MultiForkByKeyProvider for all datagen

provider/cldr/src/cldr_paths.rs

@@ -61,9 +61,6 @@ pub trait CldrPaths: std::fmt::Debug {
         // more calendars here
         vec
     }
-
-    /// Path to uprops TOML data, which is required by some CLDR transformers
-    fn uprops(&self) -> Result<PathBuf, Error>;
 }
 
 /// An implementation of [`CldrPaths`] for multiple separate local CLDR JSON directories per
@@ -112,12 +109,6 @@ impl CldrPaths for CldrPathsLocal {
     fn cldr_misc(&self) -> Result<PathBuf, Error> {
         self.cldr_misc.clone().map_err(|e| e.into())
     }
-    fn uprops(&self) -> Result<PathBuf, Error> {
-        Err(Error::Custom(
-            "This implementation does not know about uprops".to_owned(),
-            None,
-        ))
-    }
 }
 
 impl Default for CldrPathsLocal {
@@ -162,8 +153,6 @@ pub struct CldrPathsAllInOne {
     pub cldr_json_root: PathBuf,
     /// CLDR JSON directory suffix: probably either "modern" or "full"
     pub locale_subset: String,
-    /// Path to uprops TOML root directory. Required by some CLDR transformers
-    pub uprops_root: Option<PathBuf>,
 }
 
 impl CldrPaths for CldrPathsAllInOne {
@@ -200,18 +189,12 @@ impl CldrPaths for CldrPathsAllInOne {
             .clone()
             .join(format!("cldr-misc-{}", self.locale_subset)))
     }
-    fn uprops(&self) -> Result<PathBuf, Error> {
-        self.uprops_root
-            .clone()
-            .ok_or_else(|| Error::Custom("The uprops root has not been set".to_owned(), None))
-    }
 }
 
 #[cfg(test)]
 pub(crate) fn for_test() -> CldrPathsAllInOne {
     CldrPathsAllInOne {
         cldr_json_root: icu_testdata::paths::cldr_json_root(),
         locale_subset: "full".to_string(),
-        uprops_root: Some(icu_testdata::paths::uprops_toml_root()),
     }
 }

provider/cldr/src/download/cldr_allinone.rs

@@ -65,13 +65,12 @@ impl CldrAllInOneDownloader {
         })
     }
 
-    pub fn download(self, uprops_root: Option<PathBuf>) -> Result<CldrPathsAllInOne, Error> {
+    pub fn download(self) -> Result<CldrPathsAllInOne, Error> {
         // TODO(#297): Implement this async.
         let downloaded = io_util::download_and_unzip(&self.url, &self.cache_dir)?;
         Ok(CldrPathsAllInOne {
             cldr_json_root: downloaded,
             locale_subset: self.locale_subset,
-            uprops_root,
         })
     }
 }

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

@@ -10,22 +10,20 @@ use icu_list::provider::*;
 use icu_locid_macros::langid;
 use icu_provider::iter::IterableResourceProvider;
 use icu_provider::prelude::*;
-use std::convert::TryFrom;
 use std::path::PathBuf;
 
 /// A data provider reading from CLDR JSON list rule files.
 #[derive(Debug)]
 pub struct ListProvider {
     cldr_misc: PathBuf,
-    uprops_path: PathBuf,
+    uprops_root: PathBuf,
 }
 
-impl TryFrom<&dyn CldrPaths> for ListProvider {
-    type Error = Error;
-    fn try_from(cldr_paths: &dyn CldrPaths) -> Result<Self, Self::Error> {
+impl ListProvider {
+    pub fn try_from(cldr_paths: &dyn CldrPaths, uprops_root: PathBuf) -> Result<Self, Error> {
         Ok(Self {
             cldr_misc: cldr_paths.cldr_misc()?,
-            uprops_path: cldr_paths.uprops()?,
+            uprops_root,
         })
     }
 }
@@ -116,8 +114,8 @@ impl<M: ResourceMarker<Yokeable = ListFormatterPatternsV1<'static>>> ResourcePro
                     &format!(
                         "[^{}]",
                         icu_properties::sets::get_for_script(
-                            &icu_provider_uprops::PropertiesDataProvider::try_new(
-                                &self.uprops_path
+                            &icu_provider_uprops::EnumeratedPropertyUnicodeSetDataProvider::try_new(
+                                &self.uprops_root
                             )
                             .map_err(|e| DataError::custom("Properties data provider error")
                                 .with_display_context(&e))?,

provider/cldr/src/transform/mod.rs

@@ -22,12 +22,14 @@ use icu_provider::iter::IterableDynProvider;
 use icu_provider::prelude::*;
 use icu_provider::serde::SerializeMarker;
 use std::convert::TryFrom;
+use std::path::PathBuf;
 
