Merge branch 'main' into std

Cargo.toml

@@ -35,7 +35,7 @@ members = [
     # Provider
     "provider/adapters",
     "provider/baked",
-    "provider/bikeshed",
+    "provider/source",
     "provider/blob",
     "provider/core",
     "provider/core/macros",
@@ -152,7 +152,7 @@ icu_harfbuzz = { version = "~0.2.0", path = "ffi/harfbuzz", default-features = f
 
 # Provider
 icu_provider_export = { version = "~1.5.0", path = "provider/export", default-features = false }
-icu_datagen_bikeshed = { version = "~1.5.0", path = "provider/bikeshed", default-features = false }
+icu_provider_source = { version = "~1.5.0", path = "provider/source", default-features = false }
 icu_provider = { version = "~1.5.0", path = "provider/core", default-features = false }
 icu_provider_macros = { version = "~1.5.0", path = "provider/core/macros", default-features = false }
 icu_provider_adapters = { version = "~1.5.0", path = "provider/adapters", default-features = false }

components/collator/tests/tests.rs

@@ -691,7 +691,7 @@ fn test_vi() {
 }
 
 #[test]
-// See DatagenProvider test_zh_non_baked for gb2312 and big5han tests
+// See SourceDataProvider test_zh_non_baked for gb2312 and big5han tests
 fn test_zh() {
     // Note: ㄅ is Bopomofo.
 

components/experimental/src/dimension/provider/units.rs

@@ -27,19 +27,7 @@ pub struct UnitsDisplayNameV1<'data> {
     // TODO: use `MeasureUnit` for the units key instead of strings.
     /// Contains the long width patterns for the units.
     #[cfg_attr(feature = "serde", serde(borrow))]
-    pub long: ZeroMap<'data, Count, str>,
-
-    // TODO: store the pattern in a SinglePattern.
-    // TODO: use `MeasureUnit` for the units key instead of strings.
-    /// Contains the short width patterns for the units.
-    #[cfg_attr(feature = "serde", serde(borrow))]
-    pub short: ZeroMap<'data, Count, str>,
-
-    // TODO: store the pattern in a SinglePattern.
-    // TODO: use `MeasureUnit` for the units key instead of strings.
-    /// Contains the narrow width patterns for the units.
-    #[cfg_attr(feature = "serde", serde(borrow))]
-    pub narrow: ZeroMap<'data, Count, str>,
+    pub patterns: ZeroMap<'data, Count, str>,
 }
 
 // TODO: revise this.

components/experimental/src/dimension/units/format.rs

@@ -15,12 +15,12 @@ use writeable::Writeable;
 
 use crate::alloc::borrow::ToOwned;
 use crate::dimension::provider::units::{Count, UnitsDisplayNameV1};
-use crate::dimension::units::options::{UnitsFormatterOptions, Width};
 
 pub struct FormattedUnit<'l> {
     pub(crate) value: &'l FixedDecimal,
     pub(crate) unit: &'l str,
-    pub(crate) options: &'l UnitsFormatterOptions,
+    // TODO: review using options and essentials.
+    // pub(crate) _options: &'l UnitsFormatterOptions,
     // pub(crate) essential: &'l UnitsEssentialsV1<'l>,
     pub(crate) display_name: &'l UnitsDisplayNameV1<'l>,
     pub(crate) fixed_decimal_formatter: &'l FixedDecimalFormatter,
@@ -35,12 +35,11 @@ impl<'l> Writeable for FormattedUnit<'l> {
         let plural_category = self.plural_rules.category_for(self.value);
         let count = Count::from(plural_category);
         let mut unit_pattern = None;
-        let display_name = match self.options.width {
-            Width::Short => self.display_name.short.get(&count),
-            Width::Long => self.display_name.long.get(&count),
-            Width::Narrow => self.display_name.narrow.get(&count),
-        }
-        .unwrap_or_else(|| unit_pattern.insert("{0} ".to_owned() + self.unit));
+        let display_name = self
+            .display_name
+            .patterns
+            .get(&count)
+            .unwrap_or_else(|| unit_pattern.insert("{0} ".to_owned() + self.unit));
 
