Support for other spectral file formats

[ximeg]

Current version (0.6.2) of speview does support only a single binary data format, which is, unsurprisingly, called "SPE". It would be really great to add support for a wide range of data formats for the application, so it could be possible to display spectra obtained from other instruments.

Proposal

The simplest, smartest and the most effective way to do it would be to use some already available software to extract data from various file formats. One of the possibilities is to use hyperSpec for reading of files, extraction of data and representation of them in a standard way.

Subtasks

This seems to be quite a challenge, which is not simple and straightforward to implement. What we are going to do:

1. Check how hyperSpec imports data
2. Get sample datafiles in a variety of different formats (do they come with hyperSpec?)
3. Check how one can pass data from R to Python
4. Write some quick and dirty script that adds support for other file formats. It should work somehow and prove the principle!
5. Think about standard data representation – some kind of data container that provides values in a specific way to speview. Implement it, write a good documentation for it.
6. If necessary, refactor speview
7. Double-check that everything works, that the dependencies are solved correctly and that the package get installed in the right way.

[glemaitre]

@ximeg I am going into the same issue. I am used to Python than R and found that the package hyperSpec is quite complete. Did you find a solution regarding porting/warping this package to Python?

[ximeg]

@glemaitre Sorry, but I abandoned this package a long time ago and use only hyperSpec for handling of spectra. I don't think that it worth it to write a Python version of hyperSpec, or at least I don't have time for this.

Theoretically one could implement a Python class called hyperSpec that contains wavelength axis, the spectral matrix and the extra data - this is quite straightforward. Although this is the core of the hyperSpec, the most precious part is the huge number of helper functions for file import, plotting, subsetting, interpolations, etc. They could be ported into Python as class methods, and this is straighforward as well, but this would be a very big amount of work!

As a workaround I would suggest to use rpy2 to load any file format supported by hyperSpec, then access its wavelenth axis and spectral matrix as @wavelength slot and $spc element and return them into Python. I would recomend to use Pandas dataframe as a base class that imitates hyperSpec structure. Then you can extend it with methods that you need, e.g. for plotting.

[glemaitre]

Thanks I will check that.