TMI-18: Better filtering of APP segments, now only takes 'Exif' APP1 segments into account.

+ Updated failing test.

imageio/imageio-jpeg/src/main/java/com/twelvemonkeys/imageio/plugins/jpeg/JPEGSegmentImageInputStream.java

@@ -29,11 +29,13 @@
 package com.twelvemonkeys.imageio.plugins.jpeg;
 
 import com.twelvemonkeys.imageio.metadata.jpeg.JPEG;
+import com.twelvemonkeys.lang.Validate;
 
 import javax.imageio.IIOException;
 import javax.imageio.stream.ImageInputStream;
 import javax.imageio.stream.ImageInputStreamImpl;
 import java.io.IOException;
+import java.nio.charset.Charset;
 import java.util.ArrayList;
 import java.util.List;
 
@@ -90,7 +92,7 @@ final class JPEGSegmentImageInputStream extends ImageInputStreamImpl {
                 int marker = stream.readUnsignedShort();
 
                 // TODO: Refactor to make various segments optional, we probably only want the "Adobe" APP14 segment, 'Exif' APP1 and very few others
-                if (isAppSegmentMarker(marker) && marker != JPEG.APP0 && marker != JPEG.APP1 && marker != JPEG.APP14) {
+                if (isAppSegmentMarker(marker) && marker != JPEG.APP0 && !(marker == JPEG.APP1 && isAppSegmentWithId("Exif", stream)) && marker != JPEG.APP14) {
                     int length = stream.readUnsignedShort(); // Length including length field itself
                     stream.seek(realPosition + 2 + length);  // Skip marker (2) + length
                 }
@@ -138,6 +140,38 @@ final class JPEGSegmentImageInputStream extends ImageInputStreamImpl {
         return segment;
     }
 
+    private static boolean isAppSegmentWithId(String segmentId, ImageInputStream stream) throws IOException {
+        Validate.notNull(segmentId, "segmentId");
+
+        stream.mark();
+
+        try {
+            int length = stream.readUnsignedShort(); // Length including length field itself
+
+            byte[] data = new byte[Math.max(20, length - 2)];
+            stream.readFully(data);
+
+            return segmentId.equals(asNullTerminatedAsciiString(data, 0));
+        }
+        finally {
+            stream.reset();
+        }
+    }
+
+    static String asNullTerminatedAsciiString(final byte[] data, final int offset) {
+        for (int i = 0; i < data.length - offset; i++) {
+            if (data[i] == 0 || i > 255) {
+                return asAsciiString(data, offset, offset + i);
+            }
+        }
+
+        return null;
+    }
+
+    static String asAsciiString(final byte[] data, final int offset, final int length) {
+        return new String(data, offset, length, Charset.forName("ascii"));
+    }
+
     private void streamInit() throws IOException {
         stream.seek(0);
 

imageio/imageio-jpeg/src/test/java/com/twelvemonkeys/imageio/plugins/jpeg/JPEGSegmentImageInputStreamTest.java

@@ -102,7 +102,7 @@ public class JPEGSegmentImageInputStreamTest {
 
         assertThat(length, new LessOrEqual<Long>(10203l)); // In no case should length increase
 
-        assertEquals(9495l, length); // May change, if more chunks are passed to reader...
+        assertEquals(9625l, length); // May change, if more chunks are passed to reader...
     }
 
     @Test
@@ -110,14 +110,17 @@ public class JPEGSegmentImageInputStreamTest {
         ImageInputStream stream = new JPEGSegmentImageInputStream(ImageIO.createImageInputStream(getClassLoaderResource("/jpeg/no-image-types-rgb-us-web-coated-v2-ms-photogallery-exif.jpg")));
         List<JPEGSegment> appSegments = JPEGSegmentUtil.readSegments(stream, JPEGSegmentUtil.APP_SEGMENTS);
 
-        assertEquals(2, appSegments.size());
+        assertEquals(3, appSegments.size());
 
         assertEquals(JPEG.APP0, appSegments.get(0).marker());
         assertEquals("JFIF", appSegments.get(0).identifier());
 
-        assertEquals(JPEG.APP14, appSegments.get(1).marker());
+        assertEquals(JPEG.APP1, appSegments.get(1).marker());
-        assertEquals("Adobe", appSegments.get(1).identifier());
+        assertEquals("Exif", appSegments.get(1).identifier());
 
-        // And thus, no Exif, no ICC_PROFILE or other segments
+        assertEquals(JPEG.APP14, appSegments.get(2).marker());
+        assertEquals("Adobe", appSegments.get(2).identifier());
+
+        // And thus, no XMP, no ICC_PROFILE or other segments
     }
 }