ISIS2 Improvements

[warmerdam]

Hi,

i have a patch for gdal 1.7.3 to improve isis2 driver. The patch had

• writing functionality for QUBE object
• grab null value
• improve the type reading
• can read qube with back plane, side plane and bottom plane
• can read a file with label detached or attached.

and i have modify nasakeyword handler to reduce sensibility of parser to little error on comment. (Like if we have char after end comment.)

but i have rewrite/move a lot of lines... I use this patch for a research project from one year ago. I work on hyperspectral data with OTB library.

I think is it possible to easily merge pdsdataset.cpp and isis2dataset.cpp pdsdataset work on IMAGE tag and isis2 on QUBE tag but it is the same workflow for all object (QUBE, IMAGE or SPECTRAL_QUBE)

Warning for writing no geographical information are writing now, and create_copy is not implemented.

Ludovic.

[warmerdam]

ISIS2 improvements from Ludovic.

[warmerdam]

Trent notes:

What is curious is that the ISIS2 reader already supported

• grab null value (hard-coded)
• can read a file with label detached or attached.

I know the current ISIS2 options didn't support:

• writing functionality for QUBE object
• can read qube with back plane, side plane and bottom plane

thanks, Trent

[warmerdam]

There is a lot of material in this patch. As a preliminary step I have incorporated the create method with a few changes and improvements (r21618). It would be very helpful to have examples for some of the read improvements so I wouldn't just have to trust them blind.

[warmerdam]

OK, I've done a bunch more improvements for the ISIS2 driver, some from the proposed patch, and included format docs for all three drivers (r21710).

I have not include support for different interleaving, nor for the SUFFIX_ITEMS.

I'd like to support either or both if read-world samples can be provided but I don't want to do it blind, and I don't want to do it if there is no practical use of these options.

[Even Rouault]

Frank, I thinks r21710 breaks the autotest/gdrivers/pds.py. I suspect the new ISIS2Dataset::Identify() to wrongly identify PDS datasets as ISIS2 ones.

I've tried to replace the && test by

. This makes pds.py to pass, but isis2.py then fail...

[Even Rouault]

The issue in autotest has been fixed by r21772. Closing