From 2e93d83149fe99f75820b4fe6e78e09d145f726a Mon Sep 17 00:00:00 2001
From: ringabout <43030857+ringabout@users.noreply.github.com>
Date: Tue, 8 Sep 2026 00:17:04 +0800
Subject: [PATCH] fixes #26175; Utf16Char incorrectly declared as signed
 (#26181)

fixes #26175
---
 lib/std/widestrs.nim          | 10 +++++-----
 tests/stdlib/twchartoutf8.nim | 17 +++++++++++++++--
 2 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/lib/std/widestrs.nim b/lib/std/widestrs.nim
index af9be00108..ca0407a8f4 100644
--- a/lib/std/widestrs.nim
+++ b/lib/std/widestrs.nim
@@ -13,7 +13,7 @@
 #  {.error: "You must not import this module explicitly".}
 
 type
-  Utf16Char* = distinct int16
+  Utf16Char* = distinct uint16
 
 when not (defined(cpu16) or defined(cpu8)):
   when defined(nimv2):
@@ -78,10 +78,10 @@ when not (defined(cpu16) or defined(cpu8)):
     ## returns the length of a widestring. This traverses the whole string to
     ## find the binary zero end marker!
     result = 0
-    while int16(w[result]) != 0'i16: inc result
+    while uint16(w[result]) != 0'u16: inc result
 
   const
-    UNI_REPLACEMENT_CHAR = Utf16Char(0xFFFD'i16)
+    UNI_REPLACEMENT_CHAR = Utf16Char(0xFFFD'u16)
     UNI_MAX_BMP = 0x0000FFFF
     UNI_MAX_UTF16 = 0x0010FFFF
     # UNI_MAX_UTF32 = 0x7FFFFFFF
@@ -188,12 +188,12 @@ when not (defined(cpu16) or defined(cpu8)):
   iterator decodeUtf16(w: WideCString; replacement: int): int =
     ## Looks for a terminating NUL for length
     var i = 0
-    while w[i].int16 != 0'i16:
+    while w[i].uint16 != 0'u16:
       var ch = ord(w[i])
       inc i
       if ch >= UNI_SUR_HIGH_START and ch <= UNI_SUR_HIGH_END:
         # If the 16 bits following the high surrogate are NOT in the source...
-        if w[i].int16 == 0'i16:
+        if w[i].uint16 == 0'u16:
           ch = replacement #invalid UTF-16
         else:
           let ch2 = ord(w[i])
diff --git a/tests/stdlib/twchartoutf8.nim b/tests/stdlib/twchartoutf8.nim
index e437177ac3..0a5b8d0af5 100644
--- a/tests/stdlib/twchartoutf8.nim
+++ b/tests/stdlib/twchartoutf8.nim
@@ -3,7 +3,21 @@ discard """
   output: '''OK'''
 """
 
-import std/[syncio, assertions]
+import std/[syncio, assertions, widestrs]
+
+const
+  face = [Utf16Char(0xD83E), Utf16Char(0xDD2A)]
+  legacyHighSurrogate = -10178'i16
+
+static:
+  doAssert sizeof(Utf16Char) == sizeof(int16)
+  doAssert alignof(Utf16Char) == alignof(int16)
+  doAssert uint16(cast[Utf16Char](legacyHighSurrogate)) == uint16(face[0])
+
+proc testUtf16CodeUnits(chars: openArray[Utf16Char]) =
+  doAssert $chars == "🤪"
+
+testUtf16CodeUnits(face)
 
 #assume WideCharToMultiByte always produce correct result
 #windows only
@@ -11,7 +25,6 @@ import std/[syncio, assertions]
 when not defined(windows):
   echo "OK"
 else:
-  import std/widestrs
   {.push gcsafe.}
 
   const CP_UTF8 = 65001'i32