Mercurial > repos > ecology > xarray_coords_info
diff xarray_netcdf2netcdf.py @ 0:fea8a53f8099 draft
"planemo upload for repository https://github.com/galaxyecology/tools-ecology/tree/master/tools/data_manipulation/xarray/ commit 57b6d23e3734d883e71081c78e77964d61be82ba"
author | ecology |
---|---|
date | Sun, 06 Jun 2021 08:50:43 +0000 |
parents | |
children | 3e73f657a998 |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/xarray_netcdf2netcdf.py Sun Jun 06 08:50:43 2021 +0000 @@ -0,0 +1,133 @@ +#!/usr/bin/env python3 +# +# Apply operations on selected variables +# - scale +# one can also select the range of time (for timeseries) +# to apply these operations over the range only +# when a range of time is selected and when scaling, one +# can choose to save the entire timeseries or +# the selected range only. +# when scaling, one can add additional filters on dimensions +# (typically used to filter over latitudes and longitudes) + + +import argparse +import warnings + +import xarray as xr # noqa: E402 + + +class netCDF2netCDF (): + def __init__(self, infile, varname, scale="", + output="output.netcdf", + write_all=False, + filter_list="", + verbose=False): + self.infile = infile + self.verbose = verbose + self.varname = varname + self.write_all = write_all + self.filter = filter_list + self.selection = {} + if scale == "" or scale is None: + self.scale = 1 + else: + self.scale = float(scale) + if output is None: + self.output = "output.netcdf" + else: + self.output = output + # initialization + self.dset = None + self.subset = None + if self.verbose: + print("infile: ", self.infile) + print("varname: ", self.varname) + print("filter_list: ", self.filter) + print("scale: ", self.scale) + print("write_all: ", self.write_all) + print("output: ", self.output) + + def dimension_selection(self, single_filter): + split_filter = single_filter.split('#') + dimension_varname = split_filter[0] + op = split_filter[1] + ll = int(split_filter[2]) + if (op == 'sl'): + rl = int(split_filter[3]) + self.selection[dimension_varname] = slice(ll, rl) + elif (op == 'to'): + self.selection[dimension_varname] = slice(None, ll) + elif (op == 'from'): + self.selection[dimension_varname] = slice(ll, None) + elif (op == 'is'): + self.selection[dimension_varname] = ll + + def filter_selection(self): + for single_filter in self.filter: + self.dimension_selection(single_filter) + if self.write_all: + self.ds[self.varname] = \ + self.ds[self.varname].isel(self.selection)*self.scale + else: + self.dset = \ + self.ds[self.varname].isel(self.selection)*self.scale + + def compute(self): + if self.dset is None: + self.ds = xr.open_dataset(self.infile) + if self.filter: + self.filter_selection() + if self.verbose: + print(self.selection) + elif self.write_all is not None: + self.dset = self.ds[self.varname] + + def save(self): + if self.write_all: + self.ds.to_netcdf(self.output) + else: + self.dset.to_netcdf(self.output) + + +if __name__ == '__main__': + warnings.filterwarnings("ignore") + parser = argparse.ArgumentParser() + parser.add_argument( + 'input', + help='input filename in netCDF format' + ) + parser.add_argument( + 'varname', + help='Specify which variable to plot (case sensitive)' + ) + parser.add_argument( + '--filter', + nargs="*", + help='Filter list variable#operator#value_s#value_e' + ) + parser.add_argument( + '--output', + help='Output filename to store the resulting netCDF file' + ) + parser.add_argument( + '--scale', + help='scale factor to apply to selection (float)' + ) + parser.add_argument( + "--write_all", + help="write all data to netCDF", + action="store_true") + parser.add_argument( + "-v", "--verbose", + help="switch on verbose mode", + action="store_true") + args = parser.parse_args() + + dset = netCDF2netCDF(infile=args.input, varname=args.varname, + scale=args.scale, output=args.output, + filter_list=args.filter, + write_all=args.write_all, + verbose=args.verbose) + dset.compute() + dset.save()