#!/usr/bin/env python
if __name__ == '__main__':
    from misc.timer import Timer
    wyt = Timer(f'start {__file__}') 
import xarray as xr
import os, os.path, sys, time
import  salem
import pandas as pd
import numpy as np
import geoxarray
from misc.shell import run_shell

cntry_name = 'China'
cntry_sname = cntry_name.replace(' ', '')
data_name = 'msdwuvrf'
dname_long = 'mean_surface_downward_uv_radiation_flux'
xname = 'longitude'
yname = 'latitude'
shdf = salem.read_shapefile('chn_admbnda_adm1_ocha/chn_admbnda_adm1_ocha.shp')
name_cols = ['ADM1_EN', 'ADM1_PCODE']
t = pd.date_range('2009-01', '2014-12', freq='MS')
years, months = t.year.values, t.month.values
print(data_name)
ofiles = []
for year,month in zip(t.year, t.month):
    print(f'{year:04d}-{month:02d}')
    ifile = f'/tigress/wenchang/data/era5/analysis/{dname_long}/daily/era5.{dname_long}.daily.{year:04d}-{month:02d}.nc'
    ofile = f'tmp/era5.{data_name}.daily.{cntry_sname}.subregion.{year:04d}-{month:02d}.nc'
    ofiles.append(ofile)
    if os.path.exists(ofile):
        print('[exists]:', ofile)
        continue
    print(ifile, 'loading...')
    ds = salem.open_xr_dataset(ifile).load()
    # transform lon [0,360) -> [-180,180)
    ds = ds.roll(longitude=ds.longitude.size//2)
    lon_new = lon = ds.longitude.where(ds.longitude<180, other=ds.longitude-360).values
    ds = ds.assign_coords(longitude=lon_new)
    
    das = []
    #region_IDs = []
    region_names = []
    N = len(shdf)
    for i in range(N):
        r = shdf.iloc[i, :]
        lon_slice = slice(r.min_x, r.max_x)
        lat_slice = slice(r.min_y, r.max_y)
        g = r.geometry
     
        try:
            da = ds[data_name].salem.roi(geometry=g).geo.fldmean()
            das.append(da)
            #region_IDs.append(int(r.GEOID))
            #region_names.append(r[name_col])
            province, pcode = r[name_cols[0]].split(' ')[0], r[name_cols[1]]#name_cols = ['ADM1_EN', 'ADM1_PCODE']
            region_name = '_'.join([province,pcode])
            region_names.append(region_name)
            print('[OK]:', i+1,'of', N, region_name)
        except:
            print('\t[Failed]', i+1,'of', N, region_name)
            pass
    
    i_record = range(len(das))
    da = xr.concat(das, pd.Index(region_names, name='region_name')).transpose()
    da.attrs = ds[data_name].attrs
    ds_new = da.to_dataset(name=data_name)
    ds_new.attrs = ds.attrs
    ds_new.attrs['region_name'] = '; '.join(region_names)
    
    #scodes = xr.DataArray(scodes, dims='i_county', coords=[i_county,])
    #ccodes = xr.DataArray(ccodes, dims='i_county', coords=[i_county,])
    #xr.Dataset(dict(precip=da, scode=scodes, ccode=ccodes)).to_netcdf(f'pr{year}.nc', unlimited_dims='time')
    ds_new.to_netcdf(ofile, unlimited_dims='time', encoding={data_name: {'_FillValue': None}})
    
    print('[saved]:', ofile)

#ncrcat
ofile = ofiles[0].replace('tmp/', '').replace(f'{years[0]}-{months[0]:02d}', f'{years[0]}-{years[-1]}')
cmd = f'ncrcat {" ".join(ofiles)} {ofile}'
run_shell(cmd)

if __name__ == '__main__':
    wyt.check('**done**')