Add a test for PngEncoder inference

tests/ImageSharp.Tests/Formats/Png/PngEncoderTests.cs

@@ -195,6 +195,49 @@ namespace SixLabors.ImageSharp.Tests.Formats.Png
             }
         }
 
+        [Theory]
+        // does the following make sense? Or is it supposed to encode a 16bpp with two 8bit channels?
+        [WithBlankImages(1, 1, PixelTypes.Alpha8, PngColorType.GrayscaleWithAlpha, PngBitDepth.Bit8)]
+        [WithBlankImages(1, 1, PixelTypes.Argb32, PngColorType.RgbWithAlpha, PngBitDepth.Bit8)]
+        // [WithBlankImages(1, 1, PixelTypes.Bgr565, Can't reasonably be inferred)]
+        // [WithBlankImages(1, 1, PixelTypes.Bgra4444, Can't reasonably be inferred)]
+        // [WithBlankImages(1, 1, PixelTypes.Byte4, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.HalfSingle, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.HalfVector2, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.HalfVector4, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.NormalizedByte2, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.NormalizedByte4, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.NormalizedShort4, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Rg32, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Rgba1010102, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Rgba32, PngColorType.RgbWithAlpha, PngBitDepth.Bit8)]
+        // [WithBlankImages(1, 1, PixelTypes.Rgba64, PngColorType.RgbWithAlpha, PngBitDepth.Bit16)]
+        // [WithBlankImages(1, 1, PixelTypes.RgbaVector, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Short2, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Short4, I'm not sure)]
+        [WithBlankImages(1, 1, PixelTypes.Rgb24, PngColorType.Rgb, PngBitDepth.Bit8)]
+        // [WithBlankImages(1, 1, PixelTypes.Bgr24, I'm not sure)]
+        // [WithBlankImages(1, 1, PixelTypes.Bgra32, I'm not sure)]
+        [WithBlankImages(1, 1, PixelTypes.Rgb48, PngColorType.Rgb, PngBitDepth.Bit16)]
+        // [WithBlankImages(1, 1, PixelTypes.Bgra5551, I'm not sure)]
+        [WithBlankImages(1, 1, PixelTypes.Gray8, PngColorType.Grayscale, PngBitDepth.Bit8)]
+        [WithBlankImages(1, 1, PixelTypes.Gray16, PngColorType.Grayscale, PngBitDepth.Bit8)]
+        public void InfersColorTypeAndBitDepth<TPixel>(TestImageProvider<TPixel> provider, PngColorType pngColorType, PngBitDepth pngBitDepth)
+            where TPixel : struct, IPixel<TPixel>
+        {
+            Stream stream = new MemoryStream();
+            PngEncoder encoder = new PngEncoder();
+            encoder.Encode<TPixel>(provider.GetImage(), stream);
+
+            stream.Seek(0, SeekOrigin.Begin);
+
+            PngDecoder decoder = new PngDecoder();
+
+            Image image = decoder.Decode(Configuration.Default, stream);
+
+            Assert.True(image is Image<TPixel>);
+        }
+
         [Theory]
         [WithFile(TestImages.Png.Palette8Bpp, nameof(PaletteLargeOnly), PixelTypes.Rgba32)]
         public void PaletteColorType_WuQuantizer<TPixel>(TestImageProvider<TPixel> provider, int paletteSize)

tests/ImageSharp.Tests/TestUtilities/PixelTypes.cs

@@ -62,6 +62,8 @@ namespace SixLabors.ImageSharp.Tests
 
         Gray8 = 1 << 23,
 
+        Gray16 = 1 << 24,
+
         // TODO: Add multi-flag entries by rules defined in PackedPixelConverterHelper
 
         // "All" is handled as a separate, individual case instead of using bitwise OR