fixes

components/locale/benches/locale_canonicalizer.rs

@@ -8,7 +8,7 @@ use icu_locale::LocaleExpander;
 use icu_locale_core::Locale;
 
 fn canonicalize_bench(c: &mut Criterion) {
-    let lc = LocaleCanonicalizer::new();
+    let lc = LocaleCanonicalizer::new_common();
 
     let mut group = c.benchmark_group("uncanonicalized");
 
@@ -38,7 +38,7 @@ fn canonicalize_bench(c: &mut Criterion) {
 }
 
 fn canonicalize_noop_bench(c: &mut Criterion) {
-    let lc = LocaleCanonicalizer::new();
+    let lc = LocaleCanonicalizer::new_common();
 
     let mut group = c.benchmark_group("canonicalized");
 

components/locale/src/canonicalizer.rs

@@ -28,7 +28,7 @@ use tinystr::TinyAsciiStr;
 /// use icu::locale::Locale;
 /// use icu::locale::{LocaleCanonicalizer, TransformResult};
 ///
-/// let lc = LocaleCanonicalizer::new();
+/// let lc = LocaleCanonicalizer::new_extended();
 ///
 /// let mut locale: Locale = "ja-Latn-fonipa-hepburn-heploc".parse().unwrap();
 /// assert_eq!(lc.canonicalize(&mut locale), TransformResult::Modified);
@@ -202,7 +202,6 @@ impl LocaleCanonicalizer<LocaleExpander> {
     ///
     /// [📚 Help choosing a constructor](icu_provider::constructors)
     #[cfg(feature = "compiled_data")]
-    #[allow(clippy::new_without_default)] // Deliberate choice, see #5554
     pub const fn new_common() -> Self {
         Self::new_with_expander(LocaleExpander::new_common())
     }
@@ -217,7 +216,7 @@ impl LocaleCanonicalizer<LocaleExpander> {
         ]
     );
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_common)]
     pub fn try_new_common_unstable<P>(provider: &P) -> Result<Self, DataError>
     where
         P: DataProvider<AliasesV2Marker>
@@ -251,7 +250,7 @@ impl LocaleCanonicalizer<LocaleExpander> {
         ]
     );
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_extended)]
     pub fn try_new_extended_unstable<P>(provider: &P) -> Result<Self, DataError>
     where
         P: DataProvider<AliasesV2Marker>
@@ -319,7 +318,7 @@ impl<Expander: AsRef<LocaleExpander>> LocaleCanonicalizer<Expander> {
     /// ```
     /// use icu::locale::{Locale, LocaleCanonicalizer, TransformResult};
     ///
-    /// let lc = LocaleCanonicalizer::new();
+    /// let lc = LocaleCanonicalizer::new_extended();
     ///
     /// let mut locale: Locale = "ja-Latn-fonipa-hepburn-heploc".parse().unwrap();
     /// assert_eq!(lc.canonicalize(&mut locale), TransformResult::Modified);

components/locale/src/lib.rs

@@ -25,7 +25,7 @@
 //! use icu::locale::Locale;
 //! use icu::locale::{LocaleCanonicalizer, TransformResult};
 //!
-//! let lc = LocaleCanonicalizer::new();
+//! let lc = LocaleCanonicalizer::new_extended();
 //!
 //! let mut locale: Locale = "ja-Latn-fonipa-hepburn-heploc"
 //!     .parse()

components/locale/tests/locale_canonicalizer.rs

@@ -58,7 +58,7 @@ fn test_minimize() {
 
 #[test]
 fn test_canonicalize() {
-    let lc = LocaleCanonicalizer::new();
+    let lc = LocaleCanonicalizer::new_extended();
 
     let testcases: Vec<fixtures::CanonicalizationTest> =
         serde_json::from_str(include_str!("fixtures/canonicalize.json"))