         // TODO: once the patterns are implemented to be used in the data side, we do not need this.
         let pattern =
@@ -60,6 +59,8 @@ fn test_basic() {
     use writeable::Writeable;
 
     use crate::dimension::units::formatter::UnitsFormatter;
+    use crate::dimension::units::options::UnitsFormatterOptions;
+    use crate::dimension::units::options::Width;
 
     let locale = locale!("en-US").into();
     let meter = "meter";

components/experimental/src/dimension/units/formatter.rs

@@ -10,10 +10,11 @@ use icu_decimal::FixedDecimalFormatter;
 use icu_plurals::PluralRules;
 use icu_provider::DataPayload;
 
-use super::super::provider::units::UnitsDisplayNameV1Marker;
 use super::format::FormattedUnit;
-use super::options::UnitsFormatterOptions;
+use super::options::{UnitsFormatterOptions, Width};
+use crate::dimension::provider::units::UnitsDisplayNameV1Marker;
 use icu_provider::prelude::*;
+use smallvec::SmallVec;
 
 extern crate alloc;
 
@@ -27,7 +28,7 @@ extern crate alloc;
 pub struct UnitsFormatter {
     /// Options bag for the units formatter to determine the behavior of the formatter.
     /// for example: width of the units.
-    options: UnitsFormatterOptions,
+    _options: UnitsFormatterOptions,
 
     // /// Essential data for the units formatter.
     // essential: DataPayload<UnitsEssentialsV1Marker>,
@@ -53,6 +54,20 @@ impl UnitsFormatter {
         ]
     );
 
+    // TODO: Remove this function once we have separate markers for different widths.
+    #[inline]
+    fn attribute(width: Width, unit: &str) -> SmallVec<[u8; 32]> {
+        let mut buffer: SmallVec<[u8; 32]> = SmallVec::new();
+        let length = match width {
+            Width::Short => "short-",
+            Width::Narrow => "narrow-",
+            Width::Long => "long-",
+        };
+        buffer.extend_from_slice(length.as_bytes());
+        buffer.extend_from_slice(unit.as_bytes());
+        buffer
+    }
+
     /// Creates a new [`UnitsFormatter`] from compiled locale data and an options bag.
     ///
     /// ✨ *Enabled with the `compiled_data` Cargo feature.*
@@ -69,8 +84,10 @@ impl UnitsFormatter {
 
         let plural_rules = PluralRules::try_new_cardinal(locale)?;
 
-        let unit_attribute = DataMarkerAttributes::try_from_str(unit)
-            .map_err(|_| DataError::custom("Failed to parse unit"))?;
+        // TODO: Remove this allocation once we have separate markers for different widths.
+        let attribute = Self::attribute(options.width, unit);
+        let unit_attribute = DataMarkerAttributes::try_from_utf8(&attribute[..attribute.len()])
+            .map_err(|_| DataError::custom("Failed to create a data marker"))?;
 
         let display_name = crate::provider::Baked
             .load(DataRequest {
@@ -83,7 +100,7 @@ impl UnitsFormatter {
             .payload;
 
         Ok(Self {
-            options,
+            _options: options,
             display_name,
             fixed_decimal_formatter,
             plural_rules,
@@ -111,8 +128,10 @@ impl UnitsFormatter {
 
         let plural_rules = PluralRules::try_new_cardinal_unstable(provider, locale)?;
 
-        let unit_attribute = DataMarkerAttributes::try_from_str(unit)
-            .map_err(|_| DataError::custom("Failed to parse unit"))?;
+        // TODO: Remove this allocation once we have separate markers for different widths.
+        let attribute = Self::attribute(options.width, unit);
+        let unit_attribute = DataMarkerAttributes::try_from_utf8(&attribute[..attribute.len()])
+            .map_err(|_| DataError::custom("Failed to create a data marker"))?;
 
         let display_name = provider
             .load(DataRequest {
@@ -125,7 +144,7 @@ impl UnitsFormatter {
             .payload;
 
         Ok(Self {
-            options,
+            _options: options,
             display_name,
             fixed_decimal_formatter,
             plural_rules,
@@ -141,7 +160,6 @@ impl UnitsFormatter {
         FormattedUnit {
             value,
             unit,
-            options: &self.options,
             display_name: self.display_name.get(),
             fixed_decimal_formatter: &self.fixed_decimal_formatter,
             plural_rules: &self.plural_rules,