# This module contains adapters and a facade for the different python netcdf libraries. 

# $Id: opendap.py 4658 2011-06-13 15:41:23Z boer_g $
# $Date: 2011-06-13 17:41:23 +0200 (ma, 13 jun 2011) $
# $Author: boer_g $
# $Revision: 4658 $
# $HeadURL: https://repos.deltares.nl/repos/OpenEarthTools/trunk/python/io/opendap/opendap.py $
# $Keywords: $

import exceptions
import warnings

['close', 'cmptypes', 'createCompoundType', 'createDimension', 'createGroup', 'createVLType', 'createVariable', 'delncattr', 'dimensions', 'file_format', 'getncattr', 'groups', 'maskanscale', 'ncattrs', 'parent', 'path', 'renameDimension', 'renameVariable', 'set_fill_off', 'set_fill_on', 'setncattr', 'sync', 'variables', 'vltypes']


    
def pydaptonetCDF4(dataset):
    """make a pydap dataset look and quack like a netcdf4 dataset
    >>> import pydap.client
    >>> url  = 'http://opendap.deltares.nl:8080/opendap/rijkswaterstaat/jarkus/profiles/transect.nc'
    >>> ds   = pydap.client.open_url(url)
    >>> ncds = pydaptonetCDF4(ds)
    >>> type(ncds)
    <class 'pydap.model.DatasetType'>
    
    You should now be able to access the dataset in a netCDF4 way (with variables)
    >>> ncds.variables['x'] is ds['x']['x']
    True
    """
    import pydap.model
    assert isinstance(dataset, pydap.model.DatasetType)
    # in pydap the dataset itself is a dict, in netCDF4 it has a variables dict
    # let's add the variables as well
    dataset.variables = {}
    for variable in dataset.keys():
        if isinstance(dataset[variable], pydap.model.GridType):
            # the pydap client returns grids for arrays with coordinates.
            # 
            dataset.variables[variable] = dataset[variable][variable]
            dataset.variables[variable].attributes.update(dataset[variable].attributes)
        else:
            dataset.variables[variable] = dataset[variable]
    for key, value in dataset.attributes['NC_GLOBAL'].items():
        if key not in dataset:
            # I think the __setitem__ might be overwritten, so we'll do it like this
            setattr(dataset, key, value)
        else:
            warnings.warn('Could not set %s to %s because it already exists as a variable' % (key, value))
    return dataset

# TO DO
# deal with fill_values: netCDF deals with them automaticaly
# by inserting into a masked arrray, whereas pydap does not.
# numpy.ma.MaskedArray or does netCDF4 doe that based on presence of fillvalue att

# TO DO
# perhaps create a netCDF4topydap adapter also

def opendap(url):
    """return the dataset looking like a netCDF4 object
    >>> url = 'http://opendap.deltares.nl:8080/opendap/rijkswaterstaat/jarkus/profiles/transect.nc'
    >>> ds = opendap(url)
    >>> ds.variables.has_key('x')
    True
    >>> ds.title
    'Jarkus Data'
    
    """
    try:
        import netCDF4
        dataset = netCDF4.Dataset(url)
    # Either netcdf is not found, or it cannot read url's
    except (ImportError, RuntimeError):
        import pydap.client
        dataset = pydaptonetCDF4(pydap.client.open_url(url))
    return dataset

if __name__ == "__main__":
    import doctest
    doctest.testmod()