def get_completions(args):
tries = corpusTries[args["corpus"]]
mode = args["trie_mode"]
if mode in ("words", "pattern"):
trie = tries[0]
prefix = args["prefix"]
else:
trie = tries[1]
prefix = tuple(args["prefix"].split())
max_results = args["max_results"]
if max_results == 0:
max_results = None
if mode == "pattern":
r = [
word for word, freq in lab.word_filter(trie, prefix)[:max_results]
]
elif args["autocorrect"] and mode == "words":
r = lab.autocorrect(trie, prefix, max_results)
else:
if mode == "sentences":
r = [
' '.join(result)
for result in lab.autocomplete(trie, prefix, max_results)
]
else:
r = lab.autocomplete(trie, prefix, max_results)
return r
def test_03_big_filter_2(self):
patterns = ('*ing', '*ing?', '****ing', '**ing**', '????', 'mon*',
'*?*?*?*', '*???')
with open(os.path.join(TEST_DIRECTORY, 'testing_data', 'frankenstein.txt'), encoding='utf-8') as f:
text = f.read()
w = lab.make_word_trie(text)
for ix, i in enumerate(patterns):
result = lab.word_filter(w, i)
expected = read_expected('frank_filter_%s.pickle' % (ix, ))
self.assertEqual(len(expected), len(result), msg='incorrect word_filter of '+repr(i))
self.assertEqual(set(expected), set(result), msg='incorrect word_filter of '+repr(i))
def test_filter_big_2():
patterns = ('*ing', '*ing?', '****ing', '**ing**', '????', 'mon*',
'*?*?*?*', '*???')
with open(os.path.join(TEST_DIRECTORY, 'testing_data', 'frankenstein.txt'),
encoding='utf-8') as f:
text = f.read()
w = lab.make_word_trie(text)
for ix, i in enumerate(patterns):
result = lab.word_filter(w, i)
expected = read_expected('frank_filter_%s.pickle' % (ix, ))
assert len(expected) == len(result), 'incorrect word_filter of %r' % i
assert set(expected) == set(result), 'incorrect word_filter of %r' % i
def test_02_big_filter_1(self):
alphabet = a = "abcdefghijklmnopqrstuvwxyz"
word_list = ["aa" + l1 + l2 + l3 + l4 for l1 in a for l2 in a for l3 in a for l4 in a]
word_list.extend(["apple", "application", "apple", "apricot", "apricot", "apple"])
word_list.append("bruteforceisbad")
trie = lab.make_word_trie(' '.join(word_list))
for i in range(20):
result = lab.word_filter(trie, "ap*")
expected = [('apple', 3), ('apricot', 2), ('application', 1)]
self.assertEqual(len(expected), len(result), msg='incorrect word_filter of ap*')
self.assertEqual(set(expected), set(result), msg='incorrect word_filter of ap*')
def test_filter_big_1():
alphabet = a = "abcdefghijklmnopqrstuvwxyz"
word_list = [
"aa" + l1 + l2 + l3 + l4 for l1 in a for l2 in a for l3 in a
for l4 in a
]
word_list.extend(
["apple", "application", "apple", "apricot", "apricot", "apple"])
word_list.append("bruteforceisbad")
trie = lab.make_word_trie(' '.join(word_list))
for i in range(1000):
result = lab.word_filter(trie, "ap*")
expected = {('apple', 3), ('apricot', 2), ('application', 1)}
assert len(expected) == len(result), 'incorrect word_filter of ap*'
assert set(expected) == set(result), 'incorrect word_filter of ap*'
def test_01_filter(self):
# Filter to select all words in trie
trie = lab.make_word_trie(
"man mat mattress map me met a man a a a map man met")
result = lab.word_filter(trie, '*')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("a", 4), ("man", 3), ("map", 2), ("mat", 1),
("mattress", 1), ("me", 1), ("met", 2)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
# All three-letter words in trie
result = lab.word_filter(trie, '???')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("man", 3), ("map", 2), ("mat", 1), ("met", 2)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
# Words beginning with 'mat'
result = lab.word_filter(trie, 'mat*')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("mat", 1), ("mattress", 1)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
# Words beginning with 'm', third letter is t
result = lab.word_filter(trie, 'm?t*')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("mat", 1), ("mattress", 1), ("met", 2)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
# Words with at least 4 letters
result = lab.word_filter(trie, '*????')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("mattress", 1)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
# All words
result = lab.word_filter(trie, '**')
self.assertIsInstance(result, list, "result not a list.")
result.sort()
expect = [("a", 4), ("man", 3), ("map", 2), ("mat", 1),
("mattress", 1), ("me", 1), ("met", 2)]
self.assertEqual(result, expect, msg="incorrect result from filter.")
def test_filter_small():
# Filter to select all words in trie
trie = lab.make_word_trie(
"man mat mattress map me met a man a a a map man met")
result = lab.word_filter(trie, '*')
assert isinstance(result, list)
result.sort()
assert result == [("a", 4), ("man", 3), ("map", 2), ("mat", 1),
("mattress", 1), ("me", 1), ("met", 2)]
# All three-letter words in trie
result = lab.word_filter(trie, '???')
assert isinstance(result, list)
result.sort()
assert result == [("man", 3), ("map", 2), ("mat", 1), ("met", 2)]
# Words beginning with 'mat'
result = lab.word_filter(trie, 'mat*')
assert isinstance(result, list)
result.sort()
assert result == [("mat", 1), ("mattress", 1)]
# Words beginning with 'm', third letter is t
result = lab.word_filter(trie, 'm?t*')
assert isinstance(result, list)
result.sort()
assert result == [("mat", 1), ("mattress", 1), ("met", 2)]
# Words with at least 4 letters
result = lab.word_filter(trie, '*????')
assert isinstance(result, list)
result.sort()
assert result == [("mattress", 1)]
# All words
result = lab.word_filter(trie, '**')
assert isinstance(result, list)
result.sort()
assert result == [("a", 4), ("man", 3), ("map", 2), ("mat", 1),
("mattress", 1), ("me", 1), ("met", 2)]
