#!/usr/bin/env python
# Wenchang Yang (wenchang@princeton.edu)
# Mon Nov 18 11:56:55 EST 2024
if __name__ == '__main__':
    import sys,os
    try:
        from misc.timer import Timer
        tt = Timer(f'[{os.getcwd()}] start ' + ' '.join(sys.argv))
    except:
        pass
import sys, os.path, os, glob, datetime
import xarray as xr, numpy as np, pandas as pd, matplotlib.pyplot as plt
#more imports
wython = '/tigress/wenchang/wython'
if wython not in sys.path: sys.path.append(wython); print('added to python path:', wython)
import geoxarray
from misc import get_kws_from_argv
#
if __name__ == '__main__':
    try:
        tt.check('end import')
    except:
        pass
#
#start from here
#ifile = 'Data-added-Ingrid-Jun2025_subset.csv' 
ifile = 'Updated-data-added-Ingrid-Jun28-2025_subset.csv' #updated data
df = pd.read_csv(ifile, index_col=0)
lat = xr.DataArray(df['Latitude'], dims='location', coords=[df.location])
lon = xr.DataArray(df['Longitude'], dims='location', coords=[df.location])
lon = lon.where(lon>0, other=lon+360) #0-180 to 180 -> 0 to 360 to be consistent with ERA5

func = lambda da: da.interp(latitude=lat, longitude=lon)
dsname = get_kws_from_argv('dsname', 'skin_temperature')
if dsname == 'skin_temperature':
    daname = 'skt'
elif dsname == '2m_temperature':
    daname = 't2m'
elif dsname == '2m_temperature_max':
    daname = 'mx2t'
elif dsname == '2m_temperature_min':
    daname = 'mn2t'
else:
    daname = dsname
expname = 'era5'
dates = pd.date_range('1979-01', '2024-12', freq='MS')
ofile = f'{expname}.{daname}.daily.Ingrid.{dates.year[0]:04d}-{dates.year[-1]:04d}.nc' #era5.t2m.daily.Colorado.1979-2024.nc
print('ofile:', ofile)
if True:
    #concat over the year dim
    das = []
    for d in dates:
        year, month = d.year, d.month
        #ifile = f'/scratch/gpfs/GEOCLIM/wenchang/analysis/TC/{model}/{expname}/en{en:02d}/analysis_wy/TCIplev/{year:04d}0101.atmos_month.{dsname}.nc'
        #ifile = f'/tigress/wenchang/MODEL_OUT/{model}/{expname}/en{en:02d}/analysis_wy/TCIplev/{year:04d}0101.atmos_month.{dsname}.nc'
        #ifile = f'/projects/w/wenchang/data/era5/analysis_wy/TCI_v202201/era5.monthly.{year:04d}.{dsname}.nc'
        #/projects/w/wenchang/data/era5/analysis_wy/2m_temperature/daily/era5.2m_temperature.daily.2024-12.nc
        ifile = f'/projects/w/wenchang/data/era5/analysis_wy/{dsname}/daily/era5.{dsname}.daily.{year:04d}-{month:02d}.nc'
        print(f'{expname} year {year:04d}-{month:02d} {dsname}: {ifile}')
        ds = xr.open_dataset(ifile)
        ds = ds[[daname]] #only keep the data variable
        if 'expver' in ds.coords: ds = ds.drop('expver') #drop the coords that cause problems
        if 'valid_time' in ds.dims: 
            ds = ds.rename(valid_time='time')
            print('[renamed]: valid_time -> time')
        if 'pressure_level' in ds.dims:
            levels_new = ds.pressure_level.values.astype(int)
            ds = ds.rename(pressure_level='level').assign_coords(level=levels_new)
            print('[converted]: pressure_level -> level')
        da = ds[daname].load()
        da0 = da.isel(longitude=slice(0,1))
        lon0 = da0.longitude.values
        da0 = da0.assign_coords(longitude=lon0+360)
        da = xr.concat([da, da0], dim='longitude')
        if year == 1979 and month == 1:
            print(da.longitude.values)
        da = da.pipe(func)
        #da.isel(time=0).plot(); plt.show(); sys.exit()
        das.append(da)
    print('concat over time...')
    da = xr.concat(das, dim='time')
#    dss_ens.append(ds)
#print('concat over ens...')
#ds = xr.concat(dss_ens, dim=pd.Index(ens, name='ens'))

#saveing
print('saving...')
ds = da.to_dataset(name=daname)
ds.to_netcdf(ofile)
print('[saved]:', ofile)
        
 
 
if __name__ == '__main__':
    #from wyconfig import * #my plot settings
    
    #savefig
    if 'savefig' in sys.argv or 's' in sys.argv:
        figname = __file__.replace('.py', f'.png')
        if 'overwritefig' in sys.argv or 'o' in sys.argv:
            wysavefig(figname, overwritefig=True)
        else:
            wysavefig(figname)
 
    try:
        tt.check(f'**Done**')
    except:
        pass
    print()
    if 'notshowfig' in sys.argv or 'n' in sys.argv:
        pass
    else:
        if 'plt' in globals(): plt.show()