def test_derived_terms_pitaa():
_defns, heads = parse_enwiktionary_page("pitää", read_data("pitaa"))
found = 0
for head in heads:
if head["tag"] != "deriv":
continue
found += 1
assert 23 <= found <= 27
def test_parse_min_results(entry):
"""
Smoke test to check parsing returns a minimum number of definitions.
"""
defns, _heads = parse_enwiktionary_page(entry, read_data(entry))
got_senses = len(flat_roundtrip_senses(defns))
min_senses = MIN_LENGTHS[entry]
assert got_senses >= min_senses, "Needed {} senses for {} but got {}".format(
min_senses, entry, got_senses
)
def test_compound_fi(compound, subwords):
defns, heads = parse_enwiktionary_page(compound, read_data(compound))
found = 0
for head in heads:
if head["tag"] != "etymology-heading":
continue
assert head["ety_idx"] is None
assert len(head["etys"]) == 1
assert len(head["etys"][0]["bits"]) == len(subwords)
for bit, subword in zip(head["etys"][0]["bits"], subwords):
assert re.match(subword, bit["headword"])
found += 1
assert found == 1
def test_parse_no_exceptions(entry):
parse_enwiktionary_page(entry, read_data(entry), skip_ety=True)
def test_gram_note_has_formatting():
defns, heads = parse_enwiktionary_page("test", THING)
assert "thing" in defns["Noun"][0].cleaned_defn
assert "elative" not in defns["Noun"][0].cleaned_defn
def test_maki_not_gram_note():
defns, heads = parse_enwiktionary_page("maki", read_data("maki"))
assert (
"a relatively large, usually rounded elevation of earth"
in defns["Noun"][0].cleaned_defn
)
def test_saattaa():
defns, heads = parse_enwiktionary_page("saattaa", read_data("saattaa"))
verb_4_1 = defns["Verb"][3].subsenses[0].cleaned_defn
assert "might" in verb_4_1
assert "do, probably do" in verb_4_1
def test_pitaa_gram_rm():
defns, _heads = parse_enwiktionary_page("pitaa", read_data("pitaa"))
to_like_defn = defns["Verb"][2]
assert "like" in to_like_defn.cleaned_defn
assert "elative" not in to_like_defn.cleaned_defn
def test_vuotta_head_gram():
defns, _heads = parse_enwiktionary_page("vuotta", read_data("vuotta"))
ety1_form = defns["Etymology 1"]["Noun"][0].morph
assert ety1_form and ety1_form["case"] == "abessive"
ety2_form = defns["Etymology 2"]["Noun"][0].morph
assert ety2_form and ety2_form["case"] == "partitive"
def parse_file(filename):
defns = parse_enwiktionary_page(filename, filename.read())
if defns is None:
print("No definitions found")
pprint(defns)