Add JPX bits per component decoding

src/UglyToad.PdfPig.Tests/Images/Jpeg2000HelperTests.cs

@@ -0,0 +1,38 @@
+namespace UglyToad.PdfPig.Tests.Images
+{
+    using PdfPig.Images;
+    using System;
+
+    public class Jpeg2000HelperTests
+    {
+        private static readonly Lazy<string> DocumentFolder = new Lazy<string>(() => Path.GetFullPath(Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "..", "..", "..", "Images", "Files", "Jpx")));
+
+        public static IEnumerable<object[]> GetAllDocuments
+        {
+            get
+            {
+                return Directory.GetFiles(DocumentFolder.Value, "*.jp2").Select(x => new object[] { Path.GetFileName(x) });
+            }
+        }
+
+        [Fact]
+        public void GetJp2BitsPerComponent_ThrowsException_WhenInputIsTooShort()
+        {
+            Assert.Throws<InvalidOperationException>(() => Jpeg2000Helper.GetJp2BitsPerComponent(new byte[11]));
+        }
+        
+        [Fact]
+        public void GetJp2BitsPerComponent_ThrowsException_WhenSignatureBoxIsInvalid()
+        {
+            Assert.Throws<InvalidOperationException>(() => Jpeg2000Helper.GetJp2BitsPerComponent(new byte[12]));
+        }
+
+        [Theory]
+        [MemberData(nameof(GetAllDocuments))]
+        public void GetJp2BitsPerComponent_ReturnsCorrectBitsPerComponent_WhenValidInput(string path)
+        {
+            byte[] image = File.ReadAllBytes(Path.Combine(DocumentFolder.Value, path));
+            Assert.Equal(8, Jpeg2000Helper.GetJp2BitsPerComponent(image));
+        }
+    }
+}

src/UglyToad.PdfPig.Tests/UglyToad.PdfPig.Tests.csproj

@@ -132,6 +132,78 @@
     <None Update="Dla\Documents\Random 2 Columns Lists Hyph - Justified.pdf">
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </None>
+    <None Update="Images\Files\Jpx\1310ba77-4dbf-4d8d-a8f4-5ba59d1221a7.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\2cee5bb6-f845-4ac1-8156-a899075c0b46.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\33fb977a-e3da-48da-ad51-89af637ab736.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\371028e4-aea3-4e1d-b76b-47b763922e2f.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\804344c3-2c63-4e9c-b7c2-8c64a14d885b.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\82811cfb-9a70-475d-8338-f20df0acd052.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\82828826-f624-4f22-8421-f8c4adac43a3.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\8e95baf6-874e-431c-9cbc-d735ccabac0c.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\91a217c9-79bb-4a4b-934b-1362344f6b89.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\99a9ea0e-c407-4336-96a0-85023f46c231.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\9ac01df9-6623-4d14-89fd-e9934d1a6c7e.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\9d5c783a-c001-40e9-91b9-630c71804a77.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\9df64d7b-4003-4d0d-8f68-b5f88de781b7.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\a6105bfd-3ace-4d6b-b2dc-f9ce4022832b.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\b650c344-bc4d-427a-94af-cfed04136f67.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\c390b9c7-a562-42bf-a592-7a7b29819a6a.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\d6b4b35c-0ceb-47fe-aba8-4360acb49fcb.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\deff000e-a14a-40fd-bf39-88ce11745260.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\e29266a2-201a-4ad6-9725-ca1b7c22224d.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\eae2cabb-f520-4be5-932f-fb19fce5b2f2.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\eb62f062-6567-48b2-b04d-6d90de120f07.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\ed5c585f-590e-4585-9ce7-25976f589ca8.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\ef0af08b-04d1-4a3e-a9d8-7916b9826f5d.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
+    <None Update="Images\Files\Jpx\fd42e6a0-5c7a-4eb2-b0e3-474cfde067a6.jp2">
+      <CopyToOutputDirectory>Never</CopyToOutputDirectory>
+    </None>
   </ItemGroup>
   <ItemGroup>
     <Using Include="Xunit" />

