New upstream version 1.52.0~beta.3+dfsg1

[rustc.git] / vendor / unicode-normalization / src / lib.rs

diff --git a/vendor/unicode-normalization/src/lib.rs b/vendor/unicode-normalization/src/lib.rs
index 6749adc1e10c2377e15e26f5daea847c866911e3..cb623ba7c8f999979ce5153911d027bd2d924211 100644 (file)
--- a/vendor/unicode-normalization/src/lib.rs
+++ b/vendor/unicode-normalization/src/lib.rs
@@ -59,6 +59,7 @@ pub use crate::quick_check::{
     IsNormalized,
 };
 pub use crate::recompose::Recompositions;
+pub use crate::replace::Replacements;
 pub use crate::stream_safe::StreamSafe;
 pub use crate::tables::UNICODE_VERSION;
 use core::str::Chars;
@@ -71,6 +72,7 @@ mod normalize;
 mod perfect_hash;
 mod quick_check;
 mod recompose;
+mod replace;
 mod stream_safe;
 
 #[rustfmt::skip]
@@ -83,7 +85,9 @@ mod test;
 
 /// Methods for composing and decomposing characters.
 pub mod char {
-    pub use crate::normalize::{compose, decompose_canonical, decompose_compatible};
+    pub use crate::normalize::{
+        compose, decompose_canonical, decompose_cjk_compat_variants, decompose_compatible,
+    };
 
     pub use crate::lookups::{canonical_combining_class, is_combining_mark};
 }
@@ -108,6 +112,18 @@ pub trait UnicodeNormalization<I: Iterator<Item = char>> {
     /// (compatibility decomposition followed by canonical composition).
     fn nfkc(self) -> Recompositions<I>;
 
+    /// A transformation which replaces CJK Compatibility Ideograph codepoints
+    /// with normal forms using Standardized Variation Sequences. This is not
+    /// part of the canonical or compatibility decomposition algorithms, but
+    /// performing it before those algorithms produces normalized output which
+    /// better preserves the intent of the original text.
+    ///
+    /// Note that many systems today ignore variation selectors, so these
+    /// may not immediately help text display as intended, but they at
+    /// least preserve the information in a standardized form, giving
+    /// implementations the option to recognize them.
+    fn cjk_compat_variants(self) -> Replacements<I>;
+
     /// An Iterator over the string with Conjoining Grapheme Joiner characters
     /// inserted according to the Stream-Safe Text Process (UAX15-D4)
     fn stream_safe(self) -> StreamSafe<I>;
@@ -134,6 +150,11 @@ impl<'a> UnicodeNormalization<Chars<'a>> for &'a str {
         recompose::new_compatible(self.chars())
     }
 
+    #[inline]
+    fn cjk_compat_variants(self) -> Replacements<Chars<'a>> {
+        replace::new_cjk_compat_variants(self.chars())
+    }
+
     #[inline]
     fn stream_safe(self) -> StreamSafe<Chars<'a>> {
         StreamSafe::new(self.chars())
@@ -161,6 +182,11 @@ impl<I: Iterator<Item = char>> UnicodeNormalization<I> for I {
         recompose::new_compatible(self)
     }
 
+    #[inline]
+    fn cjk_compat_variants(self) -> Replacements<I> {
+        replace::new_cjk_compat_variants(self)
+    }
+
     #[inline]
     fn stream_safe(self) -> StreamSafe<I> {
         StreamSafe::new(self)