Puppy 0.1.4

A DSL for creating NetCDF files. Here’s a simple example:

from pup import *

class Test(NetCDF):
    # NC_GLOBAL attributes go here
    history = 'Created for a test'

    # dimensions need to be set explicitly only when they
    # have no variable associated with them
    dim0 = Dimension(2)

    # variables that don't specify dimensions are assumed to
    # be their own dimension
    time = Variable(range(10), record=True, units='days since 2008-01-01')

    # now a variable with dimensions (time,)
    temperature = Variable(range(10), (time,), units='deg C')

Test.save('simple.nc')

This will produce the following NetCDF file:

netcdf simple {
dimensions:
    dim0 = 2 ;
    time = UNLIMITED ; // (10 currently)
variables:
    int time(time) ;
        time:units = "days since 2008-01-01" ;
    int temperature(time) ;
        temperature:units = "deg C" ;

// global attributes:
        :history = "Created for a test" ;
data:

 time = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 ;

 temperature = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 ;
}

By default it uses pupynere for creating files, but this can be overloaded:

from netCDF4 import Dataset

class Test(NetCDF):
    loader = Dataset
    ...
Test.save('simple.nc', format='NETCDF4')

Changelog:

0.1.4:

Added support for masked arrays.

0.1.3:

Pass keyword arguments in save() to the loader.

0.1.2:

Improved optional loader detection.

0.1.1:

Added pupynere dependency.

0.1:

Initial release.