def test_no_variation_is_same_as_key(self):
"""Are any variations identical to the keys they're supposed to be
variations of?
"""
for variation, keys in VARIATIONS_ONLY.items():
for key in keys:
self.assertNotEqual(
variation, key,
"The variation '%s' is identical to the key it's supposed "
"to be a variation of." % variation)
def test_no_variation_is_same_as_key(self):
"""Are any variations identical to the keys they're supposed to be
variations of?
"""
for variation, keys in VARIATIONS_ONLY.items():
for key in keys:
self.assertNotEqual(
variation,
key,
"The variation '%s' is identical to the key it's supposed "
"to be a variation of." % variation
)
def test_extract_citations(case_factory, tmpdir, settings, elasticsearch):
from scripts.extract_cites import EDITIONS as processed_editions
settings.MISSED_CITATIONS_DIR = str(tmpdir)
blocked_by_date = set(
k for k in list(EDITIONS.keys()) + list(VARIATIONS_ONLY.keys())
if all(c['start_year'] > 2000 for c in processed_editions[k]))
legitimate_cites = [
"225 F. Supp. 552", # correct
["125 f supp 152", "125 F. Supp. 152"], # normalized
["125 Burnett (Wis.) 152", "125 Bur. 152"], # normalized
["1 F. 2d 2", "1 F.2d 2"], # not matched as "1 F. 2"
"2 1/2 Mass. 1", # special volume numbers
"3 Suppl. Mass. 2", # special volume numbers
"1 La.App. 5 Cir. 2", # not matched as "1 La.App. 5"
"2000 WL 12345", # vendor cite
]
legitimate_cites += [
"1 %s 1" % c for c in EDITIONS.keys() if c not in blocked_by_date
]
legitimate_cites += [["1 %s 1" % k, "1 %s 1" % v]
for k, vv in VARIATIONS_ONLY.items() for v in vv
if k not in blocked_by_date]
legitimate_cites_normalized = set(
normalize_cite(c if type(c) is str else c[1])
for c in legitimate_cites)
legitimate_cites = [
c if type(c) is str else c[0] for c in legitimate_cites
]
illegitimate_cites = [
"2 Dogs 3", # unrecognized reporter
"3 Dogs 4", # duplicate unrecognized reporter
"1 or 2", # not matched as 1 Or. 2
"word1 Mass. 2word", # not matched if part of larger word
"1 Mass.\n 2", # no match across newlines
"1 A.3d 1", # no match to reporter that started publishing in 2010
]
illegitimate_cites += ["1 %s 1" % c for c in blocked_by_date]
case = case_factory(
body_cache__text=", some text, ".join(legitimate_cites +
illegitimate_cites),
decision_date=datetime(2000, 1, 1))
fabfile.extract_all_citations()
update_elasticsearch_from_queue()
# check extracted cites
cites = list(ExtractedCitation.objects.all())
cite_set = set(c.cite for c in cites)
normalized_cite_set = set(c.normalized_cite for c in cites)
assert cite_set == set(legitimate_cites)
assert normalized_cite_set == legitimate_cites_normalized
assert all(c.cited_by_id == case.pk for c in cites)
