Add string optimizations

QRCoder/QRCodeGenerator.cs

@@ -745,14 +745,22 @@ bool IsUtf8()
             }
         }
 
+        private static readonly Encoding _iso88591ExceptionFallback = Encoding.GetEncoding(28591, new EncoderExceptionFallback(), new DecoderExceptionFallback()); // ISO-8859-1
         /// <summary>
         /// Checks if the given string can be accurately represented and retrieved in ISO-8859-1 encoding.
         /// </summary>
         private static bool IsValidISO(string input)
         {
-            var bytes = Encoding.GetEncoding("ISO-8859-1").GetBytes(input);
-            var result = Encoding.GetEncoding("ISO-8859-1").GetString(bytes);
-            return String.Equals(input, result);
+            // No heap allocations if the string is ISO-8859-1
+            try
+            {
+                _ = _iso88591ExceptionFallback.GetByteCount(input);
+                return true;
+            }
+            catch (EncoderFallbackException) // The exception is a heap allocation and not ideal
+            {
+                return false;
+            }
         }
 
         /// <summary>
@@ -866,18 +874,13 @@ private static BitArray PlainTextToBinaryAlphanumeric(string plainText)
             return codeText;
         }
 
-        /// <summary>
-        /// Returns a string that contains the original string, with characters that cannot be encoded by a
-        /// specified encoding (default of ISO-8859-2) with a replacement character.
-        /// </summary>
-        private static string ConvertToIso8859(string value, string Iso = "ISO-8859-2")
-        {
-            Encoding iso = Encoding.GetEncoding(Iso);
-            Encoding utf8 = Encoding.UTF8;
-            byte[] utfBytes = utf8.GetBytes(value);
-            byte[] isoBytes = Encoding.Convert(utf8, iso, utfBytes);
-            return iso.GetString(isoBytes);
-        }
+        private static readonly Encoding _iso8859_1 =
+#if NET5_0_OR_GREATER
+            Encoding.Latin1;
+#else
+            Encoding.GetEncoding(28591); // ISO-8859-1
+#endif
+        private static Encoding _iso8859_2;
 
         /// <summary>
         /// Converts plain text into a binary format using byte mode encoding, which supports various character encodings through ECI (Extended Channel Interpretations).
@@ -894,35 +897,69 @@ private static string ConvertToIso8859(string value, string Iso = "ISO-8859-2")
         /// </remarks>
         private static BitArray PlainTextToBinaryByte(string plainText, EciMode eciMode, bool utf8BOM, bool forceUtf8)
         {
-            byte[] codeBytes;
+            Encoding targetEncoding;
 
             // Check if the text is valid ISO-8859-1 and UTF-8 is not forced, then encode using ISO-8859-1.
             if (IsValidISO(plainText) && !forceUtf8)
-                codeBytes = Encoding.GetEncoding("ISO-8859-1").GetBytes(plainText);
+            {
+                targetEncoding = _iso8859_1;
+                utf8BOM = false;
+            }
             else
             {
                 // Determine the encoding based on the specified ECI mode.
                 switch (eciMode)
                 {
                     case EciMode.Iso8859_1:
                         // Convert text to ISO-8859-1 and encode.
-                        codeBytes = Encoding.GetEncoding("ISO-8859-1").GetBytes(ConvertToIso8859(plainText, "ISO-8859-1"));
+                        targetEncoding = _iso8859_1;
+                        utf8BOM = false;
                         break;
                     case EciMode.Iso8859_2:
+                        // Note: ISO-8859-2 is not natively supported on .NET Core
+                        //
+                        // Users must install the System.Text.Encoding.CodePages package and call Encoding.RegisterProvider(CodePagesEncodingProvider.Instance)
+                        // before using this encoding mode.
+                        if (_iso8859_2 == null)
+                            _iso8859_2 = Encoding.GetEncoding(28592); // ISO-8859-2
                         // Convert text to ISO-8859-2 and encode.
-                        codeBytes = Encoding.GetEncoding("ISO-8859-2").GetBytes(ConvertToIso8859(plainText, "ISO-8859-2"));
+                        targetEncoding = _iso8859_2;
+                        utf8BOM = false;
                         break;
                     case EciMode.Default:
                     case EciMode.Utf8:
                     default:
                         // Handle UTF-8 encoding, optionally adding a BOM if specified.
-                        codeBytes = utf8BOM ? Encoding.UTF8.GetPreamble().Concat(Encoding.UTF8.GetBytes(plainText)).ToArray() : Encoding.UTF8.GetBytes(plainText);
+                        targetEncoding = Encoding.UTF8;
                         break;
                 }
             }
 
