Follow-up #4985 (#4995)

unicode-org · Jun 3, 2024 · 24761aa · 24761aa
1 parent fffa624
commit 24761aa
Show file tree

Hide file tree

Showing 23 changed files with 102 additions and 67 deletions.
diff --git a/provider/datagen/src/provider.rs b/provider/datagen/src/provider.rs
@@ -8,6 +8,7 @@ use elsa::sync::FrozenMap;
 use icu_provider::datagen::IterableDataProvider;
 use icu_provider::prelude::*;
 use source::{AbstractFs, SerdeCache};
+use std::borrow::Cow;
 use std::collections::HashSet;
 use std::fmt::Debug;
 use std::path::PathBuf;
@@ -41,8 +42,17 @@ pub struct DatagenProvider {
     trie_type: TrieType,
     collation_han_database: CollationHanDatabase,
     #[allow(clippy::type_complexity)] // not as complex as it appears
-    supported_locales_cache:
-        Arc<FrozenMap<DataKey, Box<Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError>>>>,
+    supported_requests_cache: Arc<
+        FrozenMap<
+            DataKey,
+            Box<
+                Result<
+                    HashSet<(Cow<'static, DataLocale>, Cow<'static, DataKeyAttributes>)>,
+                    DataError,
+                >,
+            >,
+        >,
+    >,
 }
 
 macro_rules! cb {
@@ -109,7 +119,7 @@ impl DatagenProvider {
             segmenter_lstm_paths: None,
             trie_type: Default::default(),
             collation_han_database: Default::default(),
-            supported_locales_cache: Default::default(),
+            supported_requests_cache: Default::default(),
         }
     }
 
@@ -348,30 +358,51 @@ impl std::fmt::Display for TrieType {
 }
 
 trait IterableDataProviderCached<M: KeyedDataMarker>: DataProvider<M> {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError>;
 }
 
+impl DatagenProvider {
+    #[allow(clippy::type_complexity)] // not as complex as it appears
+    fn populate_supported_requests_cache<M: KeyedDataMarker>(
+        &self,
+    ) -> Result<&HashSet<(Cow<'static, DataLocale>, Cow<'static, DataKeyAttributes>)>, DataError>
+    where
+        DatagenProvider: IterableDataProviderCached<M>,
+    {
+        self.supported_requests_cache
+            .insert_with(M::KEY, || {
+                Box::new(self.supported_requests_cached().map(|m| {
+                    m.into_iter()
+                        .map(|(k, v)| (Cow::Owned(k), Cow::Owned(v)))
+                        .collect()
+                }))
+            })
+            .as_ref()
+            .map_err(|&e| e)
+    }
+}
+
 impl<M: KeyedDataMarker> IterableDataProvider<M> for DatagenProvider
 where
     DatagenProvider: IterableDataProviderCached<M>,
 {
     fn supported_requests(&self) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        self.supported_locales_cache
-            .insert_with(M::KEY, || Box::new(self.supported_locales_cached()))
-            .clone()
+        Ok(self
+            .populate_supported_requests_cache()?
+            .iter()
+            .map(|(k, v)| (k.clone().into_owned(), v.clone().into_owned()))
+            .collect())
     }
 
     fn supports_request(
         &self,
         locale: &DataLocale,
         key_attributes: &DataKeyAttributes,
     ) -> Result<bool, DataError> {
-        self.supported_locales_cache
-            .insert_with(M::KEY, || Box::new(self.supported_locales_cached()))
-            .as_ref()
-            .map_err(|e| *e)
-            .map(|v| v.contains(&(locale.clone(), key_attributes.clone())))
+        Ok(self
+            .populate_supported_requests_cache()?
+            .contains(&(Cow::Borrowed(locale), Cow::Borrowed(key_attributes))))
     }
 }
diff --git a/provider/datagen/src/transform/cldr/characters/mod.rs b/provider/datagen/src/transform/cldr/characters/mod.rs
@@ -48,7 +48,7 @@ macro_rules! exemplar_chars_impls {
         }
 
         impl IterableDataProviderCached<$data_marker_name> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
                 Ok(self

diff --git a/provider/datagen/src/transform/cldr/currency/essentials.rs b/provider/datagen/src/transform/cldr/currency/essentials.rs
@@ -107,7 +107,7 @@ impl DataProvider<CurrencyEssentialsV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<CurrencyEssentialsV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/datetime/mod.rs b/provider/datagen/src/transform/cldr/datetime/mod.rs
@@ -228,7 +228,7 @@ macro_rules! impl_data_provider {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
                 let mut r = HashSet::new();

diff --git a/provider/datagen/src/transform/cldr/datetime/neo.rs b/provider/datagen/src/transform/cldr/datetime/neo.rs
@@ -168,26 +168,27 @@ impl DatagenProvider {
         })
     }
 
