def test_chain_union(self):
'''Test the WordChain's union method.
'''
doc_path_list = []
for i in range(3):
file_name = 'test_data{0:d}.txt'.format((i + 1))
doc_path_list.append(os.path.join(DATA_FILE_DIR, file_name))
doc_list = (process_doc(doc_path_list[i], i + 1) for i in range(3))
iitable = build_iitable(build_sitable(doc_list))
chain = iitable['a'].union(iitable['in'])
chain_str = '(a OR in, freq:3) * --> 1 --> 2 --> 3'
self.assertEqual(str(chain), chain_str)
chain = iitable['a'].union(iitable['deepen'])
chain_str = '(a OR deepen, freq:3) * --> 1 --> 2 --> 3'
self.assertEqual(str(chain), chain_str)
chain = iitable['a'].union(WordChain('test'))
chain_str = '(a OR test, freq:3) * --> 1 --> 2 --> 3'
self.assertEqual(str(chain), chain_str)
chain = WordChain('test').union(iitable['a'])
chain_str = '(test OR a, freq:3) * --> 1 --> 2 --> 3'
self.assertEqual(str(chain), chain_str)
chain = iitable['nature'].union(iitable['a'])
chain_str = '(nature OR a, freq:3) * --> 1 --> 2 --> 3'
self.assertEqual(str(chain), chain_str)
with self.assertRaises(ValueError):
iitable['a'].union(iitable['a'])
def get_iitable():
'''Build the inverted index table.
Returns:
An iitable.
'''
doc_path_list = []
for i in range(3):
doc_path_list.append(doc_loc(i + 1))
doc_list = (process_doc(doc_path_list[i], i + 1) for i in range(3))
iitable = build_iitable(build_sitable(doc_list))
return iitable
def test_build_iitable(self):
'''Test the build_iitable function.
'''
doc_path_list = []
for i in range(3):
file_name = 'test_data{0:d}.txt'.format((i + 1))
doc_path_list.append(os.path.join(DATA_FILE_DIR, file_name))
doc_list = (process_doc(doc_path_list[i], i + 1) for i in range(3))
iitable = build_iitable(build_sitable(doc_list))
chain_a = iitable['a']
self.assertEqual('a', chain_a.word)
self.assertEqual(chain_a.freq, 3)
chain_deepen = iitable['deepen']
self.assertEqual(chain_deepen.word, 'deepen')
self.assertEqual(chain_deepen.freq, 1)