def test_drop(self):
pc = PatternCreator()
pc.add("ET1", "hello")
pc.add("ET1", "there")
pc.add("ET2", "world")
pc.add("ET4", "dog")
assert pc.to_df().index.to_list() == [0, 1, 2, 3]
pc.drop([1, 2])
assert pc.to_df().index.to_list() == [0, 1]
def test_to_df(self):
pc = PatternCreator()
pc.add("ET1", "hello")
pc.add("ET1", "there")
df_1 = pc.to_df()
df_2 = pc.to_df()
df_2.loc[0, "label"] = "REPLACED_LABEL"
df_3 = pc.to_df()
assert not df_1.equals(df_2)
assert df_1.equals(df_3)
def test_overall(self, tmpdir):
tmpdir_p = pathlib.Path(str(tmpdir)) / "patterns.json"
pc = PatternCreator()
assert len(pc.to_df()) == 0
pc.add("NEW_ENTITY_TYPE", "cake")
assert len(pc.to_df()) == 1
assert set(pc.to_df().columns) == {
"label",
"attribute_0",
"value_0",
"value_type_0",
"op_0",
}
pc.add("COOL_ENTITY_TYPE", {"LEMMA": "pancake", "OP": "*"})
assert len(pc.to_df()) == 2
pc.add("SOME_ENTITY_TYPE", [{"TEXT": "good"}, {"TEXT": "pizza"}])
assert len(pc.to_df()) == 3
assert set(pc.to_df().columns) == {
"label",
"attribute_0",
"value_0",
"value_type_0",
"op_0",
"attribute_1",
"value_1",
"value_type_1",
"op_1",
}
pc.to_jsonl(tmpdir_p)
pc_loaded = PatternCreator.from_jsonl(tmpdir_p)
pc_manual = PatternCreator(storage=pc.to_df())
assert pc == pc_loaded == pc_manual