-    fn supported_locales_neo(
+    fn supported_requests_neo(
         &self,
         calendar: Value,
         keylengths: &'static [TinyAsciiStr<8>],
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        let mut r = HashSet::new();
-
         let cldr_cal = supported_cals()
             .get(&calendar)
             .ok_or_else(|| DataErrorKind::MissingLocale.into_error())?;
-        r.extend(self.cldr()?.dates(cldr_cal).list_langs()?.flat_map(|lid| {
-            keylengths.iter().map(move |&length| {
-                (
-                    DataLocale::from(lid.clone()),
-                    DataKeyAttributes::from_tinystr(length),
-                )
+        Ok(self
+            .cldr()?
+            .dates(cldr_cal)
+            .list_langs()?
+            .flat_map(|lid| {
+                keylengths.iter().map(move |&length| {
+                    (
+                        DataLocale::from(lid.clone()),
+                        DataKeyAttributes::from_tinystr(length),
+                    )
+                })
             })
-        }));
-
-        Ok(r)
+            .collect())
     }
 }
 
@@ -671,7 +672,7 @@ impl DataProvider<TimePatternV1Marker> for DatagenProvider {
 // and we can use a union of the H12/H24 key lengths arrays, instead checking for preferred hc
 // in timepattern_convert
 impl IterableDataProviderCached<TimePatternV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         let calendar = value!("gregory");
@@ -736,10 +737,10 @@ macro_rules! impl_symbols_datagen {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-                self.supported_locales_neo(value!($calendar), $lengths)
+                self.supported_requests_neo(value!($calendar), $lengths)
             }
         }
     };
@@ -754,10 +755,10 @@ macro_rules! impl_pattern_datagen {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-                self.supported_locales_neo(value!($calendar), $lengths)
+                self.supported_requests_neo(value!($calendar), $lengths)
             }
         }
     };

diff --git a/provider/datagen/src/transform/cldr/datetime/neo_skeleton.rs b/provider/datagen/src/transform/cldr/datetime/neo_skeleton.rs
@@ -206,7 +206,7 @@ impl DataProvider<TimeNeoSkeletonPatternsV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<TimeNeoSkeletonPatternsV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         self.neo_time_skeleton_supported_locales()
@@ -230,7 +230,7 @@ macro_rules! impl_neo_skeleton_datagen {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
                 self.neo_date_skeleton_supported_locales(&value!($calendar))

diff --git a/provider/datagen/src/transform/cldr/datetime/week_data.rs b/provider/datagen/src/transform/cldr/datetime/week_data.rs
@@ -16,7 +16,7 @@ use icu_provider::prelude::*;
 use std::collections::HashSet;
 
 impl IterableDataProviderCached<WeekDataV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         let week_data: &cldr_serde::week_data::Resource = self
@@ -200,10 +200,10 @@ impl DataProvider<WeekDataV2Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<WeekDataV2Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        IterableDataProviderCached::<WeekDataV1Marker>::supported_locales_cached(self)
+        IterableDataProviderCached::<WeekDataV1Marker>::supported_requests_cached(self)
     }
 }
 

diff --git a/provider/datagen/src/transform/cldr/decimal/compact.rs b/provider/datagen/src/transform/cldr/decimal/compact.rs
@@ -108,18 +108,18 @@ impl DataProvider<LongCompactDecimalFormatDataV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<ShortCompactDecimalFormatDataV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        self.supported_locales_for_numbers()
+        self.supported_requests_for_numbers()
     }
 }
 
 impl IterableDataProviderCached<LongCompactDecimalFormatDataV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        self.supported_locales_for_numbers()
+        self.supported_requests_for_numbers()
     }
 }
 

diff --git a/provider/datagen/src/transform/cldr/decimal/mod.rs b/provider/datagen/src/transform/cldr/decimal/mod.rs
@@ -77,7 +77,7 @@ impl DatagenProvider {
             .collect())
     }
 
