#!/usr/bin/env python
# Wenchang Yang (wenchang@princeton.edu)
# Wed Jun 25 04:14:42 PM EDT 2025
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)
#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.csv'
ofile = ifile.replace('.csv', '_subset.csv')
df = pd.read_csv(ifile)
"""
In [75]: df.columns
Out[75]:
Index(['Reference (authors, date, journal)',
       'Year of data collection (Final Year of Data Collection)',
       'Location (country)', 'Lower bound of age', 'Upper bound of age',
       'Number of people seropositive', 'Number of people tested', 'Latitude',
       'Longitude', 'Temp ©', 'Location (town)', 'Location (state)',
       'link (if helpful)', 'notes', 'Unnamed: 14', 'Unnamed: 15',
       'Unnamed: 16', 'Unnamed: 17', 'Unnamed: 18'],
      dtype='object')
"""
df = df.loc[~df['Longitude'].isna()] #select valid values
#create a new column named "location"
locations = []
for ii,r in df.iterrows():
    print(ii, r)
    country = r['Location (country)']
    country = country.replace(' ', '').replace('-', '')
    town = r['Location (town)']
    print(type(town))
    if type(town) != str:
        state = r['Location (state)']
        state = state.replace(' ', '').replace('-', '')
        location = f'{country}_{state}'
    else:
        town = town.replace(' ', '').replace('-', '')
        location = f'{country}_{town}'
    locations.append(location)
df['location'] = locations
#select only useful columns
df = df[['location', 'Latitude', 'Longitude']]
#drop duplicates
df = df.drop_duplicates()
#reindex
#df = df.reindex(range(df.index.size))
#save
df.to_csv(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()