def test_content():
deal = salvador_extractor.gazetteDeal(test_deal)
expected = test_deal
result = deal.filetext
assert result == expected
def test_value():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_value()
expected = "4.375.000,00"
result = deal.date
assert result == expected
def test_date():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_date()
expected = "25/05/2020"
result = deal.date
assert result == expected
def test_company_id():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_company_id()
expected = "00.740.696/0001-92"
result = deal.company_id
assert result == expected
def test_company():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_company()
expected = "PMH PRODUTOS MÉDICOS HOSPITALARES LTDA"
result = deal.company
assert result == expected
def test_process():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_process()
expected = "6567/2020"
result = deal.process
assert result == expected
def test_object():
deal = salvador_extractor.gazetteDeal(test_deal)
deal.get_object()
expected = """Aquisição de Kit para Laboratório: Teste rápido IGM/IGG para coronavirus, para garantir
-o atendimento do Laboratório Central / SMS, no combate ao COVID-19, conforme CI DGAS / LAB.
-CENTRAL Nº 033/2020
-"""
result = deal.object
assert result == expected
import pandas as pd
deals = {}
with open("../data/teste_rapido_linear/1306.txt") as f:
gazette = f.read()
pattern = r'(DISPENSA DE LICITAÇÃO [\w\W]*? DATA DO ATO: [0-9/]+)'
text = gazette
match = re.findall(pattern, text, )
df = pd.DataFrame()
for i in match:
try:
deal = salvador_extractor.gazetteDeal(i)
deal.get_process()
print(deal.process)
deal.get_company()
print(deal.company)
deal.get_company_id()
print(deal.company_id)
deal.get_object()
print(deal.object)
deal.get_date()
deal_dict = deal.__dict__
del deal_dict['filetext']
df = df.append(deal_dict,ignore_index=True )
except Exception as e: print(e)
df.to_csv("temp.csv")