def main():
dbfile = sys.argv[1]
with open(dbfile, 'rb') as f:
passages = pickle.load(f)
nouns = set()
for p in passages:
sc = scenes.extract_possible_scenes(p)
heads = [scenes.extract_head(x) for x in sc]
heads = [x for x in heads if x is not None]
nouns.update(scenes.filter_noun_heads(heads))
print('\n'.join(nouns))
def run_file(path, eng, stats):
"""Site XML file ==> prints list of sceneness results"""
with open(path) as f:
root = ETree.ElementTree().parse(f)
passage = convert.from_site(root)
sc = scenes.extract_possible_scenes(passage)
heads = [scenes.extract_head(x) for x in sc]
for s, h in zip(sc, heads):
if h is None:
stats.heads.append(Result(s))
continue
out = eng.get_categories(s, h)
if out == 'implicit':
stats.heads.append(Result(s))
elif out == 'no base form':
stats.lemmas.append(Result(s, h))
elif out[2]:
stats.fulls.append(Result(s, h, *out))
else:
stats.no_cats.append(Result(s, h, *out))
def test_extract_head(self):
"""Tests that the API isn't broken, not validity of the result."""
passage = TestUtil.create_passage()
for x in scenes.extract_possible_scenes(passage):
scenes.extract_head(x)
def test_possible_scenes(self):
"""Tests that the API isn't broken, not validity of the result."""
elem = TestUtil.load_xml('test_files/site3.xml')
passage = convert.from_site(elem)
scenes.extract_possible_scenes(passage)
def test_extract_head(self):
"""Tests that the API isn't broken, not validity of the result."""
passage = TestUtil.create_passage()
for x in scenes.extract_possible_scenes(passage):
scenes.extract_head(x)
def test_possible_scenes(self):
"""Tests that the API isn't broken, not validity of the result."""
elem = TestUtil.load_xml('test_files/site3.xml')
passage = convert.from_site(elem)
scenes.extract_possible_scenes(passage)
