Reworked the rune operations to not spam char[1 or 2] allocations and leverage simd for BMP-only strings

Author: pl752

src/FirebirdSql.Data.FirebirdClient/Client/Managed/Version10/GdsStatement.cs

@@ -1533,16 +1533,7 @@ protected object ReadRawValue(IXdrReader xdr, DbField field)
 				else
 				{
 					var s = xdr.ReadString(innerCharset, field.Length);
-					var runes = s.EnumerateRunesToChars().ToList();
-					if ((field.Length % field.Charset.BytesPerCharacter) == 0 &&
-						runes.Count > field.CharCount)
-					{
-						return new string([.. runes.Take(field.CharCount).SelectMany(x => x)]);
-					}
-					else
-					{
-						return s;
-					}
+					return TruncateStringByRuneCount(s, field);
 				}
 
 			case DbDataType.VarChar:
@@ -1631,16 +1622,7 @@ protected async ValueTask<object> ReadRawValueAsync(IXdrReader xdr, DbField fiel
 				else
 				{
 					var s = await xdr.ReadStringAsync(innerCharset, field.Length, cancellationToken).ConfigureAwait(false);
-					var runes = s.EnumerateRunesToChars().ToList();
-					if ((field.Length % field.Charset.BytesPerCharacter) == 0 &&
-						runes.Count > field.CharCount)
-					{
-						return new string([.. runes.Take(field.CharCount).SelectMany(x => x)]);
-					}
-					else
-					{
-						return s;
-					}
+					return TruncateStringByRuneCount(s, field);
 				}
 
 			case DbDataType.VarChar:
@@ -1797,6 +1779,22 @@ protected virtual async ValueTask<DbValue[]> ReadRowAsync(CancellationToken canc
 		return row;
 	}
 
+	private static string TruncateStringByRuneCount(string s, DbField field)
+	{
+		if ((field.Length % field.Charset.BytesPerCharacter) != 0)
+		{
+			return s;
+		}
+
+		var runeCount = s.CountRunes();
+		if (runeCount <= field.CharCount)
+		{
+			return s;
+		}
+
+		return new string(s.TruncateStringToRuneCount(field.CharCount));
+	}
+
 	#endregion
 
 	#region Protected Internal Methods

src/FirebirdSql.Data.FirebirdClient/Common/DbField.cs

@@ -325,12 +325,12 @@ public void SetValue(byte[] buffer)
 						else
 						{
 							var s = Charset.GetString(buffer, 0, buffer.Length);
-
-							var runes = s.EnumerateRunesToChars().ToList();
-							if ((Length % Charset.BytesPerCharacter) == 0 &&
-								runes.Count > CharCount)
-							{
-								s = new string([.. runes.Take(CharCount).SelectMany(x => x)]);
+							if((Length % Charset.BytesPerCharacter) == 0)
+							{ 
+								var runes = s.CountRunes();
+								if(runes > CharCount) {
+									s = new string(s.TruncateStringToRuneCount(CharCount));
+								}
 							}
 
 							DbValue.SetValue(s);

src/FirebirdSql.Data.FirebirdClient/Common/DbValue.cs

@@ -424,7 +424,7 @@ public byte[] GetBytes()
 					else
 					{
 						var svalue = GetString();
-						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.EnumerateRunesToChars().Count() > Field.CharCount)
+						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.CountRunes() > Field.CharCount)
 						{
 							throw IscException.ForErrorCodes(new[] { IscCodes.isc_arith_except, IscCodes.isc_string_truncation });
 						}
@@ -460,7 +460,7 @@ public byte[] GetBytes()
 					else
 					{
 						var svalue = GetString();
-						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.EnumerateRunesToChars().Count() > Field.CharCount)
+						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.CountRunes() > Field.CharCount)
 						{
 							throw IscException.ForErrorCodes(new[] { IscCodes.isc_arith_except, IscCodes.isc_string_truncation });
 						}
@@ -639,7 +639,7 @@ public async ValueTask<byte[]> GetBytesAsync(CancellationToken cancellationToken
 					else
 					{
 						var svalue = await GetStringAsync(cancellationToken).ConfigureAwait(false);
-						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.EnumerateRunesToChars().Count() > Field.CharCount)
+						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.CountRunes() > Field.CharCount)
 						{
 							throw IscException.ForErrorCodes(new[] { IscCodes.isc_arith_except, IscCodes.isc_string_truncation });
 						}
@@ -675,7 +675,7 @@ public async ValueTask<byte[]> GetBytesAsync(CancellationToken cancellationToken
 					else
 					{
 						var svalue = await GetStringAsync(cancellationToken).ConfigureAwait(false);
-						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.EnumerateRunesToChars().Count() > Field.CharCount)
+						if ((Field.Length % Field.Charset.BytesPerCharacter) == 0 && svalue.CountRunes() > Field.CharCount)
 						{
 							throw IscException.ForErrorCodes(new[] { IscCodes.isc_arith_except, IscCodes.isc_string_truncation });
 						}

src/FirebirdSql.Data.FirebirdClient/Common/Extensions.cs

@@ -102,4 +102,60 @@ public static Encoding GetANSIEncoding()
 			}
 		}
 	}
+
+	public static int CountRunes(this ReadOnlySpan<char> text)
+	{
+		var length = text.Length;
+		if(length == 0)
+			return 0;
+
+		var i = text.IndexOfAnyInRange('\uD800', '\uDBFF');
+		if(i < 0)
+			return length;
+
+		var count = i;
+		while(i < length)
+		{
+			if(char.IsHighSurrogate(text[i]) && i + 1 < length && char.IsLowSurrogate(text[i + 1]))
+			{
+				i += 2;
+			}
+			else
+			{
+				i++;
+			}
+			count++;
+		}
+		return count;
+	}
+
+	public static ReadOnlySpan<char> TruncateStringToRuneCount(this ReadOnlySpan<char> text, int maxRuneCount)
+	{
+		if(maxRuneCount <= 0 || text.IsEmpty)
+			return ReadOnlySpan<char>.Empty;
+
+		var length = text.Length;
+		if(maxRuneCount >= length)
+			return text;
+
+		var prefix = text[..maxRuneCount];
+		var i = prefix.IndexOfAnyInRange('\uD800', '\uDBFF');
+		if(i < 0)
+			return prefix;
+
+		var remaining = maxRuneCount - i;
+		while(i < length && remaining > 0)
+		{
+			if(char.IsHighSurrogate(text[i]) && i + 1 < length && char.IsLowSurrogate(text[i + 1]))
+			{
+				i += 2;
+			}
+			else
+			{
+				i++;
+			}
+			remaining--;
+		}
+		return text[..i];
+	}
 }