src/UglyToad.PdfPig/Images/Jpeg2000Helper.cs

@@ -0,0 +1,103 @@
+namespace UglyToad.PdfPig.Images
+{
+    using System;
+    using System.Buffers.Binary;
+
+    internal static class Jpeg2000Helper
+    {
+        /// <summary>
+        /// Get bits per component values for Jp2 (Jpx) encoded images (first component).
+        /// </summary>
+        public static byte GetJp2BitsPerComponent(ReadOnlySpan<byte> jp2Bytes)
+        {
+            // Ensure the input has at least 12 bytes for the signature box
+            if (jp2Bytes.Length < 12)
+            {
+                throw new InvalidOperationException("Input is too short to be a valid JP2 file.");
+            }
+
+            // Verify the JP2 signature box
+            uint length = BinaryPrimitives.ReadUInt32BigEndian(jp2Bytes.Slice(0, 4));
+            uint type = BinaryPrimitives.ReadUInt32BigEndian(jp2Bytes.Slice(4, 4));
+            uint magic = BinaryPrimitives.ReadUInt32BigEndian(jp2Bytes.Slice(8, 4));
+
+            if (length != 0x0000000C || type != 0x6A502020 || magic != 0x0D0A870A)
+            {
+                throw new InvalidOperationException("Invalid JP2 signature box.");
+            }
+
+            // Proceed to parse JP2 boxes
+            return ParseBoxes(jp2Bytes.Slice(12));
+        }
+
+        private static byte ParseBoxes(ReadOnlySpan<byte> jp2Bytes)
+        {
+            int offset = 0;
+            while (offset < jp2Bytes.Length)
+            {
+                if (offset + 8 > jp2Bytes.Length)
+                {
+                    throw new InvalidOperationException("Invalid JP2 box structure.");
+                }
+
+                // Read box length and type
+                uint boxLength = BinaryPrimitives.ReadUInt32BigEndian(jp2Bytes.Slice(offset, 4));
+                uint boxType = BinaryPrimitives.ReadUInt32BigEndian(jp2Bytes.Slice(offset + 4, 4));
+
+                // Check for the contiguous codestream box ('jp2c')
+                if (boxType == 0x6A703263) // 'jp2c' in ASCII
+                {
+                    // Parse the codestream to find the SIZ marker
+                    return ParseCodestream(jp2Bytes.Slice(offset + 8));
+                }
+
+                // Move to the next box
+                offset += (int)(boxLength > 0 ? boxLength : 8); // Box length of 0 means the rest of the file
+            }
+
+            throw new InvalidOperationException("Codestream box not found in JP2 file.");
+        }
+
+        private static byte ParseCodestream(ReadOnlySpan<byte> codestream)
+        {
+            int offset = 0;
+            while (offset + 2 <= codestream.Length)
+            {
+                // Read marker (2 bytes)
+                ushort marker = BinaryPrimitives.ReadUInt16BigEndian(codestream.Slice(offset, 2));
+
+                // Check for SIZ marker (0xFF51)
+                if (marker == 0xFF51)
+                {
+                    if (offset + 38 > codestream.Length)
+                    {
+                        throw new InvalidOperationException("Invalid SIZ marker structure.");
+                    }
+
+                    // Skip marker length (2 bytes), capabilities (4 bytes), and reference grid size (8 bytes)
+                    // Skip image offset (8 bytes), tile size (8 bytes), and tile offset (8 bytes)
+                    offset += 38;
+
+                    // Read number of components (2 bytes)
+                    ushort numComponents = BinaryPrimitives.ReadUInt16BigEndian(codestream.Slice(offset, 2));
+
+                    offset += 2;
+                    if (numComponents < 1)
+                    {
+                        throw new InvalidOperationException("Invalid number of components in SIZ marker.");
+                    }
+
+                    // Read bits per component for the first component (1 byte per component)
+                    byte bitsPerComponent = codestream[offset];
+
+                    // Bits per component is stored as (bits - 1)
+                    return ++bitsPerComponent;
+                }
+                // Move to the next marker
+                offset += 2;
+            }
+
+            throw new InvalidOperationException("SIZ marker not found in JPEG2000 codestream.");
+        }
+    }
+}

src/UglyToad.PdfPig/XObjects/XObjectFactory.cs

@@ -8,6 +8,7 @@
     using Graphics;
     using Graphics.Colors;
     using Graphics.Core;
+    using Images;
     using Tokenization.Scanner;
     using Tokens;
     using Util;
@@ -52,20 +53,37 @@
             var isJpxDecode = dictionary.TryGet(NameToken.Filter, pdfScanner, out NameToken filterName)
                 && filterName.Equals(NameToken.JpxDecode);
 
-            int bitsPerComponent = 0;
-            if (!isImageMask && !isJpxDecode)
-            {
-                if (!dictionary.TryGet(NameToken.BitsPerComponent, pdfScanner, out NumericToken? bitsPerComponentToken))
-                {
-                    throw new PdfDocumentFormatException($"No bits per component defined for image: {dictionary}.");
-                }
-
-                bitsPerComponent = bitsPerComponentToken.Int;
-            }
-            else if (isImageMask)
+            int bitsPerComponent;
+            if (isImageMask)
             {
                 bitsPerComponent = 1;
             }
+            else
+            {
+                if (isJpxDecode)
+                {
+                    // Optional for JPX
+                    if (dictionary.TryGet(NameToken.BitsPerComponent, pdfScanner, out NumericToken? bitsPerComponentToken))
+                    {
+                        bitsPerComponent = bitsPerComponentToken.Int;
+                        System.Diagnostics.Debug.Assert(bitsPerComponent == Jpeg2000Helper.GetJp2BitsPerComponent(xObject.Stream.Data.Span));
+                    }
+                    else
+                    {
+                        bitsPerComponent = Jpeg2000Helper.GetJp2BitsPerComponent(xObject.Stream.Data.Span);
+                        System.Diagnostics.Debug.Assert(new int[] { 1, 2, 4, 8, 16 }.Contains(bitsPerComponent));
+                    }
+                }
+                else
+                {
+                    if (!dictionary.TryGet(NameToken.BitsPerComponent, pdfScanner, out NumericToken? bitsPerComponentToken))
+                    {
+                        throw new PdfDocumentFormatException($"No bits per component defined for image: {dictionary}.");
+                    }
+
+                    bitsPerComponent = bitsPerComponentToken.Int;
+                }
+            }
 
             var intent = xObject.DefaultRenderingIntent;
             if (dictionary.TryGet(NameToken.Intent, pdfScanner, out NameToken renderingIntentToken))