Add CodePointInversionListAndStringList::contains_utf8 (#7363)

For #7358

Author: robertbastian

Cargo.lock

@@ -25,6 +25,7 @@ yoke = { workspace = true, features = ["derive"] }
 zerofrom = { workspace = true, features = ["derive"] }
 zerovec = { workspace = true, features = ["derive", "yoke"] }
 potential_utf = { workspace = true, features = ["zerovec"] }
+utf8_iter = { workspace = true }
 
 serde = { workspace = true, features = ["derive"], optional = true }
 databake = { workspace = true, features = ["derive"], optional = true }

components/collections/Cargo.toml

@@ -173,6 +173,20 @@ impl<'data> CodePointInversionListAndStringList<'data> {
         self.str_list.binary_search(s).is_ok()
     }
 
+    /// See [`Self::contains_str`]
+    pub fn contains_utf8(&self, s: &[u8]) -> bool {
+        use utf8_iter::Utf8CharsEx;
+        let mut chars = s.chars();
+        if let Some(first_char) = chars.next() {
+            if chars.next().is_none() {
+                return self.contains(first_char);
+            }
+        }
+        self.str_list
+            .binary_search_by(|t| t.as_bytes().cmp(s))
+            .is_ok()
+    }
+
     ///
     /// # Examples
     /// ```

components/collections/src/codepointinvliststringlist/mod.rs

@@ -118,6 +118,12 @@ impl EmojiSetDataBorrowed<'_> {
         self.set.contains_str(s)
     }
 
+    /// See [`Self::contains_str`].
+    #[inline]
+    pub fn contains_utf8(self, s: &[u8]) -> bool {
+        self.set.contains_utf8(s)
+    }
+
     /// Check if the set contains the code point.
     #[inline]
     pub fn contains(self, ch: char) -> bool {

components/properties/src/emoji.rs

@@ -1113,6 +1113,13 @@ impl<'data> PropertyUnicodeSet<'data> {
         }
     }
 
+    #[inline]
+    pub(crate) fn contains_utf8(&self, s: &[u8]) -> bool {
+        match *self {
+            Self::CPInversionListStrList(ref l) => l.contains_utf8(s),
+        }
+    }
+
     #[inline]
     pub(crate) fn contains32(&self, cp: u32) -> bool {
         match *self {

components/properties/src/provider.rs

@@ -27,10 +27,7 @@ pub mod ffi {
         )]
         #[diplomat::attr(supports = method_overloading, rename = "contains")]
         pub fn contains_str(&self, s: &DiplomatStr) -> bool {
-            let Ok(s) = core::str::from_utf8(s) else {
-                return false;
-            };
-            self.0.as_borrowed().contains_str(s)
+            self.0.as_borrowed().contains_utf8(s)
         }
         /// Checks whether the code point is in the set.
         #[diplomat::rust_link(

ffi/capi/src/exemplar_chars.rs

@@ -24,12 +24,14 @@ pub mod ffi {
     impl EmojiSetData {
         /// Checks whether the string is in the set.
         #[diplomat::rust_link(icu::properties::EmojiSetDataBorrowed::contains_str, FnInStruct)]
+        #[diplomat::rust_link(
+            icu::properties::EmojiSetDataBorrowed::contains_utf8,
+            FnInStruct,
+            hidden
+        )]
         #[diplomat::attr(supports = method_overloading, rename = "contains")]
         pub fn contains_str(&self, s: &DiplomatStr) -> bool {
-            let Ok(s) = core::str::from_utf8(s) else {
-                return false;
-            };
-            self.0.as_borrowed().contains_str(s)
+            self.0.as_borrowed().contains_utf8(s)
         }
         /// Checks whether the code point is in the set.
         #[diplomat::rust_link(icu::properties::EmojiSetDataBorrowed::contains, FnInStruct)]