neovim

commit 610e9666af0ef16f3677b23a5c341c8817c89965
parent 69553f7bf55c060733553d96a068c1104c885bce
Author: bfredl <bjorn.linse@gmail.com>
Date:   Sun, 29 Sep 2024 11:03:36 +0200

Merge pull request #30401 from bfredl/casefold2

refactor(multibyte): neo-casefolding without allocation
Diffstat:
M
src/nvim/mbyte.c
 | 
17
+++--------------
M
test/unit/mbyte_spec.lua
 | 
6
++++++

2 files changed, 9 insertions(+), 14 deletions(-)
diff --git a/src/nvim/mbyte.c b/src/nvim/mbyte.c
@@ -1379,22 +1379,11 @@ int utf_fold(int a)
     return a;
   }
 
-  utf8proc_uint8_t input_str[16] = { 0 };
-  if (utf8proc_encode_char(a, input_str) <= 0) {
-    return a;
-  }
-
-  utf8proc_uint8_t *fold_str_utf;
-  if (utf8proc_map((utf8proc_uint8_t *)input_str, 0, &fold_str_utf,
-                   UTF8PROC_NULLTERM | UTF8PROC_CASEFOLD) < 0) {
-    return a;
-  }
-
-  int fold_codepoint_utf = utf_ptr2char((char *)fold_str_utf);
+  utf8proc_int32_t result[1];
 
-  xfree(fold_str_utf);
+  utf8proc_ssize_t res = utf8proc_decompose_char(a, result, 1, UTF8PROC_CASEFOLD, NULL);
 
-  return fold_codepoint_utf;
+  return (res == 1) ? result[0] : a;
 }
 
 // Vim's own character class functions.  These exist because many library
diff --git a/test/unit/mbyte_spec.lua b/test/unit/mbyte_spec.lua
@@ -351,6 +351,12 @@ describe('mbyte', function()
   describe('utf_fold', function()
     itp('does not crash with surrogates #30527', function()
       eq(0xDDFB, lib.utf_fold(0xDDFB))
+      eq(0xd800, lib.utf_fold(0xd800)) -- high surrogate, invalid as a character
+    end)
+
+    itp("doesn't crash on invalid codepoints", function()
+      eq(9000000, lib.utf_fold(9000000))
+      eq(0, lib.utf_fold(0))
     end)
   end)
 end)