OutOfMemoryError while reading tif

Hi, we were getting following OOME while reading one of our tif images:

Exception in thread "main" java.lang.OutOfMemoryError: Java heap space
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readValue(TIFFReader.java:461)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readValueAt(TIFFReader.java:372)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readEntry(TIFFReader.java:345)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readIFD(TIFFReader.java:178)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readSubIFDs(TIFFReader.java:229)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readIFD(TIFFReader.java:194)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.readLinkedIFDs(TIFFReader.java:151)
	at com.twelvemonkeys.imageio.metadata.tiff.TIFFReader.read(TIFFReader.java:132)
	at com.twelvemonkeys.imageio.plugins.tiff.TIFFImageReader.readMetadata(TIFFImageReader.java:187)
	at com.twelvemonkeys.imageio.plugins.tiff.TIFFImageReader.readIFD(TIFFImageReader.java:380)
	at com.twelvemonkeys.imageio.plugins.tiff.TIFFImageReader.read(TIFFImageReader.java:916)
	at javax.imageio.ImageIO.read(ImageIO.java:1462)

I suppose that the image metadata are corrupted as I was able to get rid of this error by altering the image (e.q. by opening the image and saving again in graphical editor or manually just with the text editor). Would be nice if you could improve the twelvemonkeys implementation and skip loading of metadata to prevent OOME in such case as other tools can read the image without crash.

Code to read image: BufferedImage image = ImageIO.read(new URL(wrong)); Tested with imageio 3.7.0 and 3.8.1.

The problematic image is attached in .zip as tif format is not supported as attachment. P07.61-3863-76_M178-Farbe.zip

[haraldk]

FIY: I added a small fix that will detect this situation and handle it gracefully, even if reading from URL. Will be included in next release.

[jan-pesta]

Hi Harald, thanks a lot. You’re right. The image can be read as a file. The exception above was my fault - I used original URL source with “file:///” protocol prefix at the beginning of the path. Now it works fine.