builder = CodePointInversionListBuilder::new();
@@ -100,8 +102,9 @@ impl DataProvider<BidiAuxiliaryPropertiesV1Marker> for crate::DatagenProvider {
     #[cfg(not(any(feature = "use_wasm", feature = "use_icu4c")))]
     fn load(
         &self,
-        _: DataRequest,
+        req: DataRequest,
     ) -> Result<DataResponse<BidiAuxiliaryPropertiesV1Marker>, DataError> {
+        self.check_req::<BidiAuxiliaryPropertiesV1Marker>(req)?;
         return Err(DataError::custom(
             "icu_datagen must be built with use_icu4c or use_wasm to build Bidi auxiliary properties data",
         ));

provider/datagen/src/transform/icuexport/uprops/bin_cp_set.rs

@@ -30,8 +30,9 @@ macro_rules! expand {
             impl DataProvider<$marker> for crate::DatagenProvider {
                 fn load(
                     &self,
-                    _: DataRequest,
+                    req: DataRequest,
                 ) -> Result<DataResponse<$marker>, DataError> {
+                    self.check_req::<$marker>(req)?;
                     let data = get_binary_prop_for_code_point_set(&self.source, $prop_name)?;
 
                     let mut builder = CodePointInversionListBuilder::new();

provider/datagen/src/transform/icuexport/uprops/bin_uniset.rs

@@ -31,8 +31,9 @@ macro_rules! expand {
             impl DataProvider<$marker> for crate::DatagenProvider {
                 fn load(
                     &self,
-                    _: DataRequest,
+                    req: DataRequest,
                 ) -> Result<DataResponse<$marker>, DataError> {
+                    self.check_req::<$marker>(req)?;
                     let data = get_binary_prop_for_unicodeset(&self.source, $prop_name)?;
 
                     let mut builder = CodePointInversionListBuilder::new();