def test_chain(self):
func = StringChain(columns=["coq_word_label_1", "coq_source_genre_1"],
value=" ")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(
val.tolist(),
["abc SPOK", "abc NEWS", "abc NEWS", "x SPOK", "x NEWS"])
def test_freq(self):
df = pd.DataFrame(df0)
func = Freq(columns=[
x for x in df.columns if not x.startswith("coquery_invisible")
])
val = FunctionList([func]).lapply(df, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [1, 2, 2, 1, 1])
def test_chain(self):
func = StringChain(
columns=["coq_word_label_1", "coq_source_genre_1"],
value=" ")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(
val.tolist(),
["abc SPOK", "abc NEWS", "abc NEWS", "x SPOK", "x NEWS"])
def test_freq_with_nan1(self):
df = pd.DataFrame(df0)
df["coq_test_label_1"] = [pd.np.nan, "A", pd.np.nan, "B", pd.np.nan]
func = Freq(columns=["coq_word_label_1", "coq_test_label_1"])
val = FunctionList([func]).lapply(df, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [2, 1, 2, 1, 1])
def test_freq(self):
df = pd.DataFrame(df0)
func = Freq(columns=[x for x in df.columns
if not x.startswith("coquery_invisible")])
val = FunctionList([func]).lapply(df, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [1, 2, 2, 1, 1])
def test_match_null(self):
func = StringMatch(columns=["coq_word_label_2"], value="[a]")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(
val.tolist(), [True, True, True, True, False])
def test_length(self):
func = StringLength(columns=["coq_word_label_1"])
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [3, 3, 3, 1, 1])
def test_count(self):
func = StringCount(columns=["coq_word_label_1"], value="x")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [0, 0, 0, 1, 1])
def test_match_null(self):
func = StringMatch(columns=["coq_word_label_2"], value="[a]")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [True, True, True, True, False])
def test_length(self):
func = StringLength(columns=["coq_word_label_1"])
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [3, 3, 3, 1, 1])
def test_count(self):
func = StringCount(columns=["coq_word_label_1"], value="x")
val = FunctionList([func]).lapply(df0, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [0, 0, 0, 1, 1])
def test_freq_with_nan1(self):
df = pd.DataFrame(df0)
df["coq_test_label_1"] = [pd.np.nan, "A", pd.np.nan, "B", pd.np.nan]
func = Freq(columns=["coq_word_label_1", "coq_test_label_1"])
val = FunctionList([func]).lapply(df, session=None)[func.get_id()]
self.assertListEqual(val.tolist(), [2, 1, 2, 1, 1])