From ed295a5654d3bc3e49382fa5d07e484155632cd9 Mon Sep 17 00:00:00 2001
From: Bela VanderVoort <twobitbela@gmail.com>
Date: Fri, 3 May 2024 10:47:54 -0500
Subject: [PATCH] Consider wide unicode characters to be size of 2 (#1237)

closes #260
---
 Src/CSharpier.Tests/CodeFormatterTests.cs     | 24 ++++++++++++++
 .../Utilities/CharacterSizeCalculator.cs      | 33 +++++++++++++++++++
 Src/CSharpier/Utilities/StringExtensions.cs   | 11 ++-----
 3 files changed, 59 insertions(+), 9 deletions(-)
 create mode 100644 Src/CSharpier/Utilities/CharacterSizeCalculator.cs

diff --git a/Src/CSharpier.Tests/CodeFormatterTests.cs b/Src/CSharpier.Tests/CodeFormatterTests.cs
index 47b18c657..da686609c 100644
--- a/Src/CSharpier.Tests/CodeFormatterTests.cs
+++ b/Src/CSharpier.Tests/CodeFormatterTests.cs
@@ -4,6 +4,8 @@
 
 namespace CSharpier.Tests;
 
+using CSharpier.Utilities;
+
 // TODO xml move these around
 [TestFixture]
 [Parallelizable(ParallelScope.All)]
@@ -62,6 +64,28 @@ public void Format_Should_Use_Width()
         result.Code.Should().Be("var someVariable =\n    someValue;\n");
     }
 
+    [Test]
+    public void Format_Should_Measure_Regular_Characters()
+    {
+        var code = """
+            var x = "123456";
+            """;
+        var result = CodeFormatter.Format(code, new CodeFormatterOptions { Width = 20 });
+
+        result.Code.Should().Be("var x = \"123456\";\n");
+    }
+
+    [Test]
+    public void Format_Should_Measure_Wide_Characters()
+    {
+        var code = """
+            var x = "가가가가가가";
+            """;
+        var result = CodeFormatter.Format(code, new CodeFormatterOptions { Width = 20 });
+
+        result.Code.Should().Be("var x =\n    \"가가가가가가\";\n");
+    }
+
     [Test]
     public void Format_Should_Use_IndentStyle()
     {
diff --git a/Src/CSharpier/Utilities/CharacterSizeCalculator.cs b/Src/CSharpier/Utilities/CharacterSizeCalculator.cs
new file mode 100644
index 000000000..b1f247d6c
--- /dev/null
+++ b/Src/CSharpier/Utilities/CharacterSizeCalculator.cs
@@ -0,0 +1,33 @@
+/* Copyright (c) Microsoft Corporation.
+ * Under MIT License
+ * From https://github.com/PowerShell/PowerShell/tree/master
+ */
+
+namespace CSharpier.Utilities;
+
+internal static class CharacterSizeCalculator
+{
+    // csharpier-ignore
+    public static int CalculateWidth(char c)
+    {
+        // The following is based on http://www.cl.cam.ac.uk/~mgk25/c/wcwidth.c
+        // which is derived from https://www.unicode.org/Public/UCD/latest/ucd/EastAsianWidth.txt
+        var isWide = c >= 0x1100 &&
+            (c <= 0x115f || /* Hangul Jamo init. consonants */
+             c == 0x2329 || c == 0x232a ||
+             ((uint)(c - 0x2e80) <= (0xa4cf - 0x2e80) &&
+              c != 0x303f) || /* CJK ... Yi */
+             ((uint)(c - 0xac00) <= (0xd7a3 - 0xac00)) || /* Hangul Syllables */
+             ((uint)(c - 0xf900) <= (0xfaff - 0xf900)) || /* CJK Compatibility Ideographs */
+             ((uint)(c - 0xfe10) <= (0xfe19 - 0xfe10)) || /* Vertical forms */
+             ((uint)(c - 0xfe30) <= (0xfe6f - 0xfe30)) || /* CJK Compatibility Forms */
+             ((uint)(c - 0xff00) <= (0xff60 - 0xff00)) || /* Fullwidth Forms */
+             ((uint)(c - 0xffe0) <= (0xffe6 - 0xffe0)));
+
+        // We can ignore these ranges because .Net strings use surrogate pairs
+        // for this range and we do not handle surrogate pairs.
+        // (c >= 0x20000 && c <= 0x2fffd) ||
+        // (c >= 0x30000 && c <= 0x3fffd)
+        return isWide ? 2 : 1;
+    }
+}
diff --git a/Src/CSharpier/Utilities/StringExtensions.cs b/Src/CSharpier/Utilities/StringExtensions.cs
index c4aaaf86f..8763958d9 100644
--- a/Src/CSharpier/Utilities/StringExtensions.cs
+++ b/Src/CSharpier/Utilities/StringExtensions.cs
@@ -5,13 +5,6 @@ namespace CSharpier.Utilities;
 
 internal static class StringExtensions
 {
-    public static string CalculateHash(this string value)
-    {
-        using var hasher = MD5.Create();
-        var hashedBytes = hasher.ComputeHash(Encoding.UTF8.GetBytes(value));
-        return BitConverter.ToString(hashedBytes).Replace("-", string.Empty).ToLower();
-    }
-
     public static bool EqualsIgnoreCase(this string value, string otherValue)
     {
         return string.Compare(value, otherValue, StringComparison.OrdinalIgnoreCase) == 0;
@@ -37,10 +30,10 @@ public static bool IsBlank(this string? value)
         return value == null || string.IsNullOrEmpty(value.Trim());
     }
 
-    // this will eventually deal with the visual width not being the same as the code width https://github.com/belav/csharpier/issues/260
+    // some unicode characters should be considered size of 2 when calculating how big this string will be when printed
     public static int GetPrintedWidth(this string value)
     {
-        return value.Length;
+        return value.Sum(CharacterSizeCalculator.CalculateWidth);
     }
 
     public static int CalculateCurrentLeadingIndentation(this string line, int indentSize)