 pub struct CldrJsonDataProvider;
 
 impl CldrJsonDataProvider {
     pub fn try_new(
         cldr_paths: &dyn CldrPaths,
+        uprops_root: PathBuf,
     ) -> Result<MultiForkByKeyProvider<Box<dyn IterableDynProvider<SerializeMarker>>>, Error> {
         Ok(MultiForkByKeyProvider {
             providers: vec![
@@ -55,7 +57,7 @@ impl CldrJsonDataProvider {
                 Box::new(plurals::PluralsProvider::try_from(cldr_paths)?),
                 Box::new(time_zones::TimeZonesProvider::try_from(cldr_paths)?),
                 #[cfg(feature = "icu_list")]
-                Box::new(list::ListProvider::try_from(cldr_paths)?),
+                Box::new(list::ListProvider::try_from(cldr_paths, uprops_root)?),
             ],
         })
     }

provider/core/src/dynutil.rs

@@ -159,10 +159,10 @@ where
 /// [`SerializeMarker`]: (crate::serde::SerializeMarker)
 #[macro_export]
 macro_rules! impl_dyn_provider {
-    ($provider:ty, { $($pat:pat => $struct_m:ty),+, }, ANY) => {
+    ($provider:ty, { $($pat:pat $(if $guard:expr)? => $struct_m:ty),+, }, ANY) => {
         $crate::impl_dyn_provider!(
             $provider,
-            { $($pat => $struct_m),+, },
+            { $($pat $(if $guard)? => $struct_m),+, },
             $crate::any::AnyMarker
         );
     };
@@ -173,11 +173,11 @@ macro_rules! impl_dyn_provider {
             $crate::any::AnyMarker
         );
     };
-    ($provider:ty, { $($pat:pat => $struct_m:ty),+, }, SERDE_SE) => {
+    ($provider:ty, { $($pat:pat $(if $guard:expr)? => $struct_m:ty),+, }, SERDE_SE) => {
         // If this fails to compile, enable the "serialize" feature on this crate.
         $crate::impl_dyn_provider!(
             $provider,
-            { $($pat => $struct_m),+, },
+            { $($pat $(if $guard)? => $struct_m),+, },
             $crate::serde::SerializeMarker
         );
     };
@@ -189,7 +189,7 @@ macro_rules! impl_dyn_provider {
             $crate::serde::SerializeMarker
         );
     };
-    ($provider:ty, { $($pat:pat => $struct_m:ty),+, }, $dyn_m:path) => {
+    ($provider:ty, { $($pat:pat $(if $guard:expr)? => $struct_m:ty),+, }, $dyn_m:path) => {
         impl $crate::DynProvider<$dyn_m> for $provider
         {
             fn load_payload(
@@ -202,7 +202,7 @@ macro_rules! impl_dyn_provider {
             > {
                 match key {
                     $(
-                        $pat => {
+                        $pat $(if $guard)? => {
                             let result: $crate::DataResponse<$struct_m> =
                                 $crate::DynProvider::<$struct_m>::load_payload(self, key, req)?;
                             Ok(DataResponse {
@@ -213,8 +213,6 @@ macro_rules! impl_dyn_provider {
                             })
                         }
                     )+,
-                    // Don't complain if the call site has its own wildcard match
-                    #[allow(unreachable_patterns)]
                     _ => Err($crate::DataErrorKind::MissingResourceKey.with_req(key, req))
                 }
             }
@@ -224,12 +222,10 @@ macro_rules! impl_dyn_provider {
             fn supported_options_for_key(&self, key: &$crate::ResourceKey) -> Result<Box<dyn Iterator<Item = $crate::ResourceOptions> + '_>, $crate::DataError> {
                 match *key {
                     $(
-                        $pat => {
+                        $pat $(if $guard)? => {
                             $crate::iter::IterableDynProvider::<$struct_m>::supported_options_for_key(self, key)
                         }
                     )+,
-                    // Don't complain if the call site has its own wildcard match
-                    #[allow(unreachable_patterns)]
                     _ => Err($crate::DataErrorKind::MissingResourceKey.with_key(*key))
                 }
             }

provider/uprops/src/bin_uniset.rs

@@ -50,7 +50,7 @@ impl DynProvider<UnicodePropertyV1Marker> for BinaryPropertyUnicodeSetDataProvid
 }
 
 icu_provider::impl_dyn_provider!(BinaryPropertyUnicodeSetDataProvider, {
-    _ => UnicodePropertyV1Marker,
+    _k if _k.get_path().starts_with("props/") => UnicodePropertyV1Marker,
 }, SERDE_SE);
 
 impl IterableDynProvider<UnicodePropertyV1Marker> for BinaryPropertyUnicodeSetDataProvider {

