"""Alisson Bento: reproduce aggregate contrasts; optional official-source verification.
See rights.md for NIC.br source-data attribution and CC BY 4.0 terms.
"""
import argparse,csv,hashlib,io,json,pathlib,zipfile
P=pathlib.Path(__file__).resolve().parent
ap=argparse.ArgumentParser();ap.add_argument('--archive');args=ap.parse_args()
rows=list(csv.DictReader((P/'estimates.csv').open(encoding='utf-8')))
assert len(rows)==34
idx={(r['indicator'],r['group'],r['category']):r for r in rows}
if args.archive:
 import openpyxl
 raw=pathlib.Path(args.archive).read_bytes()
 expected=json.loads((P/'sources.json').read_text())['official_archive']['sha256']
 assert hashlib.sha256(raw).hexdigest()==expected,'Source archive hash differs'
 with zipfile.ZipFile(io.BytesIO(raw)) as z:
  for kind,field in [('proporcao','percent'),('margem_de_erro','margin95_pp')]:
   name=next(n for n in z.namelist() if 'tabela_'+kind+'_v' in n)
   w=openpyxl.load_workbook(io.BytesIO(z.read(name)),read_only=True,data_only=True)
   for r in rows:
    s=w[r['indicator']];s.reset_dimensions()
    assert abs(float(s[r['source_excel_column']+r['source_excel_row']].value)-float(r[field]))<1e-9
   w.close()
out=[]
for group,a,b in [('ÁREA','Urbana','Rural'),('RENDA FAMILIAR','Mais de 5 SM até 10 SM','Até 1 SM'),('RENDA FAMILIAR','Mais de 10 SM','Até 1 SM')]:
 for ind in ['H2','H3']:
  x,y=idx[ind,group,a],idx[ind,group,b]
  out.append(dict(indicator=ind,group=group,category_a=a,category_b=b,percent_a=x['percent'],margin95_a_pp=x['margin95_pp'],percent_b=y['percent'],margin95_b_pp=y['margin95_pp'],descriptive_difference_pp=round(float(x['percent'])-float(y['percent']),4),difference_inference='Not computed; joint covariance unavailable'))
with (P/'descriptive-contrasts.csv').open('w',encoding='utf-8',newline='') as f:
 w=csv.DictWriter(f,fieldnames=out[0],lineterminator='\n');w.writeheader();w.writerows(out)
print('Verified 34 aggregate rows and reproduced six descriptive contrasts. No significance tests.')