Fix and optimize EscapeUnescapeIri

src/libraries/System.Private.Uri/src/System/IriHelper.cs

@@ -109,19 +109,11 @@ internal static unsafe string EscapeUnescapeIri(char* pInput, int start, int end
             ValueStringBuilder dest = new ValueStringBuilder(size);
             byte[]? bytes = null;
 
-            const int percentEncodingLen = 3; // Escaped UTF-8 will take 3 chars: %AB.
-            int bufferRemaining = 0;
-
             int next = start;
             char ch;
-            bool escape = false;
-            bool surrogatePair = false;
 
             for (; next < end; ++next)
             {
-                escape = false;
-                surrogatePair = false;
-
                 if ((ch = pInput[next]) == '%')
                 {
                     if (next + 2 < end)
@@ -226,56 +218,61 @@ internal static unsafe string EscapeUnescapeIri(char* pInput, int start, int end
                 {
                     // unicode
 
-                    char ch2;
+                    bool escape;
+                    bool surrogatePair = false;
+
+                    char ch2 = '\0';
 
                     if ((char.IsHighSurrogate(ch)) && (next + 1 < end))
                     {
                         ch2 = pInput[next + 1];
                         escape = !CheckIriUnicodeRange(ch, ch2, ref surrogatePair, component == UriComponents.Query);
-                        if (!escape)
-                        {
-                            // copy the two chars
-                            dest.Append(pInput[next++]);
-                            dest.Append(pInput[next]);
-                        }
                     }
                     else
                     {
-                        if (CheckIriUnicodeRange(ch, component == UriComponents.Query))
+                        escape = !CheckIriUnicodeRange(ch, component == UriComponents.Query);
+                    }
+
+                    if (escape)
+                    {
+                        Span<byte> encodedBytes = stackalloc byte[4];
+
+                        Rune rune;
+                        if (surrogatePair)
                         {
-                            // copy it
-                            dest.Append(pInput[next]);
+                            rune = new Rune(ch, ch2);
                         }
-                        else
+                        else if (!Rune.TryCreate(ch, out rune))
                         {
-                            // escape it
-                            escape = true;
+                            rune = Rune.ReplacementChar;
                         }
-                    }
-                }
-                else
-                {
-                    // just copy the character
-                    dest.Append(pInput[next]);
-                }
 
-                if (escape)
-                {
-                    const int MaxNumberOfBytesEncoded = 4;
+                        int bytesWritten = rune.EncodeToUtf8(encodedBytes);
+                        encodedBytes = encodedBytes.Slice(0, bytesWritten);
 
-                    byte[] encodedBytes = new byte[MaxNumberOfBytesEncoded];
-                    fixed (byte* pEncodedBytes = &encodedBytes[0])
+                        foreach (byte b in encodedBytes)
+                        {
+                            UriHelper.EscapeAsciiChar(b, ref dest);
+                        }
+                    }
+                    else
                     {
-                        int encodedBytesCount = Encoding.UTF8.GetBytes(pInput + next, surrogatePair ? 2 : 1, pEncodedBytes, MaxNumberOfBytesEncoded);
-                        Debug.Assert(encodedBytesCount <= MaxNumberOfBytesEncoded, "UTF8 encoder should not exceed specified byteCount");
-
-                        bufferRemaining -= encodedBytesCount * percentEncodingLen;
-
-                        for (int count = 0; count < encodedBytesCount; ++count)
+                        dest.Append(ch);
+                        if (surrogatePair)
                         {
-                            UriHelper.EscapeAsciiChar(encodedBytes[count], ref dest);
+                            dest.Append(ch2);
                         }
                     }
+
+                    if (surrogatePair)
+                    {
+                        next++;
+                    }
+                }
+                else
+                {
+                    // just copy the character
+                    dest.Append(pInput[next]);
                 }
             }
 

src/libraries/System.Private.Uri/tests/FunctionalTests/EscapeUnescapeIriTests.cs

@@ -0,0 +1,33 @@
+using System.Collections.Generic;
+using Xunit;
+
+namespace System.PrivateUri.Tests
+{
+    public class EscapeUnescapeIriTests
+    {
+        public static IEnumerable<object[]> ReplacesStandaloneSurrogatesWithReplacementChar()
+        {
+            const string UrlEncodedReplacementChar = "%EF%BF%BD";
+            const string HighSurrogate = "\ud83f";
+            const string LowSurrogate = "\udffe";
+
+            yield return new object[] { "a", "a" };
+            yield return new object[] { HighSurrogate + LowSurrogate, "%F0%9F%BF%BE" };
+            yield return new object[] { HighSurrogate, UrlEncodedReplacementChar };
+            yield return new object[] { LowSurrogate, UrlEncodedReplacementChar };
+            yield return new object[] { LowSurrogate + HighSurrogate, UrlEncodedReplacementChar + UrlEncodedReplacementChar };
+            yield return new object[] { LowSurrogate + LowSurrogate, UrlEncodedReplacementChar + UrlEncodedReplacementChar };
+            yield return new object[] { HighSurrogate + HighSurrogate, UrlEncodedReplacementChar + UrlEncodedReplacementChar };
+        }
+
+        [Theory]
+        [MemberData(nameof(ReplacesStandaloneSurrogatesWithReplacementChar))]
+        public static void ReplacesStandaloneSurrogatesWithReplacementChar(string input, string expected)
+        {
+            const string Prefix = "scheme:";
+            Uri uri = new Uri(Prefix + input);
+            string actual = uri.AbsoluteUri.Substring(Prefix.Length);
+            Assert.Equal(expected, actual);
+        }
+    }
+}

src/libraries/System.Private.Uri/tests/FunctionalTests/System.Private.Uri.Functional.Tests.csproj

@@ -4,6 +4,7 @@
   </PropertyGroup>
   <ItemGroup>
     <Compile Include="AppxUriValue.cs" />
+    <Compile Include="EscapeUnescapeIriTests.cs" />
     <Compile Include="IdnCheckHostNameTest.cs" />
     <Compile Include="IdnDnsSafeHostTest.cs" />
     <Compile Include="IdnHostNameValidationTest.cs" />