try:
# framework is running
from .startup_choice import *
except ImportError as _excp:
# class is imported by itself
if (
'attempted relative import with no known parent package' in str(_excp)
or 'No module named \'omfit_classes\'' in str(_excp)
or "No module named '__main__.startup_choice'" in str(_excp)
):
from startup_choice import *
else:
raise
from omfit_classes import utils_math
from omfit_classes.utils_math import smooth, nu_conv, is_uncertain, get_array_hash, uinterp1d
from omfit_classes.sortedDict import OMFITdataset
import numpy as np
import numbers
from uncertainties import unumpy
import warnings
warnings.filterwarnings('always', category=FutureWarning, message='The Panel class is removed from pandas.*')
import xarray.core.dataset
Dataset = xarray.core.dataset.Dataset
_load = Dataset.load
import xarray.core.dataarray
DataArray = xarray.core.dataarray.DataArray
[docs]def exportDataset(data, path, complex_dim='i', *args, **kw):
r"""
Method for saving xarray Dataset to NetCDF, with support for boolean, uncertain, and complex data
Also, attributes support OMFITexpressions, lists, tuples, dicts
:param data: Dataset object to be saved
:param path: filename to save NetCDF to
:param complex_dim: str. Name of extra dimension (0,1) assigned to (real, imag) complex data.
:param \*args: arguments passed to Dataset.to_netcdf function
:param \**kw: keyword arguments passed to Dataset.to_netcdf function
:return: output from Dataset.to_netcdf function
**ORIGINAL Dataset.to_netcdf DOCUMENTATION**
"""
from omfit_classes.omfit_base import OMFITexpression, evalExpr
if os.path.dirname(path) and not os.path.exists(os.path.dirname(path)):
os.makedirs(os.path.dirname(path))
data = copy.deepcopy(data)
def clean_attrs(da):
for k, v in list(da.attrs.items()):
if '/' in k:
da.attrs[k.replace('/', '_slash_')] = v
del da.attrs[k]
k = k.replace('/', '_slash_')
if isinstance(v, OMFITexpression):
v = evalExpr(v)
da.attrs[k] = v
if isinstance(v, bool):
da.attrs[k + '__boolean'] = str(v)
del da.attrs[k]
elif isinstance(v, dict):
da.attrs[k + '__dict'] = repr(v)
del da.attrs[k]
elif isinstance(v, list):
da.attrs[k + '__list'] = repr(v)
del da.attrs[k]
elif isinstance(v, tuple):
da.attrs[k + '__list'] = repr(v)
del da.attrs[k]
elif v is None:
da.attrs[k] = ''
else:
try:
if is_uncertain(v):
da.attrs[k + '__uncertainty'] = unumpy.std_devs(v)
da.attrs[k] = unumpy.nominal_values(v)
elif np.any(np.atleast_1d(np.iscomplex(v))):
da.attrs[k + '__imaginary'] = np.imag(v)
da.attrs[k] = np.real(v)
elif hasattr(v, 'dtype') and v.dtype == np.complex: # catches complex array with all 0 imaginary
da.attrs[k + '__imaginary'] = np.imag(v)
da.attrs[k] = np.real(v)
except Exception:
pass
# Remove illegal characters from variable names
for k in list(data.variables.keys()):
if '/' in k:
data[k.replace('/', '_slash_')] = data[k]
del data[k]
# Deal with boolean and uncertain values
for k in list(data.variables.keys()):
clean_attrs(data[k])
if data[k].values.dtype.name == 'bool':
data[k + '__boolean'] = data[k].astype('u1')
del data[k]
else:
try:
if len(data[k].values) and is_uncertain(data[k].values):
data[k + '__uncertainty'] = copy.deepcopy(data[k])
data[k + '__uncertainty'].values[:] = unumpy.std_devs(data[k + '__uncertainty'].values)
data[k].values[:] = unumpy.nominal_values(data[k].values)
except Exception:
pass
clean_attrs(data)
# Deal with complex values
ida = DataArray([0, 1], coords={complex_dim: [0, 1]}, dims=(complex_dim,))
rda = DataArray([1, 0], coords={complex_dim: [0, 1]}, dims=(complex_dim,))
for k, v in list(data.data_vars.items()):
if np.any(np.iscomplex(v.values)) or (hasattr(v.values, 'dtype') and v.values.dtype == complex):
data[k] = v.real * rda + v.imag * ida
# Use the Dataset.to_netcdf method
import warnings
with warnings.catch_warnings():
import yaml
if hasattr(yaml, 'YAMLLoadWarning'):
warnings.filterwarnings('ignore', category=yaml.YAMLLoadWarning)
return data.to_netcdf(path=path, *args, **kw)
[docs]def importDataset(filename_or_obj=None, complex_dim='i', *args, **kw):
r"""
Method for loading from xarray Dataset saved as netcdf file, with support for boolean, uncertain, and complex data.
