def update_database(today):
pcrs_report = SpainCovid19MinistryReport(today, 1)
deaths_report = SpainCovid19MinistryReport(
today, 5, (142, 490, 142 + 343, 490 + 265))
hospital_report = SpainCovid19MinistryReport(
today, 3, (169, 77, 169 + 280, 77 + 707))
try:
accumulated_pcrs_today = pcrs_report.get_column_data(1)
except Exception:
# With some PDFs, tabula-pdf auto table detection fails.
# We need to specify a custom area.
pcrs_report = SpainCovid19MinistryReport(
today, 1, (239, 56, 239 + 283, 56 + 756))
accumulated_pcrs_today = pcrs_report.get_column_data(1)
accumulated_deaths_today = deaths_report.get_column_data(1)
today_percentage_admitted = hospital_report.get_column_data(3, cast=float)
today_percentage_icu = hospital_report.get_column_data(6, cast=float)
today_pcrs_last_24h = pcrs_report.get_column_data(2)
accumulated_incidence = pcrs_report.get_column_data(3, 1, float)
update_stat(Measurement.PCRS, accumulated_pcrs_today, today)
update_stat(Measurement.DEATHS, accumulated_deaths_today, today)
influx.insert_stats(Measurement.PCRS_LAST_24H, today, today_pcrs_last_24h)
influx.insert_stats(Measurement.ACCUMULATED_INCIDENCE, today,
accumulated_incidence)
influx.insert_stats(Measurement.PERCENTAGE_ADMITTED, today,
today_percentage_admitted)
influx.insert_stats(Measurement.PERCENTAGE_ICU, today,
today_percentage_icu)
def test_given_no_data_frame_acceded_when_access_then_tabula_used(
self, tabula_mock):
date = datetime(2020, 5, 5)
page = "1"
area = MagicMock()
report = SpainCovid19MinistryReport(date, page, area)
report._get_url = MagicMock()
valid_data = {
"Col1*": ["value1", "value2*", "value3", "value4*"] +
list(CCAA_POPULATION.keys())
}
d1 = DataFrame(data={"Col1": [1, 2, 3]})
d2 = DataFrame(data=valid_data)
tabula_mock.read_pdf.return_value = [d1, d2]
returned_data_frame = report.data_frame
tabula_mock.read_pdf.assert_called_once_with(
report._get_url.return_value,
pages=str(page),
area=area,
pandas_options={'dtype': str})
report._get_url.assert_called_once_with()
self.assertEqual(valid_data["Col1*"],
list(returned_data_frame['Col1']))
def test_given_cases_pdf_when_get_url_then_cases_pdf_returned(self):
date = datetime(2020, 5, 5)
report = SpainCovid19MinistryReport(date, 1)
pdf_id = 14
report.get_cases_pdf_id_for_date = MagicMock(return_value=pdf_id)
self.assertEqual(
"https://www.mscbs.gob.es/en/profesionales/saludPublica/ccayes/alertasActual/nCov-China/"
f"documentos/Actualizacion_{pdf_id}_COVID-19.pdf",
report._get_url())
def _get_hospitals_report(date):
for height in [160, 150, 170, 135, 200]:
try:
hospital_report = SpainCovid19MinistryReport(
date, 3, (height, 33, height + 260, 33 + 790))
hospital_report.get_column_data(7, cast=float)
return hospital_report
except HTTPError as e:
raise e
except Exception as e:
pass
raise Exception("Impossible to find an appropriate ")
def test_given_dataframe_when_get_column_data_then_map_returned(self):
with patch.object(SpainCovid19MinistryReport, 'data_frame'):
report = SpainCovid19MinistryReport(None, None)
headers = ["heade1", "header2", "header3"]
ccaas = list(map(lambda x: x + "*", CCAA_POPULATION.keys()))
data1 = list(map(lambda x: str(x), range(1, 20)))
data2 = list(map(lambda x: str(x) + ".000", range(21, 40)))
report.data_frame = DataFrame(
data={
"Unnamed: 0": headers + ccaas,
"Column1": headers + data1,
"Column2": headers + data2
})
result = report.get_column_data(2)
self.assertEqual(list(CCAA_POPULATION.keys()), list(result.keys()))
self.assertEqual(list(range(21000, 40000, 1000)),
list(result.values()))
def test_given_14_5_when_get_id_then_105_returned(self):
self.assertEqual(
SpainCovid19MinistryReport.get_cases_pdf_id_for_date(
datetime(2020, 5, 14)), 105)
def test_given_9_12_and_7_12_and_8_12_without_report_when_get_id_then_266_returned(
self):
self.assertEqual(
SpainCovid19MinistryReport.get_cases_pdf_id_for_date(
datetime(2020, 12, 9)), 266)
def test_given_13_7_when_get_id_then_161_returned(self):
self.assertEqual(
SpainCovid19MinistryReport.get_cases_pdf_id_for_date(
datetime(2020, 7, 13)), 161)