components/segmenter/src/grapheme.rs

@@ -138,6 +138,7 @@ impl GraphemeClusterSegmenter {
     ///
     /// [📚 Help choosing a constructor](icu_provider::constructors)
     #[cfg(feature = "compiled_data")]
+    #[allow(clippy::new_without_default)] // Deliberate choice, see #5554
     pub fn new() -> Self {
         Self {
             payload: DataPayload::from_static_ref(

components/segmenter/src/line.rs

@@ -398,9 +398,9 @@ impl LineSegmenter {
         Self::new_auto_with_options(Default::default())
     }
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_auto)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_auto)]
     #[cfg(feature = "auto")]
-    pub fn try_new_auto_unstable<D>(provider: &D) -> Result<Self, DataError>
+    pub fn try_new_root_auto_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<LineBreakDataV2Marker>
             + DataProvider<LstmForWordLineAutoV1Marker>
@@ -428,8 +428,8 @@ impl LineSegmenter {
     }
 
     #[cfg(feature = "lstm")]
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_lstm)]
-    pub fn try_new_lstm_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_lstm)]
+    pub fn try_new_root_lstm_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<LineBreakDataV2Marker>
             + DataProvider<LstmForWordLineAutoV1Marker>
@@ -455,8 +455,8 @@ impl LineSegmenter {
         Self::new_dictionary_with_options(Default::default())
     }
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_dictionary)]
-    pub fn try_new_dictionary_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_dictionary)]
+    pub fn try_new_root_dictionary_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<LineBreakDataV2Marker>
             + DataProvider<DictionaryForWordLineExtendedV1Marker>
@@ -471,7 +471,7 @@ impl LineSegmenter {
     ///
     /// The current behavior, which is subject to change, is to use the LSTM model when available.
     ///
-    /// See also [`Self::new_auto`].
+    /// See also [`Self::new_root_auto`].
     ///
     /// ✨ *Enabled with the `compiled_data` and `auto` Cargo features.*
     ///
@@ -515,7 +515,7 @@ impl LineSegmenter {
     /// The LSTM, or Long Term Short Memory, is a machine learning model. It is smaller than
     /// the full dictionary but more expensive during segmentation (inference).
     ///
-    /// See also [`Self::new_dictionary`].
+    /// See also [`Self::new_root_lstm`].
     ///
     /// ✨ *Enabled with the `compiled_data` and `lstm` Cargo features.*
     ///
@@ -569,7 +569,7 @@ impl LineSegmenter {
     /// The dictionary model uses a list of words to determine appropriate breakpoints. It is
     /// faster than the LSTM model but requires more data.
     ///
-    /// See also [`Self::new_dictionary`].
+    /// See also [`Self::new_root_dictionary`].
     ///
     /// ✨ *Enabled with the `compiled_data` Cargo feature.*
     ///
@@ -1522,7 +1522,7 @@ mod tests {
 
     #[test]
     fn linebreak() {
-        let segmenter = LineSegmenter::try_new_dictionary_unstable(&crate::provider::Baked)
+        let segmenter = LineSegmenter::try_new_root_dictionary_unstable(&crate::provider::Baked)
             .expect("Data exists");
 
         let mut iter = segmenter.segment_str("hello world");

components/segmenter/src/sentence.rs

@@ -128,8 +128,8 @@ impl SentenceSegmenter {
         }
     }
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new)]
-    pub fn try_new_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root)]
+    pub fn try_new_root_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<SentenceBreakDataV2Marker> + ?Sized,
     {

components/segmenter/src/word.rs

@@ -222,8 +222,8 @@ impl WordSegmenter {
     }
 
     #[cfg(feature = "auto")]
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_auto)]
-    pub fn try_new_auto_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_auto)]
+    pub fn try_new_root_auto_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<WordBreakDataV2Marker>
             + DataProvider<WordBreakDataOverrideV1Marker>
@@ -252,7 +252,7 @@ impl WordSegmenter {
     );
 
