write reader and writer for common format as csv and xls

[fernandezc]

We need to export to tabulated format (csv, xls... ) or to export form these formats. Other formats could also be interesting.

Solution:
Instead of writing completely new module, we can take advantage that Pandas is already present in the required package and write some wrappers to the function present in this library:
https://pandas.pydata.org/pandas-docs/stable/reference/io.html

[fernandezc]

By the way, In the next upcoming commit I will suppress the dependence to xlrd (used in only one orphan util file)

[fernandezc]

ooups correction! xlrd is used in read_carrousel.

May be we could look if we cannot use pandas instead to avoid importing additional librairies? @atravert

[atravert]

Yes. but i can change it for the panda reader.

when i think about this issue, the main "difficulty" we will have is to define where/how to write the data and maybe the meta data. Because these methods will be mostly used to read/export from/to third party applications that will all have specific requirements

For instance in csv/excel:

• 1D, 2D: col1=xcoord, col2 ... = data ; but depending on applications, line1 contains a tittl or is the 1st coord and the corresponding data
• 3D: it can be one csv file (or excel sheet) per 'z' élement , but where to store the z coord ?

And in any case where to store the other meta data ? in excel: on a specific sheet ; in csv: on a specific file ? or on the first lines of the csv ? , but then it will be less readable by 3d party apps.

For the, reading, the main difficulty is that the locations of data and metadata and th eformats will change a lot depending on the applications, and on the version of the apps, etc... So for the reader part, I was actually wondering if it is worth wrapping existing libraries for reading - that will probably lead to less flexibility ?, instead of using directly pandas ?

On a closely related subject: you mentioned the CSDM developped by D. Massiot et al. It would be also interesting to look at the open netCDF format (also used by xarray) , which is binary and widely used (though not much in chemistry). Before writing routines to read/write csv of excel to/from Spectrochempy , it could be interesting to see which are the current solutions adopted to convert netCDF to/from csv or excel ? ;