#!/usr/bin/env python
# Wenchang Yang (wenchang@princeton.edu)
# Wed Oct  5 11:09:59 EDT 2022
if __name__ == '__main__':
    import sys
    from misc.timer import Timer
    tt = Timer('start ' + ' '.join(sys.argv))
import sys, os.path, os, glob, datetime
import xarray as xr, numpy as np, pandas as pd, matplotlib.pyplot as plt
#more imports
from modelout import get_modelout_data, update_modelout_data
from modelout.getdata import funcs
import xfilter
nwindow, dimlp = 9, 'year'
#lowpass = lambda x: x.filter.lowpass(1/nwindow, dim=dimlp, padtype='odd')
lowpass = lambda x: x.rolling(year=nwindow, center=True, min_periods=1).mean() if x.year.size>9 else x
import geoxarray
from misc import get_kws_from_argv
import xesmf as xe
from misc.landmask import whereocean
#
if __name__ == '__main__':
    tt.check('end import')
#
#start from here
daname = get_kws_from_argv('daname', 't_surf')
#func = lambda x: x.load().geo.fldmean()
#funcname = 'glbmean'
#func = funcs['nino34']
#funcname = 'nino34'
func = funcs['annualmean']
funcname = 'annualmean'

cleanup = False #remove cache or not

das = []
labels = []

model = 'FLOR'
#ctl1990_v201905
label = 'FLOR_ctl_1990_v201905_CTL1990'
expname = 'CTL1990_v201905_tigercpu_intelmpi_18_576PE'
da = get_modelout_data(daname=daname, model=model, expname=expname, func=func, funcname=funcname, years=range(1901,2000+1))
labels.append(label)
das.append(da)
#ctl1990_v201905_FA
label = 'FLOR_ctl_1990_v201905_FA'
expname = 'CTL1990_v201905_FA_tigercpu_intelmpi_18_576PE'
da = update_modelout_data(daname=daname, model=model, expname=expname, func=func, funcname=funcname)#, years=range(100,201))
labels.append(label)
das.append(da)
#ctl1990_v201905_FA05
label = 'FLOR_ctl_1990_v201905_FA05'
expname = 'CTL1990_v201905_FA05_tigercpu_intelmpi_18_576PE'
da = update_modelout_data(daname=daname, model=model, expname=expname, func=func, funcname=funcname)#, years=range(100,201))
labels.append(label)
das.append(da)
#ctl1990_v201905_FA05trop
label = 'FLOR_ctl_1990_v201905_FA05trop'
expname = 'CTL1990_v201905_FA05trop_tigercpu_intelmpi_18_576PE'
da = update_modelout_data(daname=daname, model=model, expname=expname, func=func, funcname=funcname)#, years=range(100,201))
labels.append(label)
das.append(da)
#ctl1990_v201905_FAtrop
label = 'FLOR_ctl_1990_v201905_FAtrop'
expname = 'CTL1990_v201905_FAtrop_tigercpu_intelmpi_18_576PE'
da = update_modelout_data(daname=daname, model=model, expname=expname, func=func, funcname=funcname)#, years=range(100,201))
labels.append(label)
das.append(da)

#units
units = da.attrs['units']

#obs
if daname == 't_surf':
    ifile = '/tigress/wenchang/modelInput/HadISST/HadISST_sst.187001-202403.wy.nc'
    da = xr.open_dataarray(ifile).sel(time=slice('1980', '2000')).mean('time')
    regridder = xe.Regridder(da, das[0], 'bilinear')
    da0 = regridder(da)
    obs_name = 'HadISST'
elif daname == 'precip':
    #ifile = '/tigress/wenchang/data/gpcp/v2p3/precip.mon.mean.197901_201808.nc'
    ifile = '/tigress/wenchang/data/cmap/precip.mon.mean.nc'
    da = xr.open_dataset(ifile)['precip'].sel(time=slice('1980', '2000')).mean('time')
    regridder = xe.Regridder(da, das[0], 'bilinear')
    da0 = regridder(da)
    obs_name = 'CMAP'
elif daname == 'slp':
    ifiles = [f'/tigress/wenchang/data/era5/raw/monthly/mean_sea_level_pressure/era5.mean_sea_level_pressure.monthly.{year}.nc'
        for year in range(1980,2000+1)]
    da = xr.open_mfdataset(ifiles)['msl'].sel(time=slice('1980', '2000')).mean('time') / 100 #Pa -> hPa
    regridder = xe.Regridder(da, das[0], 'bilinear')
    da0 = regridder(da)
    obs_name = 'ERA5'


 
def wyplot(da, label, ax=None, **kws):
    if ax is None: ax = plt.subplots()[1]
    #da = da.groupby('time.year').mean('time')
    daa = da.mean('time') - da0 #anomal from obs
    if daname == 't_surf':
        daa = daa.pipe(whereocean)
    #da = da.pipe(lowpass)
    #da = da.sel(year=slice(1900,2000))
    daa.assign_attrs(units=units).plot(ax=ax, **kws)

    
 
if __name__ == '__main__':
    from wyconfig import * #my plot settings
    from geoplots import mapplot
    if daname == 't_surf':
        vmax = 4
    elif daname == 'slp':
        vmax = 3
    elif daname == 'precip':
        vmax = 2 
    else:
        vmax = None
    if daname == 'precip':
        cmap = 'BrBG'
    else:
        cmap = None
    for da,label in zip(das, labels):
        fig,ax = plt.subplots()
        year_start = da.time[0].item().year
        year_stop = da.time[-1].item().year
        wyplot(da, label, ax=ax, robust=True, levels=21, vmax=vmax, cmap=cmap)
        plt.sca(ax)
        mapplot()

        ax.set_title(f"{label.split('_')[-1]} {daname} bias from {obs_name}: {year_start}-{year_stop}")

        #savefig
        if 'savefig' in sys.argv or 's' in sys.argv:
            figname = __file__.replace('.py', f'__{daname}_{label.split("_")[-1]}.png')
            if 'overwritefig' in sys.argv or 'o' in sys.argv:
                wysavefig(figname, overwritefig=True)
            else:
                wysavefig(figname)
    tt.check(f'**Done**')
    print()
    if 'notshowfig' in sys.argv:
        pass
    else:
        plt.show()