Also, attributes support OMFITexpressions, lists, tuples, dicts
:param filename_or_obj: str, file or xarray.backends.*DataStore
Strings are interpreted as a path to a netCDF file or an OpenDAP URL
and opened with python-netCDF4, unless the filename ends with .gz, in
which case the file is gunzipped and opened with scipy.io.netcdf (only
netCDF3 supported). File-like objects are opened with scipy.io.netcdf
(only netCDF3 supported).
:param complex_dim: str, name of length-2 dimension (0,1) containing (real, imag) complex data.
:param \*args: arguments passed to xarray.open_dataset function
:param \**kw: keywords arguments passed to xarray.open_dataset function
:return: xarray Dataset object containing the loaded data
**ORIGINAL xarray.open_dataset DOCUMENTATION**
"""
import xarray
# If appropriate call the original load function
if not len(args) and not len(kw) and not isinstance(filename_or_obj, str):
_load(filename_or_obj)
return
# automatically close files to avoid OS Error of too many files being open
if isinstance(filename_or_obj, str) and os.path.isfile(filename_or_obj):
if (compare_version(xarray.__version__, '0.9.2') >= 0) and (compare_version(xarray.__version__, '0.11.0') < 0):
kw.setdefault("autoclose", True)
data = xarray.open_dataset(filename_or_obj, *args, **kw)
if (compare_version(xarray.__version__, '0.9.2') >= 0) and (compare_version(xarray.__version__, '0.11.0') < 0):
# Manually trigger loading in view of `autoclose`
data.load()
def clean_attrs(da):
for k, v in list(da.attrs.items()):
if k.endswith('__list') or k.endswith('__dict') or k.endswith('__tuple'):
del da.attrs[k]
k = '__'.join(k.split('__')[:-1])
da.attrs[k] = eval(v)
elif k.endswith('__boolean'):
k = k[: -len('__boolean')]
da.attrs[k] = bool(v)
del da.attrs[k + '__boolean']
elif k.endswith('__uncertainty'):
k = k[: -len('__uncertainty')]
da.attrs[k] = unumpy.uarray(da.attrs[k], v)
del da.attrs[k + '__uncertainty']
elif k.endswith('__imaginary'):
k = k[: -len('__imaginary')]
da.attrs[k] = da.attrs[k] + 1j * da.attrs[k + '__imaginary']
del da.attrs[k + '__imaginary']
if '_slash_' in k:
da.attrs[k.replace('_slash_', '/')] = da.attrs[k]
del da.attrs[k]
# Deal with illegal characters in names, boolean, and uncertain values
for k in sorted(list(data.variables.keys())):
clean_attrs(data[k])
if k.endswith('__boolean'):
k = k[: -len('__boolean')]
data[k] = data[k + '__boolean']
data[k].values = data[k].values[:].astype(bool)
del data[k + '__boolean']
elif k.endswith('__uncertainty'):
k = k[: -len('__uncertainty')]
a = data[k].values
# For some versions of uncertainties/numpy nan uncerainties are not allowed
b = data[k + '__uncertainty'].values
b[np.where(np.isnan(b))] = 0
data[k].values = unumpy.uarray(a[:], b[:])
del data[k + '__uncertainty']
for k in list(data.variables.keys()):
if '_slash_' in k:
k = k.replace('_slash_', '/')
data[k] = data[k.replace('/', '_slash_')]
del data[k.replace('/', '_slash_')]
clean_attrs(data)
# Make a copy to avoid any open files that prevent pickling due to a thread.lock problem
newset = Dataset()
newset.attrs = copy.deepcopy(data.attrs)
for k, v in list(data.data_vars.items()):
if complex_dim in v.dims:
newdata = v.loc[{complex_dim: 0}] + 1j * v.loc[{complex_dim: 1}]
newdata.attrs = v.attrs
newdata.name = v.name
newset.update(newdata.to_dataset())
else:
newset.update(v.to_dataset())
data.close()
data = newset
# load bytes as strings
for k, v in data.variables.items():
if isinstance(v.values, bytes):
data.assign(**{k: b2s(v.values)})
for k, v in data.attrs.items():
data.attrs[k] = b2s(v)
return data
xarray.core.dataset.Dataset = Dataset
