From 104b51b9122f0cb142a07bd9a1dc2106fe817519 Mon Sep 17 00:00:00 2001
From: matcool <26722564+matcool@users.noreply.github.com>
Date: Thu, 6 Oct 2022 14:09:51 -0300
Subject: [PATCH] LibUnicode: Fix Hangul syllable composition for specific
 cases

This fixes `combine_hangul_code_points` which would try to combine
a LVT syllable with a trailing consonant, resulting in a wrong
character.

Also added a test for this specific case.
---
 Tests/LibUnicode/TestUnicodeNormalization.cpp | 1 +
 Userland/Libraries/LibUnicode/Normalize.cpp   | 3 ++-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/Tests/LibUnicode/TestUnicodeNormalization.cpp b/Tests/LibUnicode/TestUnicodeNormalization.cpp
index 8157b6d9660..daef39d7652 100644
--- a/Tests/LibUnicode/TestUnicodeNormalization.cpp
+++ b/Tests/LibUnicode/TestUnicodeNormalization.cpp
@@ -63,6 +63,7 @@ TEST_CASE(normalize_nfc)
 
     EXPECT_EQ(normalize("\u1103\u1161\u11B0"sv, NormalizationForm::NFC), "닭"sv);
     EXPECT_EQ(normalize("\u1100\uAC00\u11A8"sv, NormalizationForm::NFC), "\u1100\uAC01"sv);
+    EXPECT_EQ(normalize("\u1103\u1161\u11B0\u11B0"sv, NormalizationForm::NFC), "닭\u11B0");
 }
 
 TEST_CASE(normalize_nfkd)
diff --git a/Userland/Libraries/LibUnicode/Normalize.cpp b/Userland/Libraries/LibUnicode/Normalize.cpp
index 8f5a59f7c3c..bc1ab70221e 100644
--- a/Userland/Libraries/LibUnicode/Normalize.cpp
+++ b/Userland/Libraries/LibUnicode/Normalize.cpp
@@ -111,7 +111,8 @@ static u32 combine_hangul_code_points(u32 a, u32 b)
         auto const leading_vowel_index = leading_index * HANGUL_BLOCK_COUNT + vowel_index * HANGUL_TRAILING_COUNT;
         return HANGUL_SYLLABLE_BASE + leading_vowel_index;
     }
-    if (is_hangul_code_point(a) && is_hangul_trailing(b)) {
+    // LV characters are the first in each "T block", so use this check to avoid combining LVT with T.
+    if (is_hangul_code_point(a) && (a - HANGUL_SYLLABLE_BASE) % HANGUL_TRAILING_COUNT == 0 && is_hangul_trailing(b)) {
         return a + b - HANGUL_TRAILING_BASE;
     }
     return 0;