Import('*')
import pandas as pd

papers = [
    'Reimers_Waldfogel', 'Finkelstein_Gentzkow_Williams',
    'Goodman-Bacon', 'Abebe_Caria_Ortiz-Ospina',
    'Mueller_Spinnewijn_Topa', 'Kostol_Myhre',
    'Bailey_Sun_Timpe', 'Seibold', 'Dinerstein_Smith',
    'Adermon_Lindahl_Palme', 'Bourreau_Sun_Verboven', 'Weaver',
    'Braguinsky_Ohyama_Okazaki_Syverson', 'Kanzig'
]
source = ['#source/analysis/bootstrap_census/calculate_statistics.py']
df_objects = pd.read_csv('../../../source/raw/bootstrap_census/source/raw/orig/objects_of_interest.csv')
df_objects = df_objects[df_objects['can_produce_replicates'].isin(['X', 'N'])]
for paper in papers:
    indir = 'source/raw/bootstrap_census/output/derived'         
    
    source += [f'#{indir}/{paper}_2021_Replicates.csv', f'#{indir}/{paper}_2021_Estimates.csv']       

target = ['#output/analysis/bootstrap_census/bootstrap_census.csv',
          '#output/analysis/bootstrap_census/bootstrap_census.log']
env.Python(target, source)


target = ['#output/autofill/bootstrap_census.tex']
source = ['#source/analysis/bootstrap_census/make_autofills.py',
          '#source/raw/bootstrap_census/source/raw/orig/objects_of_interest.csv',
          '#output/analysis/bootstrap_census/bootstrap_census.csv',
          '#source/lib/JMSLab/autofill.py']
env.Python(target, source)


target = [
    '#output/analysis/bootstrap_census/distance_sk.eps',
    '#output/analysis/bootstrap_census/distance_sk.png'
]
source = [
    '#source/analysis/bootstrap_census/summary_plots.py',
    '#source/lib/helpers/remove_eps_dates.py',
    '#output/analysis/bootstrap_census/bootstrap_census.csv'
]
env.Python(target, source)