import xarray
importDataset.__doc__ += xarray.open_dataset.__doc__
exportDataset.__doc__ += Dataset.to_netcdf.__doc__
xarray.Dataset = Dataset
from xarray import *
__all__ = [
'reindex_interp',
'reindex_conv',
'split_data',
'smooth_data',
'exportDataset',
'importDataset',
'OMFITncDataset',
'OMFITncDynamicDataset',
'pandas_read_json',
'DataFrame',
]
if not np.any([('sphinx' in k and not 'sphinxcontrib' in k) for k in sys.modules]):
__all__.extend(['DataArray', 'Dataset'])
from scipy.interpolate import RegularGridInterpolator
[docs]def reindex_interp(data, method='linear', copy=True, interpolate_kws={'fill_value': np.nan}, **indexers):
r"""Conform this object onto a new set of indexes, filling in
missing values using interpolation. If only one indexer is specified,
utils.uinterp1d is used. If more than one indexer is specified,
utils.URegularGridInterpolator is used.
:params copy : bool, optional
If `copy=True`, the returned array's dataset contains only copied
variables. If `copy=False` and no reindexing is required then
original variables from this array's dataset are returned.
:params method : {'linear'}, optional
Method to use for filling index values in ``indexers`` not found on
this data array:
* linear: Linear interpolation between points
:params interpolate_kws : dict, optional
Key word arguments passed to either uinterp1d (if len(indexers)==1) or
URegularGridInterpolator.
:params \**indexers : dict
Dictionary with keys given by dimension names and values given by
arrays of coordinates tick labels. Any mis-matched coordinate values
will be filled in with NaN, and any mis-matched dimension names will
simply be ignored.
:return: Another dataset array, with new coordinates and interpolated data.
See Also:
DataArray.reindex_like
align
"""
# see if it is already handled
try:
ds = data.reindex(method=method, copy=copy, **indexers)
# xarray method converts complex arrays into object arrays
if method is None:
if isinstance(data, Dataset):
for k in list(data.data_vars.keys()):
ds[k] = ds[k].astype(data[k].dtype)
else:
ds = ds.astype(data.dtype)
return ds
except Exception:
pass
# handle Datasets for user's convenience
if isinstance(data, Dataset):
ds = data.apply(reindex_interp, keep_attrs=True, method=method, copy=copy, interpolate_kws=interpolate_kws, **indexers)
elif isinstance(data, DataArray):
# Reindexing to get the copy doesn't work if the indexers are not in the dims
if np.all([k not in data.dims for k in list(indexers.keys())]):
if copy:
return data.copy()
return data
# dumb reindexing to get the new object
ds = data.reindex(method=None, copy=copy, **indexers)
# form 1D interpolator
if len(indexers) == 1:
akey = list(indexers.keys())[0]
axis = data.dims.index(akey)
pts = data[data.dims[axis]]
values = data.values
fi = uinterp1d(pts, values, axis=axis, kind=method, copy=copy, **interpolate_kws)
newpts = indexers[akey]
# form multi-dimensional interpolator
else:
args = np.array([data[k].data for k in data.dims])
values = data.values
fi = URegularGridInterpolator(tuple(args), values, method=method, **interpolate_kws)
newpts = np.array(np.meshgrid(*[ds[k].data for k in ds.dims])).T.reshape(-1, ds.ndim)
# perform actual interpolation
ds.values = fi(newpts).reshape(ds.shape)
else:
raise TypeError('Input must be DataArray or Dataset')
return ds
[docs]def reindex_conv(data, method='gaussian', copy=True, window_sizes={}, causal=False, interpolate=False, std_dev=2, **indexers):
r"""Conform this object onto a new set of indexes, filling values along changed dimensions
using nu_conv on each in the order they are kept in the data object.
