Regression opening RLE and deflate encoded images.

[JimBobSquarePants]

Magick.NET version

Magick.NET-Q16-AnyCPU 14.9.1

Environment (Operating system, version and so on)

Windows 11 .NET 8

Description

I'm upgrading the version of Magick.NET-Q16-AnyCPU used by the ImageSharp tests reference decoder from v13.10.0 to v14.9.1 to avoid security warnings and am seeing new exceptions for four images that were previously possible to decode.

I've attached the images in a zip file. The exceptions returned are listed below:

RunLengthEncoded-inverted.bmp

      Message: 
        ImageMagick.MagickCoderErrorException : compression not supported `' @ error/bmp.c/ReadBMPImage/843

      Stack Trace: 
        NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller, Void* data)
        NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller)
        MagickImageCollection.AddImages(Stream stream, IMagickReadSettings`1 readSettings, Boolean ping)
        MagickImageCollection.Read(Stream stream, IMagickReadSettings`1 readSettings)
        MagickImageCollection.ctor(Stream stream, IMagickReadSettings`1 readSettings)

pal8rlecut.bmp

      Message: 
ImageMagick.MagickCorruptImageErrorException : unable to runlength decode image `' @ error/bmp.c/ReadBMPImage/1163

      Stack Trace: 
NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller, Void* data)
MagickImageCollection.AddImages(Stream stream, IMagickReadSettings`1 readSettings, Boolean ping)

BigTIFF_Indexed4_Deflate.tif

      Message: 
ImageMagick.MagickCoderErrorException : Decoding error at scanline 0. `ZIPDecode' @ error/tiff.c/TIFFErrors/575

      Stack Trace: 
NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller, Void* data)
MagickImageCollection.AddImages(Stream stream, IMagickReadSettings`1 readSettings, Boolean ping)

multipage_deflate_withPreview.tiff

      Message: 
ImageMagick.MagickCoderErrorException : Decoding error at scanline 0. `ZIPDecode' @ error/tiff.c/TIFFErrors/575

      Stack Trace: 
NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller, Void* data)
MagickImageCollection.AddImages(Stream stream, IMagickReadSettings`1 readSettings, Boolean ping)

rgb_small_deflate.tiff

      Message: 
ImageMagick.MagickCoderErrorException : Decoding error at scanline 0. `ZIPDecode' @ error/tiff.c/TIFFErrors/575

      Stack Trace: 
NativeMagickImageCollection.ReadStream(IMagickSettings`1 settings, ReadWriteStreamDelegate reader, SeekStreamDelegate seeker, TellStreamDelegate teller, Void* data)
MagickImageCollection.AddImages(Stream stream, IMagickReadSettings`1 readSettings, Boolean ping)

Steps to Reproduce

The crux of the code is as follows:

BmpReadDefines bmpReadDefines = new()
{
    IgnoreFileSize = !this.validate,
};
PngReadDefines pngReadDefines = new()
{
    IgnoreCrc = !this.validate
};

MagickReadSettings settings = new()
{
    FrameCount = options.MaxFrames
};
settings.SetDefines(bmpReadDefines);
settings.SetDefines(pngReadDefines);

using MagickImageCollection magickImageCollection = new(stream, settings);

Images

issues.zip

[dlemstra]

These files are failing because extra checks where added:

• RunLengthEncoded-inverted.bmp is now failing because it has a a negative height and is compressed with BI_RLE8. But the negative height is only allowed with BI_RGB or BI_BITFIELDS.
• pal8rlecut.bmp can no longer be read because there are not enough pixels to decode. It looks like the RLE stream is missing data. This now results in an error because a check for this was fixed.

The tiff files are failing inside this block of decompress_template.h:

} else if (block_type == DEFLATE_BLOCKTYPE_UNCOMPRESSED) {
	u16 len, nlen;

	/*
	 * Uncompressed block: copy 'len' bytes literally from the input
	 * buffer to the output buffer.
	 */

	bitsleft -= 3; /* for BTYPE and BFINAL */

	/*
	 * Align the bitstream to the next byte boundary.  This means
	 * the next byte boundary as if we were reading a byte at a
	 * time.  Therefore, we have to rewind 'in_next' by any bytes
	 * that have been refilled but not actually consumed yet (not
	 * counting overread bytes, which don't increment 'in_next').
	 */
	bitsleft = (u8)bitsleft;
	SAFETY_CHECK(overread_count <= (bitsleft >> 3));
	in_next -= (bitsleft >> 3) - overread_count;
	overread_count = 0;
	bitbuf = 0;
	bitsleft = 0;

	SAFETY_CHECK(in_end - in_next >= 4);
	len = get_unaligned_le16(in_next);
	nlen = get_unaligned_le16(in_next + 2);
	in_next += 4;

I enabled support for libdeflate between those two versions and that broke things. When I remove libdefate as a dependency all the files can be read again. Thanks for reporting this and it will be resolved in the next release.

[JimBobSquarePants]

Thanks @dlemstra for getting back to me so quickly, and on a weekend!

I'm a little confused by the new stance on the bmp decoder. Quite a lot of software can open both of those files without issue, including browsers and native Windows software. File Explorer even generates thumbnails for them.

That might surprise a lot of users.

[dlemstra]

The libdefate dependency has been removed in the current release. The other issues will remain because ImageMagick is more strict when reading certain files.