     #[cfg(feature = "auto")]
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_auto)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_auto)]
     pub fn try_new_auto_with_options_unstable<D>(
         provider: &D,
         options: WordBreakOptions,
@@ -334,8 +334,8 @@ impl WordSegmenter {
     }
 
     #[cfg(feature = "lstm")]
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_lstm)]
-    pub fn try_new_lstm_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_lstm)]
+    pub fn try_new_root_lstm_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<WordBreakDataV2Marker>
             + DataProvider<WordBreakDataOverrideV1Marker>
@@ -363,7 +363,7 @@ impl WordSegmenter {
     );
 
     #[cfg(feature = "lstm")]
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_lstm)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_lstm)]
     pub fn try_new_lstm_with_options_unstable<D>(
         provider: &D,
         options: WordBreakOptions,
@@ -437,8 +437,8 @@ impl WordSegmenter {
         }
     }
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_dictionary)]
-    pub fn try_new_dictionary_unstable<D>(provider: &D) -> Result<Self, DataError>
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_dictionary)]
+    pub fn try_new_root_dictionary_unstable<D>(provider: &D) -> Result<Self, DataError>
     where
         D: DataProvider<WordBreakDataV2Marker>
             + DataProvider<WordBreakDataOverrideV1Marker>
@@ -465,7 +465,7 @@ impl WordSegmenter {
         ]
     );
 
-    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_dictionary)]
+    #[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_root_dictionary)]
     pub fn try_new_dictionary_with_options_unstable<D>(
         provider: &D,
         options: WordBreakOptions,

ffi/capi/src/locale_directionality.rs

@@ -51,7 +51,12 @@ pub mod ffi {
             )))
         }
         /// Construct a new LocaleDirectionality instance using compiled data.
-        #[diplomat::rust_link(icu::locale::LocaleDirectionality::new_with_expander, FnInStruct)]
+        #[diplomat::rust_link(icu::locale::LocaleDirectionality::new_extended, FnInStruct)]
+        #[diplomat::rust_link(
+            icu::locale::LocaleDirectionality::new_with_expander,
+            FnInStruct,
+            hidden
+        )]
         #[diplomat::attr(auto, named_constructor = "extended")]
         #[cfg(feature = "compiled_data")]
         pub fn create_extended() -> Box<LocaleDirectionality> {
@@ -61,7 +66,12 @@ pub mod ffi {
         }
 
         /// Construct a new LocaleDirectionality instance using a particular data source.
-        #[diplomat::rust_link(icu::locale::LocaleDirectionality::new_with_expander, FnInStruct)]
+        #[diplomat::rust_link(icu::locale::LocaleDirectionality::new_extended, FnInStruct)]
+        #[diplomat::rust_link(
+            icu::locale::LocaleDirectionality::new_with_expander,
+            FnInStruct,
+            hidden
+        )]
         #[diplomat::attr(supports = fallible_constructors, named_constructor = "extended_with_provider")]
         #[cfg(feature = "buffer_provider")]
         pub fn create_extended_with_provider(

provider/source/src/segmenter/lstm.rs

@@ -246,7 +246,11 @@ mod tests {
             provider.as_any_provider(),
         );
 
-        let segmenter = LineSegmenter::try_new_lstm_with_any_provider(&provider).unwrap();
+        let segmenter = LineSegmenter::try_new_lstm_with_options_with_any_provider(
+            &provider,
+            Default::default(),
+        )
+        .unwrap();
 
         const TEST_STR: &str = "ภาษาไทยภาษาไทย";
         let utf16: Vec<u16> = TEST_STR.encode_utf16().collect();

tutorials/c-tiny/segmenter/test.c

@@ -8,7 +8,7 @@
 #include <stdio.h>
 
 int main(int argc, char *argv[]) {
-    LineSegmenter* segmenter = icu4x_LineSegmenter_create_auto_mv1();
+    LineSegmenter* segmenter = icu4x_LineSegmenter_create_root_auto_mv1();
 
 
     char output[40];