:param copy: bool, optional
If `copy=True`, the returned array's dataset contains only copied
variables. If `copy=False` and no reindexing is required then
original variables from this array's dataset are returned.
:param method: str/function, optional
Window function used in nu_conv for filling index values in ``indexers``.
:param window_sizes: dict, optional
Window size used in nu_conv along each dimension specified in indexers.
Note, no convolution is performed on dimensions not explicitly given in indexers.
:param causal: bool, optional
Passed to nu_conv, where it forces window function f(x>0)=0.
:param \**indexers: dict
Dictionary with keys given by dimension names and values given by
arrays of coordinate's tick labels. Any mis-matched coordinate values
will be filled in with NaN, and any mis-matched dimension names will
simply be ignored.
:param interpolate: False or number
Parameter indicating to interpolate data so that there are `interpolate`
number of data points within a time window
:param std_dev: str/int.
Accepted strings are 'propagate' or 'none'. Future options will include 'mean', and 'population'.
Setting to an integer will convolve the error uncertainties to the std_dev power before taking the std_dev root.
:return: DataArray
Another dataset array, with new coordinates and interpolated data.
See Also:
DataArray.reindex_like
align
"""
# handle Datasets for user's convinience
if isinstance(data, Dataset):
ds = data.apply(
reindex_conv,
keep_attrs=True,
method=method,
copy=copy,
window_sizes=window_sizes,
causal=causal,
interpolate=interpolate,
std_dev=std_dev,
**indexers,
)
elif isinstance(data, DataArray):
# Reindexing to get the copy doesn't work if the indexers are not in the dims
if np.all([k not in data.dims for k in list(indexers.keys())]):
if copy:
return data.copy()
return data
# dumb reindexing to get the new object
ds = data.reindex(method=None, copy=copy, **indexers)
values = data.values
for k, v in list(indexers.items()):
if k in data.dims:
axis = data.dims.index(k)
xi = data[k]
xo = ds[k]
ws = window_sizes.get(k, None)
values = nu_conv(
values,
xi=data[k].values,
xo=ds[k].values,
causal=causal,
window_function=method,
window_size=ws,
axis=axis,
interpolate=interpolate,
std_dev=std_dev,
)
ds.values = values
else:
raise TypeError('Input must be DataArray or Dataset')
return ds
[docs]def split_data(data, **indexers):
"""
Split the OMFITdataArray in two wherever the step in a given coordinate
exceeds the specified limit.
:param indexers: dict-like with key,value pairs corresponding to dimension labels and step size.
Example:
>> dat = OMFITdataArray([0.074,-0.69,0.32,-0.12,0.14],coords=[[1,2,5,6,7],dims=['time'],name='random_data')
>> dats=dat.split(time=2)])
>> print(dats)
"""
coords = SortedDict(indexers)
k = list(coords.keys())[0]
v = coords.pop(k)
newdata = []
start = 0
for i, d in enumerate(np.diff(data[k])):
if d > v:
temp = data.isel(**{k: slice(start, i + 1)})
if coords:
temp = temp.split(coords)
newdata.append(temp)
start = i + 1
temp = data.isel(**{k: slice(start, i + 1)})
if coords:
temp = temp.split(coords)
newdata.append(temp)
return newdata
[docs]def smooth_data(data, window_size=11, window_function='hanning', axis=0):
"""
One dimensional smoothing. Every projection of the DataArray values
in the specified dimension is passed to the OMFIT nu_conv smoothing function.
:param axis: Axis along which 1D smoothing is applied.
