def test_build_index2(mock_file):
""" Check if we can build index from string with repeated word. """
my_index = InvertedIndex()
my_index.build(mock_file)
assert my_index.index_data == {'test': [1], 'me': [1], 'first': [1]}
def test_many_queries_not_in_one_article(query):
index = InvertedIndex()
article_id = '12'
with open(ONE_ARTICLE_PATH, 'r') as fd:
index.build(fd)
assert article_id not in index.find_articles(
query), 'find article in query that not in one article'
def test_build_index(mock_file):
""" Check if we can build index from simple string. """
my_index = InvertedIndex()
my_index.build(mock_file)
assert my_index.index_data == {'test': [1], 'me': [1], 'first': [1]}
def test_query2(mock_file):
""" Check if we can query some correct words with repeats. """
my_index = InvertedIndex()
my_index.build(mock_file)
assert my_index.query(['string', 'long']) == {1, 10, 100500}
def main():
os.chdir(RELATIVE_PATH_TO_CORPUS)
docs = os.listdir(os.getcwd())
index = InvertedIndex(docs)
index.build()
QueryHandler(index).loop()
def test_build_index4(mock_file):
""" Check if we can build index from file with several lines. """
my_index = InvertedIndex()
my_index.build(mock_file)
# Check the index
assert len(my_index.index_data) == 12
assert len(my_index.index_data.values()) == 12
def test_one_article():
index = InvertedIndex()
with open(ONE_ARTICLE_PATH, 'r') as fd:
index.build(fd)
with open(ONE_ARTICLE_PATH, 'r') as fd:
article_id, words = fd.readline().split(maxsplit=1)
words = words.split()
assert article_id == index.find_articles(
words), 'didnt find article in query of all words in article'
def test_query(mock_file):
""" Check if we can query some correct words. """
my_index = InvertedIndex()
my_index.build(mock_file)
assert my_index.query(['test']) == {
1,
}
def test_build_index3(mock_file):
""" Check if we can build index from string with special characters. """
my_index = InvertedIndex()
my_index.build(mock_file)
assert my_index.index_data == {
'test': [1],
'test,': [1],
'me': [1],
'first': [1]
}
def test_dump_index(mock_file):
""" Check if we can dump index. """
my_index = InvertedIndex()
my_index.build(mock_file)
my_index.dump('my_Test.index')
expected_calls = [
"call('my_Test.index', 'wb')", "call().write(b'\\x03\\x00\\x00\\x00')",
"call().write(b'\\x00\\x02')", "call().write(b'me')",
"call().write(b'\\x00\\x01')", "call().write(b'\\x00\\x04')",
"call().write(b'test')", "call().write(b'\\x00\\x05')",
"call().write(b'first')"
]
all_calls = [str(call) for call in mock_file.mock_calls]
for call in expected_calls:
assert call in all_calls
def test_build_small_index():
index = InvertedIndex()
wiki_path = 'small_test.txt'
with open(wiki_path, 'r') as fd:
index.build(fd)
assert index.inverted_index == _SMALL_INDEX, f'wrong build index from {wiki_path}'