#!/usr/bin/env python
# Wenchang Yang (wenchang@princeton.edu)
# Wed Oct 17 09:45:40 EDT 2018
import os, os.path, sys, xarray as xr, numpy as np

data_name = 'sst'
units = 'degrees C'
long_name = 'Daily sea surface temperature'
dimx, dimy, dimz, dimt = 'lon', 'lat', 'zlev', 'time'
years = range(1981, 2018+1)

nc_encoding = {dimt: dict(_FillValue=None, dtype='double'),
    dimx: dict(_FillValue=None),
    dimy: dict(_FillValue=None),
    data_name: dict(_FillValue=np.nan, dtype='float32'),
}

for year in years:
    ifiles = f'raw/avhrr-only-v2.{year}*.nc'
    ofile = f'{data_name}/by_year/{data_name}.avhrr-only-v2.{year}.nc'
    if os.path.exists(ofile):
        print('[File exists]:', ofile)
        continue
    else:
        pass
    ds = xr.open_mfdataset(ifiles)
    da = ds[data_name].squeeze(dim=dimz, drop=True)
    da.attrs = dict(units=units, long_name=long_name)
    da.to_dataset(name=data_name).to_netcdf(ofile, unlimited_dims=dimt, encoding=nc_encoding) 
    print('[File created]:', ofile)