:type axis: int,str (if data is DataArray or Dataset)
Documentation for the smooth function is below.
"""
if isinstance(data, Dataset):
ds = data.apply(smooth, keep_attrs=True, axis=axis, window_size=window_size, window=window_function)
else:
if isinstance(axis, str):
if axis in data.dims:
axis = list(data.dims).index(axis)
else:
raise ValueError('{dim} not in dims {dims}'.format(dim=dim, dims=data.dims))
akey = data.dims[axis]
x = data[akey]
# reorder args so y (data) is first
def afun(*args, **kwargs):
args = list(args)
args.insert(0, args.pop(1))
return nu_conv(*args, **kwargs)
ds = apply_along_axis(afun, axis, data, x, x, window_size, window_function=window_function)
return ds
smooth_data.__doc__ += utils_math.nu_conv.__doc__
[docs]class OMFITncDataset(OMFITobject, OMFITdataset):
"""
Class that merges the power of Datasets with OMFIT dynamic loading of objects
"""
def __init__(self, filename, lock=False, exportDataset_kw={}, data_vars=None, coords=None, attrs=None, **kw):
r"""
:param filename: Path to file
:param lock: Prevent in memory changes to the DataArray entries contained
:param exportDataset_kw: dictionary passed to exportDataset on save
:param data_vars: see xarray.Dataset
:param coords: see xarray.Dataset
:param attrs: see xarray.Dataset
:param \**kw: arguments passed to OMFITobject
"""
if exportDataset_kw:
kw['exportDataset_kw'] = exportDataset_kw
OMFITobject.__init__(self, filename, **kw)
self.lock = lock
if data_vars or coords or attrs:
if not (os.stat(self.filename).st_size):
self.dynaLoad = False
else:
raise ValueError('Cannot specify `filename` with data and `data_vars`, or `coords` or `attrs` at the same time')
else:
self.dynaLoad = True
OMFITdataset.__init__(self, data_vars=data_vars, coords=coords, attrs=attrs)
self._loaded_hash = hash(())
self._dynamic_keys = []
def __str__(self):
return "File {:}\n{:}".format(self.filename, str(self.to_dataset()))
[docs] def set_lock(self, lock=True):
"""Make all the DataArrays immutable and disable inplace updates"""
self.lock = lock
# lock/unlock all the array data and attributes
for k, v in self.variables.items():
# assumes all values are DataArrays
v.values.flags.writeable = not lock
def __hash__(self):
"""
Return an hash representing the current state of the data
"""
hashes = [hash(frozenset(list(self._dataset.attrs.items())))] # assumes attrs is not nested (i think the netcdf does too)
for k, v in self.variables.items():
hashes.append(get_array_hash(v.values)) # assumes all items are DataArrays
hashes.append(hash(frozenset(list(v.attrs.items()))))
return hash(tuple(hashes))
[docs] @dynaLoad
def load(self):
"""
Loads the netcdf into memory using the importDataset function.
"""
# To speedup the saving, we calculate a hash of the data at load time
# so that we can verify if the data ever changed, and if not then we do not
# need to save from the OMFIT tree to NetCDF, with great speedup benefit
lock = self.lock
self.set_lock(False)
data = Dataset()
if len(self.filename) and os.path.exists(self.filename) and os.stat(self.filename).st_size:
data = importDataset(self.filename)
self._dataset.update(data)
self._dataset.attrs = data.attrs
# save a small memory footprint record of what things looked like on loading
self._loaded_hash = self.__hash__()
# let decorated methods know it doesn't need to be (re)loaded
self.dynaLoad = False
# make values immutable if was originally locked
self.set_lock(lock)
def _check_need_to_save(self):
"""
Determine if the data has changed and the file needs to be re-written.
:return: bool. True if data has changed in some way since loading or if filename changed.
"""
# check if file path has changed
if not (len(self.link) and os.path.exists(self.link) and os.stat(self.link).st_size):
return True
elif not self.__hash__() == self._loaded_hash:
# check if any of the keys, values or attrs changed
return True
return False
[docs] @dynaSave
def save(self, force_write=False, **kw):
r"""
Saves file using system move and copy commands if data in memory is unchanged,
and the exportDataset function if it has changed.
