def test_index_hits():
src = """
Europe
Russia [url=www.russia.ru]
Moscow
Europe [info="a place in Moscow"]
England
London
"""
text = 'Europe: London, EUROPE CENTER in Moscow'
root = loads(src)
normalizer = lambda txt: ' '.join(re.compile('\w+').findall(txt.lower()))
terms_idx = Index(root, normalizer=normalizer)
hits = terms_idx.hits(text, indirect_hits=False)
assert 4 == len(hits)
assert [
Hit(root[0], dhits=2, ihits=None),
Hit(root[0, 0, 0, 0], dhits=2, ihits=None),
Hit(root[0, 1, 0], dhits=1, ihits=None),
Hit(root[0, 0, 0], dhits=1, ihits=None)
] == hits
hits = terms_idx.hits(text)
assert 6 == len(hits)
assert [
Hit(root[0], dhits=2, ihits=4),
Hit(root[0, 0, 0, 0], dhits=2, ihits=0),
Hit(root[0, 1, 0], dhits=1, ihits=0),
Hit(root[0, 0, 0], dhits=1, ihits=2),
Hit(root[0, 0], dhits=0, ihits=3),
Hit(root[0, 1], dhits=0, ihits=1),
] == hits
def test_index_normalization():
src = """Europe\n\tGermany\n\tMunich\n\tBerlin"""
text = 'munich BERLIN'
root = loads(src)
normalizer = lambda txt: txt.lower()
terms_idx = Index(root, normalizer=normalizer)
hits = terms_idx.hits(text, indirect_hits=False)
assert 2 == len(hits)
assert 'Munich' == hits[0].node.term
assert 'Berlin' == hits[1].node.term
def test_index_build():
root = loads(TEST_SRC)
terms_idx = Index(root)
assert 'A' in terms_idx
assert 'B' in terms_idx
assert 'C' in terms_idx
assert 'D' in terms_idx
assert 'E' in terms_idx
assert 'X' not in terms_idx
nodes = terms_idx.get('D')
assert 1 == len(nodes)
assert 'D' == nodes[0].term
assert None == terms_idx.get('X')
def test_base_case():
src = """
A
B
# comment 1 for C
C # comment 2 for C
D
# comment for E
E [key1=value1 "key 2"="value 2" ]
"""
root = loads(src)
assert 'A' == root[0].term
assert 'B' == root[0][0].term
assert 'B' == root[0, 0].term
assert 'C' == root[0, 0, 0].term
assert 'D' == root[0, 0, 1].term
assert 'D' == root[0, 0, 1].term
assert 'E' == root[1].term
assert {'key1': 'value1', 'key 2': 'value 2'} == root[1].data
def test_empty_content_parsing():
tree = loads('')
assert [] == tree.children
assert None == tree.term
def test_parse_src_with_invalid_data():
s = 'A [key1 = value1]' # space before or after = is not allowed
with pytest.raises(SyntaxError) as ex:
loads(s)
assert 'invalid data string format' in str(ex)