A handy in-memory netCDF Dataset for fortran, for 1-D arrays only (so making it more similar to a pandas DataFrame).
Useful commands:
use dataset, only: Dataset
type(Dataset) :: ds
Allocate memory for an underlying ds%values(:,:) pointer
call ds%alloc(nvar=3, nlen=5)
Define variable names (default to "v1", "v2"...)
ds%names(1) = "x"
ds%names(2) = "v1"
ds%names(3) = "v2"
Can set and get variables by name
call ds%setitem('x', [1.d0, 2.d0, 3.d0, 4.d0, 5.d0])
call ds%setitem('v1',ds%getitem('x')**2)
Or directly access the underlying values attribute, using the iname method
to obtain the integer position index corresponding to a variable name.
call ds%values(:, ds%iname('v2')) = ds%values(:, ds%iname('x'))+0.d0
Includes pretty printing or arrays a la pandas' DataFrame
call ds%print()
Dataset(nvar=3, nlen=5)
x v1 v2
1 1.0000E+00 1.0000E+00 1.0000E+00
2 2.0000E+00 4.0000E+00 2.0000E+00
3 3.0000E+00 9.0000E+00 3.0000E+00
4 4.0000E+00 1.6000E+01 4.0000E+00
5 5.0000E+00 2.5000E+01 5.0000E+00
As well as netCDF I/O based on Alex Robinson's ncio module
call ds%write_nc('test.nc')
call ds%read_nc('test.nc')
And some basic functionality such as interpolation
call ds%set_index(ds%iname("x")) ! set the x variable as index
ds2 = ds%interp([1.d0, 1.5d0,4.d0]) ! interpolate on new axis
Dataset(nvar=3, nlen=3)
x v1 v2
1 1.0000E+00 1.0000E+00 1.0000E+00
2 1.5000E+00 2.5000E+00 1.5000E+00
3 4.0000E+00 1.6000E+01 4.0000E+00
It is also possible to use methods such as slice, compress, take, to index the dataset as a whole along the first (index) or second (variable) dimension.
More functions such as align or reidnex could be built in. The current limitation is that only a single type can be represented in the DataFrame (currently double precision), but this has the advantage of being efficient. A more, complex, mixed-type and mixed-dimensions dataset will be presented in another project.