Auto merge of #124773 - Marcondiro:master, r=joboet

bors · bors · commit 238c1e798d37 · 2024-05-09T16:59:56.000Z
fix #124714 str.to_lowercase sigma handling Hello, This PR fixes issue #124714 about 'Σ' handling in `str.to_lowercase()`. The fix consists in considering the full original string during 'Σ' handling instead of considering just the substring left after the optimized ascii handling. A new test is added to avoid regression. Thanks!
diff --git a/library/alloc/src/str.rs b/library/alloc/src/str.rs
@@ -375,14 +375,16 @@ impl str {
         // Safety: We have written only valid ASCII to our vec
         let mut s = unsafe { String::from_utf8_unchecked(out) };
 
-        for (i, c) in rest[..].char_indices() {
+        for (i, c) in rest.char_indices() {
             if c == 'Σ' {
                 // Σ maps to σ, except at the end of a word where it maps to ς.
                 // This is the only conditional (contextual) but language-independent mapping
                 // in `SpecialCasing.txt`,
                 // so hard-code it rather than have a generic "condition" mechanism.
                 // See https://github.com/rust-lang/rust/issues/26035
-                map_uppercase_sigma(rest, i, &mut s)
+                let out_len = self.len() - rest.len();
+                let sigma_lowercase = map_uppercase_sigma(&self, i + out_len);
+                s.push(sigma_lowercase);
             } else {
                 match conversions::to_lower(c) {
                     [a, '\0', _] => s.push(a),
@@ -400,13 +402,13 @@ impl str {
         }
         return s;
 
-        fn map_uppercase_sigma(from: &str, i: usize, to: &mut String) {
+        fn map_uppercase_sigma(from: &str, i: usize) -> char {
             // See https://www.unicode.org/versions/Unicode7.0.0/ch03.pdf#G33992
             // for the definition of `Final_Sigma`.
             debug_assert!('Σ'.len_utf8() == 2);
             let is_word_final = case_ignorable_then_cased(from[..i].chars().rev())
                 && !case_ignorable_then_cased(from[i + 2..].chars());
-            to.push_str(if is_word_final { "ς" } else { "σ" });
+            if is_word_final { 'ς' } else { 'σ' }
         }
 
         fn case_ignorable_then_cased<I: Iterator<Item = char>>(iter: I) -> bool {
diff --git a/library/alloc/tests/str.rs b/library/alloc/tests/str.rs
@@ -1848,6 +1848,9 @@ fn to_lowercase() {
     assert_eq!("ΑΣ'Α".to_lowercase(), "ασ'α");
     assert_eq!("ΑΣ''Α".to_lowercase(), "ασ''α");
 
+    // https://github.com/rust-lang/rust/issues/124714
+    assert_eq!("abcdefghijklmnopΣ".to_lowercase(), "abcdefghijklmnopς");
+
     // a really long string that has it's lowercase form
     // even longer. this tests that implementations don't assume
     // an incorrect upper bound on allocations
