def test_20_million_insert_of_invoices(random_invoices):
invoice_stats = InvoiceStats()
t_insert_20_million_rows = process_time()
invoice_stats.add_invoices(invoices=random_invoices)
elapsed_time_insert_20_million_rows = process_time() - t_insert_20_million_rows
LOGGER.info(
f"elapsed_time_insert_20_million_rows: {elapsed_time_insert_20_million_rows} seconds"
)
invoice_stats.invoice_schema.validate(invoice_stats.invoices)
t_get_median_20_million_rows = process_time()
invoice_stats.get_median()
elapsed_time_get_median_20_million_rows = process_time() - t_get_median_20_million_rows
LOGGER.info(
f"elapsed_time_get_median_20_million_rows: {elapsed_time_get_median_20_million_rows} seconds"
)
t_get_mean_20_million_rows = process_time()
invoice_stats.get_mean()
elapsed_time_get_mean_20_million_rows = process_time() - t_get_mean_20_million_rows
LOGGER.info(f"elapsed_time_get_mean_20_million_rows: {elapsed_time_get_mean_20_million_rows}")
LOGGER.info(f"\n{invoice_stats.invoice_stats.to_string()}\n")
# Check dimension of stats table
assert invoice_stats.invoice_stats.shape[0] == 1
assert invoice_stats.invoice_stats.shape[1] == 2
# Check dimension of invoices
assert invoice_stats.invoices.shape[0] == INVOICE_LIMIT
assert invoice_stats.invoices.shape[1] == 2
def test_get_mean_failure(incorrect_invoice):
invoice_stats = InvoiceStats()
invoice_stats.add_invoice(incorrect_invoice)
# force wrong value
invoice_stats.invoices["invoice_value"] = "abc"
with pytest.raises(Exception):
invoice_stats.get_mean()
def test_set_agg_value_invoice_values_failure(incorrect_invoice):
invoice_stats = InvoiceStats()
invoice_stats.add_invoice(invoice=incorrect_invoice)
# force wrong value
invoice_stats.invoices["invoice_value"] = "abc"
with pytest.raises(Exception):
invoice_stats._set_agg_value_invoice_values(row=0,
column="asdasd",
agg_type="mean")
def invoice_instance(
mock_pandas,
mock_valid_min_invoice_val,
mock_valid_max_invoice_val,
mock_valid_invoice_limit,
mock_create_invoice_schema,
mock_create_invoice_stats_schema,
) -> Generator[InvoiceStats, None, None]:
"""PyTest fixture to create invoice instance to use in tests"""
invoice_stats: InvoiceStats = InvoiceStats(
max_invoice_value=MAX_VALUE, min_invoice_value=MIN_VALUE, invoice_limit=LIMIT,
)
yield invoice_stats
def test_mean_and_median_of_invoices(random_invoices, expected_mean, expected_median):
invoice_stats = InvoiceStats()
invoice_stats.add_invoices(invoices=random_invoices)
# Check dimension of invoices
assert invoice_stats.invoices.shape[0] == len(random_invoices)
assert invoice_stats.invoices.shape[1] == 2
invoice_stats.get_mean()
invoice_stats.get_median()
# Check dimension of stats table
assert invoice_stats.invoice_stats.shape[0] == 1
assert invoice_stats.invoice_stats.shape[1] == 2
# Validate that mean and median is correct
LOGGER.info(
f"{invoice_stats.invoice_stats.values.tolist()[0]} == {[expected_mean, expected_median]}"
)
assert invoice_stats.invoice_stats.values.tolist()[0] == [expected_mean, expected_median]
def test_add_remove_add(random_invoice, random_invoices, expected_mean, expected_median):
invoice_stats = InvoiceStats()
invoice_stats.add_invoices(invoices=random_invoices)
# Check dimension of invoices
assert invoice_stats.invoices.shape[0] == len(random_invoices)
assert invoice_stats.invoices.shape[1] == 2
invoice_stats.get_mean()
invoice_stats.get_median()
# Check dimension of stats table
assert invoice_stats.invoice_stats.shape[0] == 1
assert invoice_stats.invoice_stats.shape[1] == 2
invoice_stats.add_invoice(invoice=random_invoice)
# Check dimension of invoices
assert invoice_stats.invoices.shape[0] == len(random_invoices) + 1
assert invoice_stats.invoices.shape[1] == 2
invoice_stats.clear()
# Check dimension of invoices
assert invoice_stats.invoices.empty
assert invoice_stats.invoices.empty
# Check dimension of stats table
assert invoice_stats.invoice_stats.empty
assert invoice_stats.invoice_stats.empty
invoice_stats.add_invoice(invoice=random_invoice)
# Check dimension of invoices
assert invoice_stats.invoices.shape[0] == 1
assert invoice_stats.invoices.shape[1] == 2
# Check dimension of stats table
assert invoice_stats.invoice_stats.empty
assert invoice_stats.invoice_stats.empty
invoice_stats.get_mean()
invoice_stats.get_median()
assert invoice_stats.invoice_stats.values.tolist()[0] == [999.0, 999.0]
def test_insert_failure(incorrect_invoices):
invoice_stats = InvoiceStats()
with pytest.raises(SchemaError):
invoice_stats.add_invoices(invoices=incorrect_invoices)