Merge pull request ClickHouse#88940 from ClickHouse/backport/25.8/88914

Backport ClickHouse#88914 to 25.8: Fix bug in `reverseUTF8`

Author: alexey-milovidov

src/Functions/reverseUTF8.cpp

@@ -46,7 +46,7 @@ struct ReverseUTF8Impl
             ColumnString::Offset j = prev_offset;
             while (j < offsets[i])
             {
-                if (data[j] < 0xBF)
+                if (data[j] < 0xC0)
                 {
                     res_data[offsets[i] + prev_offset - 1 - j] = data[j];
                     j += 1;
@@ -63,8 +63,8 @@ struct ReverseUTF8Impl
                 }
                 else
                 {
-                    res_data[offsets[i] + prev_offset - 1 - j] = data[j];
-                    j += 1;
+                    memcpy(&res_data[offsets[i] + prev_offset - 1 - j - 3], &data[j], 4);
+                    j += 4;
                 }
             }
 

tests/queries/0_stateless/03699_reverse_utf8.reference

@@ -0,0 +1,4 @@
+тівирп
+🌈🇧🇬
+🌈
+ेत्समन

tests/queries/0_stateless/03699_reverse_utf8.sql

@@ -0,0 +1,6 @@
+-- The function reverses the sequence of UTF-8 code points (that is different from bytes or full characters):
+
+SELECT reverseUTF8('привіт');
+SELECT reverseUTF8('🇬🇧🌈');
+SELECT reverseUTF8('🌈');
+SELECT reverseUTF8('नमस्ते');