def test_sql_date_parsing(text_files):
connection_string = dc.get_connection_info(
os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_texts = dc.get_sql_texts(tests_folder)
result = dc.CompareDataFrame.from_sql(sql_texts['sql_use_date_prompt'],
connection_string,
params=['2014-04-01', '2014-04-04'])
expected = dc.CompareDataFrame(
{
"Nums": [1, 2, 3, 4],
"Chars": ["right", "b", "c", "d"],
"Extra": ["one", "two", "three", "four"]
},
columns=['Nums', 'Chars', 'Extra'])
result.get_member_difference(expected, to_file=True)
in_right_not_in_left = pd.read_csv('in_right_not_in_left.txt', sep='\t')
in_left_not_in_right = pd.read_csv('in_left_not_in_right.txt', sep='\t')
# Due to date filtering, expect two rows although 3 in t1
assert in_right_not_in_left.shape[0] == 2
assert in_left_not_in_right.shape[0] == 0
def test_dump_values_same(text_files):
# Test that NaN/null are equivalent as required for ETL testing
connection_string = dc.get_connection_info(
os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
sql_texts = dc.get_sql_texts(tests_folder)
result = dc.CompareDataFrame.from_sql(sql_texts['sql_lite_nulls'],
connection_string)
# Side effect of above that ints and floats may not always be equal when expected
expected = dc.CompareDataFrame(
{
'Nums': [1, 2, 3],
'NumNulls': [np.nan, 3.3, 4.7],
'SomeInt': [1.0, 2.0, 3.0]
},
columns=['Nums', 'NumNulls', 'SomeInt'])
result.create_value_comparable_lists(expected,
to_file=True,
value_precision=1)
value_difference = pd.read_csv('value_difference.txt', sep='\t')
print('The different values are: {}'.format(value_difference))
assert value_difference.shape[0] == 0
def test_sql_dump_value_difference(text_files):
connection_string = dc.get_connection_info(os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_text = "SELECT * FROM t1"
result = dc.CompareDataFrame.from_sql(sql_text, connection_string)
expected = dc.CompareDataFrame(
pd.DataFrame(
{"Nums": [1, 2, 3, 4], "Chars": ["right", "b", "c", "d"], "Extra": ["one", "two", "three", "four"]},
columns=['Nums', 'Chars', 'Extra']))
result.create_value_comparable_lists(expected, to_file=True)
value_difference = pd.read_csv('value_difference.txt', sep='\t')
assert "left | right" in value_difference['Chars_1'].values
def test_compare_value_difference(text_files):
connection_string = dc.get_connection_info(os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_text = "SELECT * FROM t1"
result = dc.CompareDataFrame.from_sql(sql_text, connection_string)
expected = dc.CompareDataFrame(
pd.DataFrame(
{"Nums": [1, 2, 3, 4], "Chars": ["right", "b", "c", "d"], "Extra": ["one", "two", "three", "four"]},
columns=['Nums', 'Chars', 'Extra']))
left, right = result.create_value_comparable_lists(expected)
assert left == right
def test_dump_values_same(text_files):
# Test that NaN/null are equivalent as required for ETL testing
connection_string = dc.get_connection_info(os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
sql_texts = dc.get_sql_texts(tests_folder)
result = dc.CompareDataFrame.from_sql(sql_texts['sql_lite_nulls'], connection_string)
# Side effect of above that ints and floats may not always be equal when expected
expected = dc.CompareDataFrame(
{'Nums': [1, 2, 3], 'NumNulls': [np.nan, 3.3, 4.7], 'SomeInt': [1.0,2.0,3.0]},
columns=['Nums', 'NumNulls','SomeInt'])
result.create_value_comparable_lists(expected, to_file=True, value_precision=1)
value_difference = pd.read_csv('value_difference.txt', sep='\t')
print('The different values are: {}'.format(value_difference))
assert value_difference.shape[0] == 0
def test_sql_date_parsing(text_files):
connection_string = dc.get_connection_info(os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_texts = dc.get_sql_texts(tests_folder)
result = dc.CompareDataFrame.from_sql(sql_texts['sql_use_date_prompt'], connection_string, params=['2014-04-01', '2014-04-04'])
expected = dc.CompareDataFrame(
{"Nums": [1, 2, 3, 4], "Chars": ["right", "b", "c", "d"], "Extra": ["one", "two", "three", "four"]},
columns=['Nums', 'Chars', 'Extra'])
result.get_member_difference(expected, to_file=True)
in_right_not_in_left = pd.read_csv('in_right_not_in_left.txt', sep='\t')
in_left_not_in_right = pd.read_csv('in_left_not_in_right.txt', sep='\t')
# Due to date filtering, expect two rows although 3 in t1
assert in_right_not_in_left.shape[0] == 2
assert in_left_not_in_right.shape[0] == 0
def test_sql_dump_value_difference(text_files):
connection_string = dc.get_connection_info(
os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_text = "SELECT * FROM t1"
result = dc.CompareDataFrame.from_sql(sql_text, connection_string)
expected = dc.CompareDataFrame(
pd.DataFrame(
{
"Nums": [1, 2, 3, 4],
"Chars": ["right", "b", "c", "d"],
"Extra": ["one", "two", "three", "four"]
},
columns=['Nums', 'Chars', 'Extra']))
result.create_value_comparable_lists(expected, to_file=True)
value_difference = pd.read_csv('value_difference.txt', sep='\t')
assert "left | right" in value_difference['Chars_1'].values
def test_compare_value_difference(text_files):
connection_string = dc.get_connection_info(
os.path.join(tests_folder, 'connection_file.ini'), 'sqlitedb')
assert connection_string == "Driver=SQLite3 ODBC Driver;Database=sqlite.db"
sql_text = "SELECT * FROM t1"
result = dc.CompareDataFrame.from_sql(sql_text, connection_string)
expected = dc.CompareDataFrame(
pd.DataFrame(
{
"Nums": [1, 2, 3, 4],
"Chars": ["right", "b", "c", "d"],
"Extra": ["one", "two", "three", "four"]
},
columns=['Nums', 'Chars', 'Extra']))
left, right = result.create_value_comparable_lists(expected)
assert left == right
# From terminal in current directory run:
# py.test -vv
# that's it!
import datacompare as dc
connection_string = dc.get_connection_info('connection_file.ini', 'salesdb')
sql_texts = dc.get_sql_texts()
result = dc.CompareDataFrame.from_sql(sql_texts['example_sales_new'],
connection_string,
params=['2015-04-01', '2015-04-02'])
expected = dc.CompareDataFrame.from_sql(sql_texts['example_sales_old'],
connection_string,
params=['2015-04-01', '2015-04-02'])
def test_with_pytest_members_are_same():
left, right = result.get_member_difference(expected)
assert left == right
def test_with_pytest_values_are_same():
left, right = result.create_value_comparable_lists(expected)
assert left == right