+#if NETCOREAPP2_1_OR_GREATER || NETSTANDARD2_1
+            // We can use stackalloc for small arrays to prevent heap allocations
+            // Note that all QR codes should fit within 3000 bytes, so this code should never trigger a heap allocation unless an exception will be thrown anyway.
+            int count = targetEncoding.GetByteCount(plainText);
+            Span<byte> codeBytes = count < 3000 ? stackalloc byte[count] : new byte[count];
+            targetEncoding.GetBytes(plainText, codeBytes);
+#else
+            byte[] codeBytes = targetEncoding.GetBytes(plainText);
+#endif
+
             // Convert the array of bytes into a BitArray.
-            return ToBitArray(codeBytes);
+            if (utf8BOM)
+            {
+                // convert to bit array, leaving 24 bits for the UTF-8 preamble
+                var bitArray = ToBitArray(codeBytes, 24);
+                // write UTF8 preamble (EF BB BF) to the BitArray
+                DecToBin(0xEF, 8, bitArray, 0);
+                DecToBin(0xBB, 8, bitArray, 8);
+                DecToBin(0xBF, 8, bitArray, 16);
+                return bitArray;
+            }
+            else
+            {
+                return ToBitArray(codeBytes);
+            }
         }
 
         /// <summary>
@@ -932,7 +969,13 @@ private static BitArray PlainTextToBinaryByte(string plainText, EciMode eciMode,
         /// <param name="byteArray">The byte array to convert into a BitArray.</param>
         /// <param name="prefixZeros">The number of leading zeros to prepend to the resulting BitArray.</param>
         /// <returns>A BitArray representing the bits of the input byteArray, with optional leading zeros.</returns>
-        private static BitArray ToBitArray(byte[] byteArray, int prefixZeros = 0)
+        private static BitArray ToBitArray(
+#if NETCOREAPP2_1_OR_GREATER || NETSTANDARD2_1
+            ReadOnlySpan<byte> byteArray, // byte[] has an implicit cast to ReadOnlySpan<byte>
+#else
+            byte[] byteArray,
+#endif
+            int prefixZeros = 0)
         {
             // Calculate the total number of bits in the resulting BitArray including the prefix zeros.
             var bitArray = new BitArray((int)((uint)byteArray.Length * 8) + prefixZeros);

QRCoderTests/QRGeneratorTests.cs

@@ -160,6 +160,26 @@ public void can_encode_byte()
             result.ShouldBe("0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001111111001011011111110000000010000010011100100000100000000101110101101101011101000000001011101001010010111010000000010111010001010101110100000000100000100000101000001000000001111111010101011111110000000000000000110110000000000000000111011111111011000100000000001001110001100010000010000000010011110001010001001000000000110011010000001000110000000001110001111001010110110000000000000000111101010011100000000111111101111011100110000000001000001010011101110010000000010111010110101110010100000000101110100110001000110000000001011101011001000100010000000010000010100000100011000000000111111101110101010111000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000");
         }
 