provider/uprops/src/enum_uniset.rs

@@ -99,7 +99,7 @@ impl DynProvider<UnicodePropertyV1Marker> for EnumeratedPropertyUnicodeSetDataPr
 }
 
 icu_provider::impl_dyn_provider!(EnumeratedPropertyUnicodeSetDataProvider, {
-    _ => UnicodePropertyV1Marker,
+    _k if _k.get_path().starts_with("props/") && _k.get_path().contains('=') => UnicodePropertyV1Marker,
 }, SERDE_SE);
 
 impl IterableDynProvider<UnicodePropertyV1Marker> for EnumeratedPropertyUnicodeSetDataProvider {

provider/uprops/src/lib.rs

@@ -26,6 +26,7 @@ mod script;
 mod uprops_helpers;
 mod uprops_serde;
 
-pub use enum_codepointtrie::EnumeratedPropertyCodePointTrieProvider;
 pub use provider::PropertiesDataProvider;
-pub use script::ScriptWithExtensionsPropertyProvider;
+
+// Required by icu_provider_cldr::transform::list
+pub use enum_uniset::EnumeratedPropertyUnicodeSetDataProvider;

provider/uprops/src/provider.rs

@@ -2,57 +2,38 @@
 // called LICENSE at the top level of the ICU4X source tree
 // (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
 
-use crate::bin_uniset::BinaryPropertyUnicodeSetDataProvider;
-use crate::enum_uniset::EnumeratedPropertyUnicodeSetDataProvider;
-use crate::uprops_helpers::get_last_component_no_version;
-use icu_properties::provider::UnicodePropertyV1Marker;
+use icu_provider::fork::by_key::MultiForkByKeyProvider;
 use icu_provider::iter::IterableDynProvider;
-use icu_provider::prelude::*;
-
+use icu_provider::serde::SerializeMarker;
 use std::path::Path;
 
 /// This data provider returns `UnicodeSet` data inside a `UnicodeProperty`
 /// data struct. The source data is in the form of a directory of TOML file(s)
 /// of data for the property(-ies) desired, as given by the ICU4C property data
 /// exporter tool.
-pub struct PropertiesDataProvider {
-    binary: BinaryPropertyUnicodeSetDataProvider,
-    enumerated: EnumeratedPropertyUnicodeSetDataProvider,
-}
-
+pub struct PropertiesDataProvider;
 impl PropertiesDataProvider {
-    /// Construct a new data provider instance. `root_dir` is the path to the
-    /// root directory containing the property data TOML files.
-    pub fn try_new(root_dir: &Path) -> eyre::Result<Self> {
-        let binary = BinaryPropertyUnicodeSetDataProvider::try_new(root_dir)?;
-        let enumerated = EnumeratedPropertyUnicodeSetDataProvider::try_new(root_dir)?;
-        Ok(Self { binary, enumerated })
-    }
-}
-
-impl DynProvider<UnicodePropertyV1Marker> for PropertiesDataProvider {
-    fn load_payload(
-        &self,
-        key: ResourceKey,
-        req: &DataRequest,
-    ) -> Result<DataResponse<UnicodePropertyV1Marker>, DataError> {
-        if get_last_component_no_version(&key).contains('=') {
-            self.enumerated.load_payload(key, req)
-        } else {
-            self.binary.load_payload(key, req)
-        }
-    }
-}
-
-icu_provider::impl_dyn_provider!(PropertiesDataProvider, {
-    _ => UnicodePropertyV1Marker,
-}, SERDE_SE);
-
-impl IterableDynProvider<UnicodePropertyV1Marker> for PropertiesDataProvider {
-    fn supported_options_for_key(
-        &self,
-        _resc_key: &ResourceKey,
-    ) -> Result<Box<dyn Iterator<Item = ResourceOptions>>, DataError> {
-        Ok(Box::new(core::iter::once(ResourceOptions::default())))
+    pub fn try_new(
+        root_dir: &Path,
+    ) -> eyre::Result<MultiForkByKeyProvider<Box<dyn IterableDynProvider<SerializeMarker>>>> {
+        Ok(MultiForkByKeyProvider {
+            providers: vec![
+                Box::new(
+                    crate::enum_codepointtrie::EnumeratedPropertyCodePointTrieProvider::try_new(
+                        root_dir,
+                    )?,
+                ),
+                Box::new(crate::script::ScriptWithExtensionsPropertyProvider::try_new(root_dir)?),
+                Box::new(
+                    crate::enum_uniset::EnumeratedPropertyUnicodeSetDataProvider::try_new(
+                        root_dir,
+                    )?,
+                ),
+                // Has to go last as it matches all props/ keys.
+                Box::new(
+                    crate::bin_uniset::BinaryPropertyUnicodeSetDataProvider::try_new(root_dir)?,
+                ),
+            ],
+        })
     }
 }