fix: Unicode Text length Calculation (#16219)

* fix: Unicode Text length Calculation

* Change files

* yarn format fix

Author: vineethkuttan

change/react-native-windows-c3de21ab-adb1-4f1d-a290-4f0c29f087ef.json

@@ -0,0 +1,7 @@
+{
+  "type": "prerelease",
+  "comment": "fix: Unicode Text length Calculation",
+  "packageName": "react-native-windows",
+  "email": "66076509+vineethkuttan@users.noreply.github.com",
+  "dependentChangeType": "patch"
+}

vnext/Common/unicode.cpp

@@ -93,6 +93,42 @@ std::wstring Utf8ToUtf16(const std::string &utf8) {
   return Utf8ToUtf16(utf8.c_str(), utf8.length());
 }
 
+size_t Utf8ToUtf16Length(const char *utf8, size_t utf8Len) {
+  if (utf8Len == 0) {
+    return 0;
+  }
+
+  if (utf8Len > static_cast<size_t>((std::numeric_limits<int>::max)())) {
+    throw std::overflow_error("Length of input string to Utf8ToUtf16Length() must fit into an int.");
+  }
+
+  const int utf8Length = static_cast<int>(utf8Len);
+
+  constexpr DWORD flags = 0;
+
+  const int utf16Length = ::MultiByteToWideChar(
+      CP_UTF8, // Source string is in UTF-8.
+      flags, // Conversion flags.
+      utf8, // Source UTF-8 string pointer.
+      utf8Length, // Length of the source UTF-8 string, in chars.
+      nullptr, // Do not convert, just request the size.
+      0 // Request size of destination buffer, in wchar_ts.
+  );
+
+  if (utf16Length == 0) {
+    throw UnicodeConversionException(
+        "Cannot get result string length when converting from UTF-8 to UTF-16 "
+        "(MultiByteToWideChar failed).",
+        GetLastError());
+  }
+
+  return static_cast<size_t>(utf16Length);
+}
+
+size_t Utf8ToUtf16Length(const std::string &utf8) {
+  return Utf8ToUtf16Length(utf8.c_str(), utf8.length());
+}
+
 #if _HAS_CXX17
 std::wstring Utf8ToUtf16(const std::string_view &utf8) {
   return Utf8ToUtf16(utf8.data(), utf8.length());

vnext/Common/unicode.h

@@ -55,6 +55,14 @@ class UnicodeConversionException : public std::runtime_error {
 /* (4) */ std::wstring Utf8ToUtf16(const std::string_view &utf8);
 #endif
 
+// The following functions return the length of the UTF-16 string that would
+// result from converting the input UTF-8 string, without actually performing
+// the conversion or allocating a temporary std::wstring. This is useful in
+// hot paths where only the length is needed (e.g. DirectWrite text ranges).
+//
+size_t Utf8ToUtf16Length(const char *utf8, size_t utf8Len);
+size_t Utf8ToUtf16Length(const std::string &utf8);
+
 // The following functions convert UTF-16BE strings to UTF-8 strings. Their
 // behaviors mirror those of the above Utf8ToUtf16 functions.
 //

vnext/Microsoft.ReactNative/Fabric/Composition/ParagraphComponentView.cpp

@@ -153,11 +153,12 @@ facebook::react::SharedViewEventEmitter ParagraphComponentView::eventEmitterAtPo
       uint32_t textPosition = metrics.textPosition;
 
       for (auto fragment : m_attributedStringBox.getValue().getFragments()) {
-        if (textPosition < fragment.string.length()) {
+        uint32_t utf16Length = static_cast<uint32_t>(::Microsoft::Common::Unicode::Utf8ToUtf16Length(fragment.string));
+        if (textPosition < utf16Length) {
           return std::static_pointer_cast<const facebook::react::ViewEventEmitter>(
               fragment.parentShadowView.eventEmitter);
         }
-        textPosition -= static_cast<uint32_t>(fragment.string.length());
+        textPosition -= utf16Length;
       }
     }
   }
@@ -210,10 +211,11 @@ bool ParagraphComponentView::IsTextSelectableAtPoint(facebook::react::Point pt)
 
       // Finds which fragment contains this text position
       for (auto fragment : m_attributedStringBox.getValue().getFragments()) {
-        if (textPosition < fragment.string.length()) {
+        uint32_t utf16Length = static_cast<uint32_t>(::Microsoft::Common::Unicode::Utf8ToUtf16Length(fragment.string));
+        if (textPosition < utf16Length) {
           return true;
         }
-        textPosition -= static_cast<uint32_t>(fragment.string.length());
+        textPosition -= utf16Length;
       }
     }
   }

vnext/Microsoft.ReactNative/Fabric/Composition/TextDrawing.cpp

@@ -66,7 +66,7 @@ void RenderText(
   unsigned int position = 0;
   unsigned int length = 0;
   for (auto fragment : attributedString.getFragments()) {
-    length = static_cast<UINT32>(fragment.string.length());
+    length = static_cast<UINT32>(::Microsoft::Common::Unicode::Utf8ToUtf16Length(fragment.string));
     DWRITE_TEXT_RANGE range = {position, length};
     if (fragment.textAttributes.foregroundColor &&
             (fragment.textAttributes.foregroundColor != textAttributes.foregroundColor) ||

vnext/Microsoft.ReactNative/Fabric/platform/react/renderer/textlayoutmanager/WindowsTextLayoutManager.cpp

@@ -262,7 +262,7 @@ void WindowsTextLayoutManager::GetTextLayout(
       attachments.push_back(attachment);
       position += 1;
     } else {
-      unsigned int length = static_cast<UINT32>(fragment.string.length());
+      unsigned int length = static_cast<UINT32>(Microsoft::Common::Unicode::Utf8ToUtf16Length(fragment.string));
       DWRITE_TEXT_RANGE range = {position, length};
       TextAttributes attributes = fragment.textAttributes;
       DWRITE_FONT_STYLE fragmentStyle = DWRITE_FONT_STYLE_NORMAL;