Fixed BOM detection

BUTR · May 29, 2024 · 24f8a11 · 24f8a11
1 parent d5fef0d
commit 24f8a11
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 13 deletions.
diff --git a/src/Bannerlord.LauncherManager.Native/ReaderUtils.cs b/src/Bannerlord.LauncherManager.Native/ReaderUtils.cs
@@ -7,20 +7,29 @@ namespace Bannerlord.LauncherManager.Native;
 
 public static class ReaderUtils2
 {
-    private static readonly byte[] BOMMarkUtf8 = Encoding.UTF8.GetPreamble();
+    private static readonly byte[] BOMMarkUtf8 = [0xEF, 0xBB, 0xBF];
+    private static readonly byte[] BOMMarkUtf16BE = [0xFE, 0xFF];
+    private static readonly byte[] BOMMarkUtf16LE = [0xFF, 0xFE];
 
     public static XmlDocument Read(ReadOnlySpan<char> xml)
     {
         var doc = new XmlDocument();
         doc.LoadXml(new string(RemoveBOM2(xml)));
         return doc;
     }
-    
+
     private static ReadOnlySpan<char> RemoveBOM2(ReadOnlySpan<char> xml)
     {
-        var bom = MemoryMarshal.Cast<byte, char>(BOMMarkUtf8.AsSpan());
-        if (xml.StartsWith(bom))
-            xml = xml.Slice(bom.Length);
+        var utf8BOM = MemoryMarshal.Cast<byte, char>(BOMMarkUtf8.AsSpan());
+        var utf16BEBOM = MemoryMarshal.Cast<byte, char>(BOMMarkUtf16BE.AsSpan());
+        var utf16LEBOM = MemoryMarshal.Cast<byte, char>(BOMMarkUtf16LE.AsSpan());
+
+        var bomOffset = xml.StartsWith(utf8BOM) ? utf8BOM.Length :
+            xml.StartsWith(utf16BEBOM) ? utf16BEBOM.Length :
+            xml.StartsWith(utf16LEBOM) ? utf16LEBOM.Length : 0;
+
+        xml = xml.Slice(bomOffset > 0 ? bomOffset : 0);
+
         return xml;
     }
 }
diff --git a/src/Bannerlord.LauncherManager/Utils/ReaderUtils.cs b/src/Bannerlord.LauncherManager/Utils/ReaderUtils.cs
@@ -1,29 +1,38 @@
 using System;
 using System.IO;
-using System.Text;
 using System.Xml;
 
 namespace Bannerlord.LauncherManager.Utils;
 
 public static class ReaderUtils
 {
-    private static readonly byte[] BOMMarkUtf8 = Encoding.UTF8.GetPreamble();
+    private static readonly byte[] BOMMarkUtf8 = [0xEF, 0xBB, 0xBF];
+    private static readonly byte[] BOMMarkUtf16BE = [0xFE, 0xFF];
+    private static readonly byte[] BOMMarkUtf16LE = [0xFF, 0xFE];
 
     public static XmlDocument Read(byte[] data)
     {
         using var ms = new MemoryStream(data);
         using var sr = new StreamReader(ms);
 
-        Span<byte> bom = BOMMarkUtf8.AsSpan();
+        var utf8BOM = BOMMarkUtf8.AsSpan();
+        var utf16BEBOM = BOMMarkUtf16BE.AsSpan();
+        var utf16LEBOM = BOMMarkUtf16LE.AsSpan();
+
 #if NETSTANDARD2_1_OR_GREATER
-        Span<byte> preamblePreallocated = stackalloc byte[BOMMarkUtf8.Length];
+        Span<byte> preamblePreallocated = stackalloc byte[4];
         ms.Read(preamblePreallocated);
 #else
-        var preamblePreallocated = new byte[BOMMarkUtf8.Length];
-        ms.Read(preamblePreallocated, 0, BOMMarkUtf8.Length);
+        var preamblePreallocatedArray = new byte[4];
+        ms.Read(preamblePreallocatedArray, 0, 4);
+        var preamblePreallocated = preamblePreallocatedArray.AsSpan();
 #endif
-
-        ms.Seek(bom.SequenceEqual(preamblePreallocated) ? BOMMarkUtf8.Length : 0, SeekOrigin.Begin);
+
+        var bomOffset = preamblePreallocated.StartsWith(utf8BOM) ? BOMMarkUtf8.Length :
+            preamblePreallocated.StartsWith(utf16BEBOM) ? BOMMarkUtf16BE.Length :
+            preamblePreallocated.StartsWith(utf16LEBOM) ? BOMMarkUtf16LE.Length : 0;
+
+        ms.Seek(bomOffset > 0 ? bomOffset : 0, SeekOrigin.Begin);
         var doc = new XmlDocument();
         doc.Load(new XmlTextReader(sr));
         return doc;