Don't fail on non-4:4:4 JPEG

components/formats-bsd/src/loci/formats/services/JPEGTurboServiceImpl.java

@@ -88,6 +88,8 @@ public class JPEGTurboServiceImpl implements JPEGTurboService {
   private long sos;
   private long imageDimensions;
 
+  private int mcuWidth;
+  private int mcuHeight;
   private int tileWidth;
   private int tileHeight;
   private int xTiles;
@@ -156,6 +158,10 @@ public void initialize(RandomAccessInputStream jpeg, int width, int height)
       }
       else if (marker == SOF0) {
         imageDimensions = in.getFilePointer() + 1;
+        parseSOF();
+      }
+      else if (marker > 0xFFC0 && marker < 0xFFD0 && marker % 4 != 0) {
+        throw new IOException("Unsupported JPEG SOF marker: " + marker);
       }
       else if (marker == SOS) {
         sos = end;
@@ -206,7 +212,7 @@ else if (marker == SOS) {
       }
     }
 
-    tileWidth = restartInterval * 8;
+    tileWidth = restartInterval * mcuWidth;
     tileHeight = (int) Math.min(tileWidth, 512);
 
     xTiles = imageWidth / tileWidth;
@@ -289,9 +295,9 @@ public byte[] getTile(int tileX, int tileY) throws IOException {
 
     long dataLength = header.length + 2;
 
-    int mult = tileHeight / 8; // was restartInterval
+    int mult = tileHeight / mcuHeight; // was restartInterval
     int start = tileX + (tileY * xTiles * mult);
-    for (int row=0; row<tileHeight/8; row++) {
+    for (int row=0; row<tileHeight/mcuHeight; row++) {
       int end = start + 1;
 
       long startOffset = restartMarkers.get(start);
@@ -314,7 +320,7 @@ public byte[] getTile(int tileX, int tileY) throws IOException {
     offset += header.length;
 
     start = tileX + (tileY * xTiles * mult);
-    for (int row=0; row<tileHeight/8; row++) {
+    for (int row=0; row<tileHeight/mcuHeight; row++) {
       int end = start + 1;
 
       long endOffset = in.length();
@@ -374,6 +380,8 @@ public void close() throws IOException {
     restartInterval = 1;
     sos = 0;
     imageDimensions = 0;
+    mcuWidth = 0;
+    mcuHeight = 0;
     tileWidth = 0;
     tileHeight = 0;
     xTiles = 0;
@@ -396,4 +404,53 @@ private byte[] getFixedHeader() throws IOException {
     return header;
   }
 
+  private void parseSOF() throws IOException {
+    // https://mykb.cipindanci.com/archive/SuperKB/1294/JPEG%20File%20Layout%20and%20Format.htm
+    // example: FFC00011 08001100 11030122 00021101 031101
+    int bpc = in.readByte() & 0xff;
+    if (bpc != 8) {
+      throw new IOException("Only 8-bit channels supported by this reader");
+    }
+    in.skipBytes(4);
+    int channels = in.readByte() & 0xff;
+    if (channels != 3) {
+      // https://stackoverflow.com/questions/51008883/is-there-a-grayscale-jpg-format
+      throw new IOException("Only images with 3 channels are supported by this reader");
+    }
+
+    // Sampling factors: https://groups.google.com/g/comp.compression/c/Q8FUSocL7nA
+    // https://stackoverflow.com/q/27918757 https://stackoverflow.com/q/43225439
+    // mcu_tool.sh shows MCU size for some images
+    // JpegSnoop or identify -verbose show subsampling rates
+    // convert -sampling rate 2x2 or ffmpeg -i .. -vf format=yuv420p for making images
+
+    // MCU size in X direction divided by 8 is the ratio of largest X to smallest X
+    // likewise for Y. The first 4 bits is X, the next 4 bits make up Y.
+    // some examples: 2x1,2x1,2x1 (0x21,0x21,0x21) is 8x8. because 2/2 = 1, 1/1 = 1
+    // 2x1,1x1,1x1 is 16x8. 3x2,1x1,1x1 is 24x16
+
+    // No need to shift now
+    int maxX = 0x00;
+    int minX = 0xf0;
+    int maxY = 0x00;
+    int minY = 0x0f;
+
+    for (int i = 0; i < channels; i++) {
+      int componentId = in.readByte() & 0xff;
+
+      int rates = in.readByte();
+      int X = rates & 0xf0;
+      int Y = rates & 0x0f;
+      maxX = Math.max(maxX, X);
+      minX = Math.min(minX, X);
+      maxY = Math.max(maxY, Y);
+      minY = Math.min(minY, Y);
+
+      int quantTableNumber = in.readByte() & 0xff;
+    }
+
+    mcuHeight = maxX / minX * 8;
+    mcuWidth = maxY / minY * 8;
+  }
+
 }