def test_cf_safe_name(self):
from pyaxiom.netcdf.utils import cf_safe_name
self.assertEquals('foo', cf_safe_name('foo'))
self.assertEquals('v_1foo', cf_safe_name('1foo'))
self.assertEquals('v_1foo_99', cf_safe_name('1foo-99'))
self.assertEquals('foo_99', cf_safe_name('foo-99'))
self.assertEquals('foo_99_', cf_safe_name('foo(99)'))
self.assertEquals('v__foo_99_', cf_safe_name('_foo(99)'))
def test_cf_safe_name(self):
from pyaxiom.netcdf.utils import cf_safe_name
self.assertEqual('foo', cf_safe_name('foo'))
self.assertEqual('v_1foo', cf_safe_name('1foo'))
self.assertEqual('v_1foo_99', cf_safe_name('1foo-99'))
self.assertEqual('foo_99', cf_safe_name('foo-99'))
self.assertEqual('foo_99_', cf_safe_name('foo(99)'))
self.assertEqual('v__foo_99_', cf_safe_name('_foo(99)'))
def from_dataframe(cls, df, output, **kwargs):
reserved_columns = ['trajectory', 't', 'x', 'y', 'z', 'distance']
data_columns = [ d for d in df.columns if d not in reserved_columns ]
with IncompleteMultidimensionalTrajectory(output, 'w') as nc:
trajectory_group = df.groupby('trajectory')
max_obs = trajectory_group.size().max()
unique_trajectories = df.trajectory.unique()
nc.createDimension('trajectory', unique_trajectories.size)
nc.createDimension('obs', max_obs)
# Metadata variables
nc.createVariable('crs', 'i4')
trajectory = nc.createVariable('trajectory', get_dtype(df.trajectory), ('trajectory',))
# Create all of the variables
time = nc.createVariable('time', 'i4', ('trajectory', 'obs'), fill_value=int(cls.default_fill_value))
z = nc.createVariable('z', get_dtype(df.z), ('trajectory', 'obs'), fill_value=df.z.dtype.type(cls.default_fill_value))
latitude = nc.createVariable('latitude', get_dtype(df.y), ('trajectory', 'obs'), fill_value=df.y.dtype.type(cls.default_fill_value))
longitude = nc.createVariable('longitude', get_dtype(df.x), ('trajectory', 'obs'), fill_value=df.x.dtype.type(cls.default_fill_value))
if 'distance' in df:
distance = nc.createVariable('distance', get_dtype(df.distance), ('trajectory', 'obs'), fill_value=df.distance.dtype.type(cls.default_fill_value))
attributes = dict_update(nc.nc_attributes(), kwargs.pop('attributes', {}))
for i, (uid, gdf) in enumerate(trajectory_group):
trajectory[i] = uid
# tolist() converts to a python datetime object without timezone
g = gdf.t.fillna(999999).tolist() # 999999 is a dummy value
NaTs = gdf.t.isnull()
timenums = np.ma.MaskedArray(nc4.date2num(g, units=cls.default_time_unit))
timenums.mask = NaTs
time[i, :] = timenums
latitude[i, :] = gdf.y.fillna(latitude._FillValue).values
longitude[i, :] = gdf.x.fillna(longitude._FillValue).values
z[i, :] = gdf.z.fillna(z._FillValue).values
if 'distance' in gdf:
distance[i, :] = gdf.distance.fillna(distance._FillValue).values
for c in data_columns:
# Create variable if it doesn't exist
var_name = cf_safe_name(c)
if var_name not in nc.variables:
if np.issubdtype(gdf[c].dtype, 'S') or gdf[c].dtype == object:
# AttributeError: cannot set _FillValue attribute for VLEN or compound variable
v = nc.createVariable(var_name, get_dtype(gdf[c]), ('trajectory', 'obs'))
else:
v = nc.createVariable(var_name, get_dtype(gdf[c]), ('trajectory', 'obs'), fill_value=gdf[c].dtype.type(cls.default_fill_value))
if var_name not in attributes:
attributes[var_name] = {}
attributes[var_name] = dict_update(attributes[var_name], {
'coordinates' : 'time latitude longitude z',
})
else:
v = nc.variables[var_name]
if hasattr(v, '_FillValue'):
vvalues = gdf[c].fillna(v._FillValue).values
else:
# Use an empty string... better than nothing!
vvalues = gdf[c].fillna('').values
sl = slice(0, vvalues.size)
v[i, sl] = vvalues
# Set global attributes
nc.update_attributes(attributes)
return IncompleteMultidimensionalTrajectory(output, **kwargs)
def from_dataframe(cls, df, output, **kwargs):
reserved_columns = ['trajectory', 'profile', 't', 'x', 'y', 'z', 'distance']
data_columns = [ d for d in df.columns if d not in reserved_columns ]
with IncompleteMultidimensionalProfile(output, 'w') as nc:
profile_group = df.groupby('profile')
max_zs = profile_group.size().max()
unique_profiles = df.profile.unique()
nc.createDimension('profile', unique_profiles.size)
nc.createDimension('z', max_zs)
# Metadata variables
nc.createVariable('crs', 'i4')
profile = nc.createVariable('profile', get_dtype(df.profile), ('profile',))
# Create all of the variables
time = nc.createVariable('time', 'i4', ('profile',))
latitude = nc.createVariable('latitude', get_dtype(df.y), ('profile',))
longitude = nc.createVariable('longitude', get_dtype(df.x), ('profile',))
if 'distance' in df:
distance = nc.createVariable('distance', get_dtype(df.distance), ('profile',))
z = nc.createVariable('z', get_dtype(df.z), ('profile', 'z'), fill_value=df.z.dtype.type(cls.default_fill_value))
attributes = dict_update(nc.nc_attributes(), kwargs.pop('attributes', {}))
for i, (uid, pdf) in enumerate(profile_group):
profile[i] = uid
time[i] = nc4.date2num(pdf.t.iloc[0], units=cls.default_time_unit)
latitude[i] = pdf.y.iloc[0]
longitude[i] = pdf.x.iloc[0]
if 'distance' in pdf:
distance[i] = pdf.distance.iloc[0]
zvalues = pdf.z.fillna(z._FillValue).values
sl = slice(0, zvalues.size)
z[i, sl] = zvalues
for c in data_columns:
# Create variable if it doesn't exist
var_name = cf_safe_name(c)
if var_name not in nc.variables:
if np.issubdtype(pdf[c].dtype, 'S') or pdf[c].dtype == object:
# AttributeError: cannot set _FillValue attribute for VLEN or compound variable
v = nc.createVariable(var_name, get_dtype(pdf[c]), ('profile', 'z'))
else:
v = nc.createVariable(var_name, get_dtype(pdf[c]), ('profile', 'z'), fill_value=pdf[c].dtype.type(cls.default_fill_value))
if var_name not in attributes:
attributes[var_name] = {}
attributes[var_name] = dict_update(attributes[var_name], {
'coordinates' : 'time latitude longitude z',
})
else:
v = nc.variables[var_name]
if hasattr(v, '_FillValue'):
vvalues = pdf[c].fillna(v._FillValue).values
else:
# Use an empty string... better than nothing!
vvalues = pdf[c].fillna('').values
sl = slice(0, vvalues.size)
v[i, sl] = vvalues
# Set global attributes
nc.update_attributes(attributes)
return IncompleteMultidimensionalProfile(output, **kwargs)
