Import('*')
import pandas as pd
import numpy as np

papers = ['Reimers_Waldfogel_2021', 'Finkelstein_Gentzkow_Williams_2021',
               'Goodman-Bacon_2021', 'Abebe_Caria_Ortiz-Ospina_2021',
	           'Mueller_Spinnewijn_Topa_2021', 'Kostol_Myhre_2021',
               'Bailey_Sun_Timpe_2021', 'Seibold_2021', 'Dinerstein_Smith_2021',
               'Adermon_Lindahl_Palme_2021', 'Bourreau_Sun_Verboven_2021', 'Weaver_2021',
               'Braguinsky_Ohyama_Okazaki_Syverson_2021', 'Kanzig_2021']
    

df_objects = pd.read_csv('../../../source/raw/bootstrap_census/source/raw/orig/objects_of_interest.csv')
df_objects = df_objects[np.invert((df_objects['can_produce_replicates'] != 'X').values ^ (df_objects['can_produce_replicates'] != 'N').values)]

target = []
source = ['#source/analysis/pp_plots/pp_plots.py']

for paper in papers:
    indir = 'source/raw/bootstrap_census/output/derived'            

    source += [f'#{indir}/{paper}_Replicates.csv', f'#{indir}/{paper}_Estimates.csv']        
    
    df_paper = df_objects[df_objects['citation'] == paper]
    objects = df_paper['shortname_object'].values
    for object in objects:
        target += [f'#output/analysis/pp_plots/{paper}/{object}.png',
                   f'#output/analysis/pp_plots/{paper}/{object}.eps']

env.Python(target, source)
