#!/usr/bin/env python
# Wenchang Yang (wenchang@princeton.edu)
# Sat May  7 12:43:43 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 misc.landmask import flagland
import geoxarray
#from misc.landmask import whereland
import xclim
#
if __name__ == '__main__':
    tt.check('end import')
def anom_diff_ts(ts, sm = 15, t_0 = 0):
    # t_0 is start of year used for accumulation of anomalies (should be in middle of dry season - default is to start from beginning of calendar year)
    # sm is number of days used to smooth the anomalies (default is 15, ie day +- 7 days)
    
    # wrap year at t_0 by shifting all values forwards
    ts = ts.shift(time = -t_0)
    
    # daily anomaly wrt annual mean
    anom = ts.groupby("time.year") - ts.groupby("time.year").mean("time")
    #print('anom', anom)
    #print()
    #anom.isel(time=slice(0,365)).plot()
       
    # cumulative daily anomalies each year
    #anom_cum = anom.groupby("time.year").cumsum()
    time_axis = anom.dims.index('time')
    anom_cum = anom.groupby("time.year").apply(np.cumsum, axis=time_axis)
    #print('anom_cum', anom_cum); print()
    #anom_cum.isel(time=slice(0,365)).plot(figsize=(6,4))
    
    # smooth the cumulative anomalies
    anom_sm = anom_cum.rolling(time = sm, center = True).mean()
    #anom_sm.isel(time=slice(0,365)).plot(figsize=(6,4))
    
    # differences of smoothed anomalies
    anom_diff = anom_sm.diff("time")
    #anom_diff.isel(time=slice(0,365)).plot(figsize=(6,4))
    
    # push the anomalies back to have the correct dates
    anom_diff = anom_diff.shift(time = t_0)
    
    return anom_diff


def rainy_season(ts, sm = 15, t_0 = 0):
        
    # get differences of smoothed anomalies
    anom_ts = anom_diff_ts(ts, sm = sm, t_0 = t_0)
    
    # count number of consecutive days of positive gradient in cumulative precip
    rs_length = xclim.indices.run_length.rle(xr.ones_like(anom_ts).where(anom_ts > 0, 0))
    #rs_length[0:365].plot(figsize=(6,4))
    
    # tag all days with length of period to which they belong
    rs_length = rs_length.ffill("time")
    #rs_length[0:365].plot(figsize=(6,4))
    
    # identify those days that are part of the longest wet period in each year
    rs = rs_length.where(rs_length.groupby("time.year") - rs_length.groupby("time.year").max() == 0)
    #rs[0:365].plot(figsize=(6,4))
        
    # convert calendar to Gregorian, matching dates (ensures that DOY is correct, regardless of model calendar)
    rs = xclim.core.calendar.convert_calendar(rs, "default", align_on = "date")
    #rs[0:365].plot(figsize=(6,4))
    
    # get start & end of rainy season
    if 'ens' in rs.dims and rs.ens.size>1:
        onsets = []
        ends = []
        for ii in rs.ens.values:
            onsets.append( rs.sel(ens=ii).time.where(rs.sel(ens=ii) > 0).dropna("time", "any").groupby("time.year").min('time').dt.dayofyear )
            ends.append( rs.sel(ens=ii).time.where(rs.sel(ens=ii) > 0).dropna("time", "any").groupby("time.year").max('time').dt.dayofyear )
        onsets = xr.concat(onsets, dim='ens')
        ends = xr.concat(ends, dim='ens')
    else:
        onsets = rs.time.where(rs > 0).dropna("time", "any").groupby("time.year").min('time').dt.dayofyear
        ends = rs.time.where(rs > 0).dropna("time", "any").groupby("time.year").max('time').dt.dayofyear
    
    # duration of rainy season
    durations = ends - onsets
    
    return onsets, ends, durations
#
#start from here
#to_degC = lambda x: (x-273.15).assign_attrs(units='degC')
#mask = lambda x: x.where(flagland(x)>0.5)
#ifile = 't_ref_max_FLOR_HistRCP45_tigercpu_intelmpi_18_576PE_5ens_1860-2100_SouthAsia.nc'
#ifile = 'precip_FLOR_HistRCP45_tigercpu_intelmpi_18_576PE_5ens_1860-2100_Pernambuco_masked.nc'
#ifile = 'precip_AM2.5C360_amipHadISSTlong_chancorr_tigercpu_intelmpi_18_1080PE_10ens_1871-2021_Pernambuco_masked.nc'
#daname = 'precip'
#units = 'mm/day'
#sel_season = lambda da: da.isel(time=(da.time.dt.month>=6)&(da.time.dt.month<=8))
#sel_season = lambda da: da.isel(time=(da.time.dt.month==6))
#daname = 't_ref' if 't_ref' in sys.argv else 't_ref_max'
daname = 'precip'
model = 'AM2.5C360' if 'AM2.5C360' in sys.argv else 'FLOR'
nens = 3 if model == 'AM2.5C360' else 10
#nens = 10
#units = '$^\circ$C'
#func_units = lambda x: (x-273.15).assign_attrs(units=units)
if model == 'FLOR':
    ifiles = 'precip_FLOR_HistRCP45_tigercpu_intelmpi_18_576PE_ens??_1860-2100_WAfrica.nc'
else:
    ifiles = 'precip_AM2.5C360_amipHadISSTrcp45_tigercpu_intelmpi_18_1080PE_ens??_1871-2050_WAfrica.nc'
def func(da):
    """onset/duration; """
    return da.load().geo.fldmean().pipe(rainy_season)
    
#ofile = ifile.replace('.nc', '_index.nc') 
ofile = ifiles.replace('ens??', f'{nens}ens').replace('.nc', f'_onsetindex.nc') 
if os.path.exists(ofile) and not 'o' in sys.argv:
    da = xr.open_dataarray(ofile)
    print('[loaded]:', ofile)
else:
    print('reading...')
    onsets, ends, durations = xr.open_mfdataset(ifiles)[daname].pipe(func)
    onsets.to_dataset(name='onset_day').to_netcdf(ofile)
    print('[saved]:', ofile)
    
    ofile_end = ofile.replace('onsetindex', 'endindex')
    ends.to_dataset(name='end_day').to_netcdf(ofile_end)
    print('[saved]:', ofile_end)
    
    ofile_duration = ofile.replace('onsetindex', 'durationindex')
    durations.to_dataset(name='duration').to_netcdf(ofile_duration)
    print('[saved]:', ofile_duration)
 
if __name__ == '__main__':
    #from wyconfig import * #my plot settings
    
    #savefig
    if len(sys.argv)>1 and 'savefig' in sys.argv[1:]:
        figname = __file__.replace('.py', f'.png')
        if 'overwritefig' in sys.argv[1:]:
            wysavefig(figname, overwritefig=True)
        else:
            wysavefig(figname)
    tt.check(f'**Done**')
    print()
    #plt.show()