Fixes needed for FIBSEM .dat format

[acardona]

In line 210 of https://github.com/fiji/IO/blob/master/src/main/java/sc/fiji/io/FIBSEM_Reader.java, it reads, in the function readChannels:

		// it is always unsigned short
		final byte[] slice = new byte[ (int)header.xRes * (int)header.yRes * numChannels * 2 ];
		file.read( slice );
		final ByteBuffer buffer = ByteBuffer.wrap( slice );
		final ShortBuffer shortBuffer = buffer.asShortBuffer();

... which is not true. I have here some .dat files that are:

• signed 16-bit (very surprisingly)
• BIG_ENDIAN, so it needs ByteBuffer.wrap( slice ).order( ByteOrder.BIG_ENDIAN ) to be parsed correctly.

[imagejan]

For what it's worth, here's a permalink to the lines in question (you can create a permalink by clicking on the line numbers on the left hand side, and then using the context menu on the ... button):

IO/src/main/java/sc/fiji/io/FIBSEM_Reader.java

Lines 210 to 214 in 3aa1c1a

	// it is always unsigned short
	final byte[] slice = new byte[ (int)header.xRes * (int)header.yRes * numChannels * 2 ];
	file.read( slice );
	final ByteBuffer buffer = ByteBuffer.wrap( slice );
	final ShortBuffer shortBuffer = buffer.asShortBuffer();

---

@acardona is there any metadata in the .dat files that can be used to argue what type and what endianness is required?

[acardona]

Hi @imagejan, there should be, but the only documentation I have are these java files in this repository.

[jefferis]

Out of curiosity @acardona are the magic/header fields in your big endian files little endian even if the image data is big endian?

[acardona]

As far as I know, Greg @jefferis, the java file doesn't decode any header field for endianness. But I think it doesn't decode all header fields, of which there are many. Someone ought to look into the matlab file that converts DAT to TIFF, there it may be specified.