Saving NetCDF files takes much longer than loading them. Since 99% of the times NetCDF files are not edited
but just read, it makes sense to check if any changes was made before re-saving the same file from scratch.
If the files has not changed, than one can just copy the "old" file with a system copy.
:param force_write: bool. Forces the (re)writing of the file, even if the data is unchanged.
:param \**kw: keyword arguments passed to Dataset.to_netcdf function
"""
changed = self._check_need_to_save()
if force_write or changed:
if changed:
printi(
'The data has been edited. Saving {:} from scratch... '.format(self.filename.split('/')[-1])
+ 'if file is big this may take some time.'
)
else:
printi('Saving {:} from scratch... '.format(self.filename.split('/')[-1]) + 'if file is big this may take some time.')
exportDataset_kw = {}
if 'exportDataset_kw' in self.OMFITproperties:
exportDataset_kw = self.OMFITproperties['exportDataset_kw']
exportDataset_kw.update(kw)
if os.path.exists(self.filename):
os.remove(self.filename)
exportDataset(self.to_dataset(), path=self.filename, **exportDataset_kw)
else:
OMFITobject.save(self)
def __setitem__(self, item, value):
if self.lock:
raise ValueError("Cannot setitem of locked OMFITncDataset")
else:
return OMFITdataset.__setitem__(self, item, value)
class dynamic_quantity(object):
def __init__(self, obj, function_name):
self.obj = obj
self.function_name = function_name
def __call__(self, *args, **kw):
tmp = getattr(self.obj, self.function_name)(*args, **kw)
self.obj._dynamic_keys.pop(self.obj._dynamic_keys.index(self.function_name))
return tmp
def __tree_repr__(self):
return getattr(self.obj, self.function_name).__doc__.strip().split('\n')[0].strip('.'), []
[docs]class OMFITncDynamicDataset(OMFITncDataset):
def __init__(self, filename, **kw):
self.update_dynamic_keys(self.__class__)
OMFITncDataset.__init__(self, filename, **kw)
[docs] def update_dynamic_keys(self, cls):
self._dynamic_keys[:] = [x[0] for x in [x for x in inspect.getmembers(cls, predicate=inspect.ismethod) if x[0].startswith('calc_')]]
def __getitem__(self, key):
"""
Dynamically call methods if quantities are not there
"""
# show dynamic quantity in the OMFIT GUI tree
if key in self._dynamic_keys:
return dynamic_quantity(self, key)
# evaluate dynamic quantities
elif 'calc_' + key in self._dynamic_keys:
getattr(self, 'calc_' + key)()
self._dynamic_keys.pop(self._dynamic_keys.index('calc_' + key))
# return entry in the Dataset
return OMFITncDataset.__getitem__(self, key)
# def calc_test_fun(self):
# """Dummy while we develop"""
# self['test_fun'] = DataArray([1, 2, 3], dims=['x'], coords={'x': [1, 2, 3]})
[docs] def keys(self):
return np.unique(self._dynamic_keys + OMFITncDataset.keys(self))
[docs] def save(self, *args, **kw):
tmp = copy.deepcopy(self._dynamic_keys)
try:
self._dynamic_keys[:] = []
OMFITncDataset.save(self, *args, **kw)
finally:
self._dynamic_keys[:] = tmp
############################################ Monkey-patch xarray plotting
from xarray.plot import plot as xplot
def _uplot(data, ax=None, **kw):
"""
Stop uarrays from killing plots.
Uses uerrorbar for 1d then uses xarray to label everything.
Uses xarray on nominal values of 2D, etc.
