def __call__(self, **kwargs) :
# input : {axisName: condition, ...}
# standardize the axisNames
for axisName, condition in kwargs.iteritems() :
del kwargs[axisName]
if type(condition) == tuple :
condition = tuple(sorted(condition[:2]))+condition[2:]
kwargs[Axes.standardize(axisName)] = condition
output = self.extract_data(**kwargs)
# do we need to interpolate along an axis ?
for axisName, condition in kwargs.iteritems() :
# did we ask for a single value for an axis
# yet still have this axis in the output variable ?
# this signifies extract_data returned the neighbouring points
# because this single value is not on the grid
if type(condition) != tuple and axisName in output.axes :
output.data # kludge for grib files
firstSlice = [slice(None)]*len(output.shape)
secondSlice = [slice(None)]*len(output.shape)
firstSlice[output.axes.index(axisName)] = 0
secondSlice[output.axes.index(axisName)] = 1
# linear interpolation !
try :
output = \
(output[secondSlice] - output[firstSlice])/\
(output.axes[axisName].data[1] - \
output.axes[axisName].data[0])\
*(condition - output.axes[axisName].data[0]) \
+ output[firstSlice]
except IndexError :
# sometimes, due to rounding errors, extract_data will return one
# value when two were expected, thus output...[1] fails
output = output[firstSlice]
output.metadata[axisName] = condition
return output
def __init__(self, variables=None, axes=None) :
if axes == None :
self.axes = Axes()
else :
self.axes = axes
if variables == None :
self.variables = {}
else :
self.variables = variables
def mean(self, axisNames) : # input can either either be like 'zy' or like ['lev', 'lat'] # turn the 'zy' into ['z', 'y'] axisNames = list(axisNames) for i in range(len(axisNames)) : axisNames[i] = Axes.standardize(axisNames[i]) # levels must be averaged first # 'level' must be at the top of the list if axisNames[i] == 'level' : del axisNames[i] axisNames = ['level'] + axisNames return self.averager(axisNames)
def write(self, filePath) :
from netCDF4 import Dataset
# making sure the variable's axes are given
# to the file is the user's responsability
with Dataset(filePath, 'w') as output :
for axisName, axis in self.axes.iteritems() :
axisName = Axes.ncStandardize(axisName)
output.createDimension(axisName, len(axis))
if axisName == 'time' :
output.createVariable('time', int, ('time',))
output.variables['time'].units = 'seconds since 1970-1-1'
from datetime import datetime
epoch = datetime(1970, 1, 1)
output.variables['time'][:] = [
(instant-epoch).total_seconds()
for instant in axis.data]
else :
# this is a real axis with information worth recording
if axis.units != 'indices' or len(axis) - 1 != axis[-1] :
output.createVariable(
axisName,
type(np.asscalar(axis.data.ravel()[0])),
(axisName,))
output.variables[axisName][:] = axis.data
if axis.units != None :
output.variables[axisName].units = axis.units
# TODO metadata...
for variableName, variable in self.variables.iteritems() :
if variableName == '~' :
variableName = 'unknown'
output.createVariable(
variableName,
type(np.asscalar(variable.data.ravel()[0])),
tuple(
[Axes.ncStandardize(axisName) for axisName in
variable.axes.keys()]))
if 'units' in variable.metadata :
output.variables[variableName].units = variable.units
output.variables[variableName][:] = variable.data
