diff --git a/spacy/tests/pipeline/test_lemmatizer.py b/spacy/tests/pipeline/test_lemmatizer.py index 0d2d3d6e5..570683329 100644 --- a/spacy/tests/pipeline/test_lemmatizer.py +++ b/spacy/tests/pipeline/test_lemmatizer.py @@ -1,8 +1,10 @@ import pytest import pickle +import spacy from spacy import util, registry +from spacy.about import __lookups_url__ from spacy.lang.en import English -from spacy.lookups import Lookups +from spacy.lookups import Lookups, load_lookups_data_from_url from ..util import make_tempdir @@ -111,3 +113,15 @@ def test_lemmatizer_serialize(nlp): # Make sure that lemmatizer cache can be pickled pickle.dumps(lemmatizer2) + + +@pytest.mark.parametrize("lang", ("ca", "en")) +def test_lemmatizer_load_lookups_from_url(lang): + nlp = spacy.blank(lang) + lemmatizer = nlp.add_pipe("lemmatizer") + req_tables, opt_tables = lemmatizer.get_lookups_config(mode=lemmatizer.mode) + lookups = load_lookups_data_from_url( + nlp.lang, req_tables + opt_tables, __lookups_url__ + ) + lemmatizer.initialize(lookups=lookups) + assert set(lemmatizer.lookups.tables) == set(req_tables + opt_tables)