diff --git a/tests/test_read_pos.py b/tests/test_read_pos.py new file mode 100644 index 000000000..5590679b1 --- /dev/null +++ b/tests/test_read_pos.py @@ -0,0 +1,15 @@ +import py.test +from spacy.pos_util import realign_tagged +from spacy.pos_util import _parse_line + +def test_realign(): + rules = [",", "can't", "``"] + tagged = "``/`` I/PRP ca/MD n't/RB" + assert realign_tagged(rules, tagged) == "``I/``_PRP can't/MD_RB" + + +def test_parse_line(): + line = "Pierre/NNP Vinken,/NNP_, isn't/VBZ_RB 61/CD years/NNS old./RB_.\n" + tokens, tags = _parse_line(line, '/') + assert len(tokens) == len(tags) + assert len(tokens) == 9