"""
from matplotlib import pyplot
if ax is None:
ax = pyplot.gca()
rm = False
da_tmp = data.copy()
# replace uncertainty objects with their nominal values and use uerrorbar
if is_uncertain(data.values):
da_tmp.values = unumpy.nominal_values(data.values)
if data.ndim == 1:
x = data[data.dims[0]]
if len(x) and not isinstance(x.values[0], numbers.Number):
x = np.arange(len(x))
from omfit_classes.utils_plot import uerrorbar
pl = uerrorbar(x, data.values, ax=ax, **kw)
rm = True
# replace string coordinates with indices
for k in da_tmp.dims:
if len(da_tmp[k].values) and not isinstance(da_tmp[k].values[0], numbers.Number):
da_tmp.coords[k + '_label'] = da_tmp[k].astype('str')
da_tmp[k] = np.arange(len(da_tmp[k]))
# contour plot complex data
if len(da_tmp.dims) > 1 and np.any(np.iscomplex(da_tmp)):
da_tmp = np.abs(da_tmp)
# check if there are any pyplot figures open
nfig = len(pyplot.get_fignums())
# use the xarray plotting magic
l = xplot(darray=da_tmp.real, ax=ax, **kw)
if len(da_tmp.dims) == 1 and np.any(np.iscomplex(da_tmp)):
kwi = dict(**kw)
kwi['color'] = l[0].get_color()
kwi.pop('linestyle', '')
kwi['ls'] = '--'
l = xplot(darray=da_tmp.imag, ax=ax, **kwi)
# xarray will have opened a new Figure if we used a FigureNotebook
if nfig == 0:
pyplot.close(pyplot.figure(1))
# for string coordinates: dynamic labeling using the original sting array
for k in data.dims:
if len(data[k].values) and not isinstance(data[k].values[0], numbers.Number):
if ax.get_xlabel() == str(k):
def myformatter(x, p, da_tmp=da_tmp, data=data, k=k):
i = np.abs(da_tmp[k] - x).argmin()
return data[k].values[i]
# if not downsampling the labels, include all the labels right on their index
if hasattr(ax.xaxis.get_major_locator(), '_nbins') and len(da_tmp[k]) <= ax.xaxis.get_major_locator()._nbins:
ax.set_xticks(da_tmp[k].values)
# enables dynamic labeling for when there are too many values to have a tick at each index
ax.xaxis._funcformatter = myformatter
ax.get_xaxis().set_major_formatter(matplotlib.ticker.FuncFormatter(ax.xaxis._funcformatter))
ax.figure.canvas.draw()
if ax.get_ylabel() == str(k):
def myformatter(x, p, da_tmp=da_tmp, data=data, k=k):
i = np.abs(da_tmp[k].values - x).argmin()
return data[k].values[i]
# if not downsampling the labels, include all the labels right on their index
if hasattr(ax.yaxis.get_major_locator(), '_nbins') and len(da_tmp[k]) <= ax.yaxis.get_major_locator()._nbins:
ax.set_yticks(da_tmp[k].values)
# enables dynamic labeling for when there are too many values to have a tick at each index
ax.yaxis._funcformatter = myformatter
ax.get_yaxis().set_major_formatter(matplotlib.ticker.FuncFormatter(ax.yaxis._funcformatter))
ax.figure.canvas.draw()
# remove the xarray line on top of uerrorbar if necessary (used xarray for labeling etc.)
if rm:
ax.lines.pop()
else:
pl = l
return pl
_uplot.__doc__ = DataArray.plot.__doc__
DataArray.plot = _uplot
############################################
from pandas import read_json as pandas_read_json
from pandas import DataFrame
############################################
if '__main__' == __name__:
test_classes_main_header()
from omfit_classes.omfit_base import OMFITexpression, evalExpr
import numpy as np
filename = OMFITsrc + '/../samples/TS.nc'
nc = importDataset(filename)
# behaviour breaks between xarray 0.13.0 and 0.14.0
nc['n_e'].data = np.zeros(nc['n_e'].data.shape)
nc['n_e'].reset_coords(['ELM_phase', 'ELM_until_next', 'ELM_since_last', 'subsystem'], drop=True)
nc['n_e'] *= 2
assert np.all(~np.isnan(nc['n_e'].data))
nc['n_e'].attrs['shot'] = OMFITexpression('12345')
import pickle
pickle.dumps(nc)
tmp_dir = tempfile.mkdtemp()
exportDataset(nc, '/%s/export_dataset.nc' % tmp_dir)
os.system('rm -rf %s' % tmp_dir)