-    fn supported_locales_for_numbers(
+    fn supported_requests_for_numbers(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/decimal/symbols.rs b/provider/datagen/src/transform/cldr/decimal/symbols.rs
@@ -49,10 +49,10 @@ impl DataProvider<DecimalSymbolsV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<DecimalSymbolsV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
-        self.supported_locales_for_numbers()
+        self.supported_requests_for_numbers()
     }
 }
 

diff --git a/provider/datagen/src/transform/cldr/displaynames/language.rs b/provider/datagen/src/transform/cldr/displaynames/language.rs
@@ -52,7 +52,7 @@ impl DataProvider<LocaleDisplayNamesV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<LanguageDisplayNamesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self
@@ -73,7 +73,7 @@ impl IterableDataProviderCached<LanguageDisplayNamesV1Marker> for DatagenProvide
 }
 
 impl IterableDataProviderCached<LocaleDisplayNamesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/displaynames/region.rs b/provider/datagen/src/transform/cldr/displaynames/region.rs
@@ -37,7 +37,7 @@ impl DataProvider<RegionDisplayNamesV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<RegionDisplayNamesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/displaynames/script.rs b/provider/datagen/src/transform/cldr/displaynames/script.rs
@@ -37,7 +37,7 @@ impl DataProvider<ScriptDisplayNamesV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<ScriptDisplayNamesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/displaynames/variant.rs b/provider/datagen/src/transform/cldr/displaynames/variant.rs
@@ -37,7 +37,7 @@ impl DataProvider<VariantDisplayNamesV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<VariantDisplayNamesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/list/mod.rs b/provider/datagen/src/transform/cldr/list/mod.rs
@@ -128,7 +128,7 @@ macro_rules! implement {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
                 Ok(self

diff --git a/provider/datagen/src/transform/cldr/percent/mod.rs b/provider/datagen/src/transform/cldr/percent/mod.rs
@@ -33,7 +33,7 @@ impl DataProvider<PercentEssentialsV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<PercentEssentialsV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self

diff --git a/provider/datagen/src/transform/cldr/personnames/person_names_format_data_providers.rs b/provider/datagen/src/transform/cldr/personnames/person_names_format_data_providers.rs
@@ -7,11 +7,11 @@ use std::borrow::Cow;
 use std::collections::HashSet;
 
 use icu_experimental::personnames::provider::*;
-use icu_provider::datagen::IterableDataProvider;
 use icu_provider::prelude::*;
 use zerovec::VarZeroVec;
 
 use crate::provider::transform::cldr::cldr_serde::personnames::person_name_format_json_struct::Resource;
+use crate::provider::IterableDataProviderCached;
 
 impl DataProvider<PersonNamesFormatV1Marker> for crate::DatagenProvider {
     fn load(&self, req: DataRequest) -> Result<DataResponse<PersonNamesFormatV1Marker>, DataError> {
@@ -34,8 +34,10 @@ impl DataProvider<PersonNamesFormatV1Marker> for crate::DatagenProvider {
     }
 }
 
-impl IterableDataProvider<PersonNamesFormatV1Marker> for crate::DatagenProvider {
-    fn supported_requests(&self) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
+impl IterableDataProviderCached<PersonNamesFormatV1Marker> for crate::DatagenProvider {
+    fn supported_requests_cached(
+        &self,
+    ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self
             .cldr()?
             .personnames()

diff --git a/provider/datagen/src/transform/cldr/plurals/mod.rs b/provider/datagen/src/transform/cldr/plurals/mod.rs
@@ -65,7 +65,7 @@ macro_rules! implement {
         }
 
         impl IterableDataProviderCached<$marker> for DatagenProvider {
-            fn supported_locales_cached(
+            fn supported_requests_cached(
                 &self,
             ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
                 Ok(self
@@ -124,7 +124,7 @@ impl DataProvider<PluralRangesV1Marker> for DatagenProvider {
 }
 
 impl IterableDataProviderCached<PluralRangesV1Marker> for DatagenProvider {
-    fn supported_locales_cached(
+    fn supported_requests_cached(
         &self,
     ) -> Result<HashSet<(DataLocale, DataKeyAttributes)>, DataError> {
         Ok(self