Speed up surrogate validation in HttpUtility (#110478)

* Speed up surrogate validation in HttpUtility
dotnet · Jan 17, 2025 · 2a28fde · 2a28fde
1 parent 9d03f9b
commit 2a28fde
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 82 deletions.
diff --git a/src/libraries/System.Web.HttpUtility/src/System.Web.HttpUtility.csproj b/src/libraries/System.Web.HttpUtility/src/System.Web.HttpUtility.csproj
@@ -11,13 +11,9 @@
     <Compile Include="System\Web\HttpUtility.cs" />
     <Compile Include="System\Web\Util\HttpEncoder.cs" />
     <Compile Include="System\Web\Util\UriUtil.cs" />
-    <Compile Include="System\Web\Util\Utf16StringValidator.cs" />
-    <Compile Include="$(CommonPath)System\HexConverter.cs"
-             Link="Common\System\HexConverter.cs" />
-    <Compile Include="$(CommonPath)System\Text\ValueStringBuilder.cs"
-             Link="Common\System\Text\ValueStringBuilder.cs" />
-    <Compile Include="$(CommonPath)System\Text\ValueStringBuilder.AppendSpanFormattable.cs"
-             Link="Common\System\Text\ValueStringBuilder.AppendSpanFormattable.cs" />
+    <Compile Include="$(CommonPath)System\HexConverter.cs" Link="Common\System\HexConverter.cs" />
+    <Compile Include="$(CommonPath)System\Text\ValueStringBuilder.cs" Link="Common\System\Text\ValueStringBuilder.cs" />
+    <Compile Include="$(CommonPath)System\Text\ValueStringBuilder.AppendSpanFormattable.cs" Link="Common\System\Text\ValueStringBuilder.AppendSpanFormattable.cs" />
   </ItemGroup>
 
   <ItemGroup>

diff --git a/src/libraries/System.Web.HttpUtility/src/System/Web/Util/HttpEncoder.cs b/src/libraries/System.Web.HttpUtility/src/System/Web/Util/HttpEncoder.cs
@@ -299,7 +299,7 @@ internal static byte[] UrlDecode(ReadOnlySpan<byte> bytes)
                 helper.AddByte(b);
             }
 
-            return Utf16StringValidator.ValidateString(helper.GetString());
+            return helper.GetString();
         }
 
         [return: NotNullIfNotNull(nameof(value))]
@@ -383,7 +383,7 @@ internal static string UrlDecode(ReadOnlySpan<char> value, Encoding encoding)
                 }
             }
 
-            return Utf16StringValidator.ValidateString(helper.GetString());
+            return helper.GetString();
         }
 
         [return: NotNullIfNotNull(nameof(bytes))]
@@ -674,7 +674,35 @@ internal string GetString()
                     FlushBytes();
                 }
 
-                return _charBuffer.Slice(0, _numChars).ToString();
+                Span<char> chars = _charBuffer.Slice(0, _numChars);
+
+                const char HIGH_SURROGATE_START = '\ud800';
+                const char LOW_SURROGATE_END = '\udfff';
+
+                // Replace any invalid surrogate chars.
+                int idxOfFirstSurrogate = chars.IndexOfAnyInRange(HIGH_SURROGATE_START, LOW_SURROGATE_END);
+                for (int i = idxOfFirstSurrogate; (uint)i < (uint)chars.Length; i++)
+                {
+                    if (char.IsHighSurrogate(chars[i]))
+                    {
+                        if ((uint)(i + 1) >= (uint)chars.Length || !char.IsLowSurrogate(chars[i + 1]))
+                        {
+                            // High surrogate not followed by a low surrogate.
+                            chars[i] = (char)Rune.ReplacementChar.Value;
+                        }
+                        else
+                        {
+                            i++;
+                        }
+                    }
+                    else if (char.IsLowSurrogate(chars[i]))
+                    {
+                        // Low surrogate not preceded by a high surrogate.
+                        chars[i] = (char)Rune.ReplacementChar.Value;
+                    }
+                }
+
+                return chars.ToString();
             }
         }
     }

diff --git a/src/libraries/System.Web.HttpUtility/src/System/Web/Util/Utf16StringValidator.cs b/src/libraries/System.Web.HttpUtility/src/System/Web/Util/Utf16StringValidator.cs