+        [Fact]
+        [Category("QRGenerator/TextEncoding")]
+        public void can_encode_utf8()
+        {
+            var gen = new QRCodeGenerator();
+            var qrData = gen.CreateQrCode("https://en.wikipedia.org/wiki/🍕", QRCodeGenerator.ECCLevel.L, true, false, QRCodeGenerator.EciMode.Utf8);
+            var result = string.Join("", qrData.ModuleMatrix.Select(x => x.ToBitString()).ToArray());
+            result.ShouldBe("0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011111110101011010011101111111000000001000001001111100001110100000100000000101110101110000011000010111010000000010111010111010111100101011101000000001011101010011010111010101110100000000100000100011010001110010000010000000011111110101010101010101111111000000000000000000101000011100000000000000000111100101011110101011100111010000000001011000101011111010011101010000000001010011101111101001111011101000000000111011011110000010001100000100000000000000010011010101100000000000000000001100110101011011111001101110000000000000011100001010101010110101000000000000111001011100110111111110011000000001110101011001011001000100011000000000000101010100001010111111000000000000010111010101001111100000001110000000000010110100010111111100100010100000000011101111010011101111111101010000000000000000110000001000100010010000000001111111001100011001010101101000000000100000100111111111011000111000000000010111010010100011010111110111000000001011101010110100011100101011000000000101110101100101111100101111010000000010000010111011001111000001101000000001111111011110000100000110101000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000");
+        }
+
+        [Fact]
+        [Category("QRGenerator/TextEncoding")]
+        public void can_encode_utf8_bom()
+        {
+            var gen = new QRCodeGenerator();
+            var qrData = gen.CreateQrCode("https://en.wikipedia.org/wiki/🍕", QRCodeGenerator.ECCLevel.L, true, true, QRCodeGenerator.EciMode.Utf8);
+            var result = string.Join("", qrData.ModuleMatrix.Select(x => x.ToBitString()).ToArray());
+            result.ShouldBe("0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000011111110010001101010101111111000000001000001011011000110000100000100000000101110100111010101111010111010000000010111010110100100010101011101000000001011101000101111000010101110100000000100000101010000111000010000010000000011111110101010101010101111111000000000000000000001010101110000000000000000111110111110101010100101010100000000000100000110000101000001100101000000000001001001011000011010000111100000000100010001111000001111110111010000000010110111010100011100100101111000000000001010001101101001000010100100000000100001101110011001010000001010000000001011001100011001111111010111000000000010001010101011110010100000100000000100100010000000000010110010000000000010110110010110000101010101100000000001001100100010010100111101101100000000101010110011000111101111100100000000000000000111011110011100011010000000001111111011100110010010101110000000000100000100100110010101000110110000000010111010110010111101111110011000000001011101010100000100010110100000000000101110101001100111110110111100000000010000010111100101111100100001000000001111111011110001110100111000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000");
+        }
+
         [Fact]
         [Category("QRGenerator/TextEncoding")]
         public void can_generate_from_bytes()
@@ -170,6 +190,32 @@ public void can_generate_from_bytes()
             var result = string.Join("", qrData.ModuleMatrix.Select(x => x.ToBitString()).ToArray());
             result.ShouldBe("0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001111111011001011111110000000010000010010010100000100000000101110101010101011101000000001011101010010010111010000000010111010111000101110100000000100000100000001000001000000001111111010101011111110000000000000000011000000000000000000111100101010010011101000000001011100001001001001110000000010101011111011111110100000000000101000000110000000000000001011001001010100110000000000000000000110001000101000000000111111100110011011110000000001000001001111110111010000000010111010011100100101100000000101110101110010010010000000001011101011010100011000000000010000010110110101000100000000111111101011100010000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000");
         }
+
+        [Fact]
+        [Category("QRGenerator/TextEncoding")]
+        public void isValidIso_works()
+        {
+            // see private method: QRCodeGenerator.IsValidISO
+
+            Encoding _iso88591ExceptionFallback = Encoding.GetEncoding(28591, new EncoderExceptionFallback(), new DecoderExceptionFallback()); // ISO-8859-1
+
+            IsValidISO("abc").ShouldBeTrue();
+            IsValidISO("äöü").ShouldBeTrue();
+            IsValidISO("🍕").ShouldBeFalse();
+
+            bool IsValidISO(string input)
+            {
+                try
+                {
+                    _ = _iso88591ExceptionFallback.GetByteCount(input);
+                    return true;
+                }
+                catch (EncoderFallbackException)
+                {
+                    return false;
+                }
+            }
+        }
     }
 
     public static class ExtensionMethods