def test_save_to_existing_file(self):
a = pd.DataFrame({
"i": [1, 2, 3, 4],
"a": ['a1', 'a2', 'a3', None],
"b": [None, 'b2', None, 'b4']
})
a.set_index('i', inplace=True)
b = pd.DataFrame({
"i": [2, 3, 4],
"b": ['b2', 'b3', None],
"c": ['c2', None, 'c4']
})
b.set_index('i', inplace=True)
expected = merge_dfs_by_index(a, b)
a.to_csv(test_directory / 'test_files' / 'a.tsv', sep='\t', index=True)
b.to_csv(test_directory / 'test_files' / 'b.tsv', sep='\t', index=True)
expected.to_csv(test_directory / 'test_files' / 'merged_ab.tsv',
sep='\t',
index=True)
# merge a -> b
save_tsv(a, test_directory / 'test_files' / 'b.tsv')
# read merged version of b
merged_ab_read = pd.read_csv(test_directory / 'test_files' / 'b.tsv',
sep='\t',
index_col=0)
self.assertTrue(expected.equals(merged_ab_read))
def test_create_dummy_file(self):
df = pd.read_csv(test_directory / 'test_files' / 'participants.tsv',
sep='\t',
index_col=0)
path_to_save = test_directory / "test_create_dummy_file.tsv"
save_tsv(df, path_to_save)
self.assertTrue(Path(path_to_save).exists())
df_read = pd.read_csv(test_directory / "test_create_dummy_file.tsv",
sep='\t',
index_col=0)
self.assertTrue(df_read.equals(df))
def test_save_index_dtype_str(self):
a = pd.DataFrame({"i": ['1'], "a": ["a"]})
a.set_index('i', inplace=True)
expected = a
# merge a -> a
save_tsv(a, test_directory / 'test_files' / 'a.tsv')
save_tsv(a, test_directory / 'test_files' / 'a.tsv')
# read merged version of b
merged_read = pd.read_csv(test_directory / 'test_files' / 'a.tsv',
sep='\t',
index_col=0)
merged_read = merged_read.set_index(merged_read.index.astype(str))
self.assertTrue(expected.equals(merged_read))
def test_create_file(self):
df = pd.DataFrame()
path_to_save = test_directory / "test_create_file.tsv"
save_tsv(df, path_to_save)
self.assertTrue(Path(path_to_save).exists())