def test_indexer_can_add_file(self, mocker):
connection = mocker.Mock()
c1 = mocker.Mock()
connection.execute.return_value = c1
c1.fetchone.return_value = (1,)
cursor = mocker.MagicMock()
connection.cursor.return_value = cursor
hasher = mocker.Mock()
hasher.get_hashes.return_value = ("hash1", "hash2")
indexer = Indexer(connection, hasher)
f = File('~/test.txt', 'test.txt')
indexer.add_file(f)
cursor.execute.assert_has_calls([
mocker.call("INSERT OR IGNORE INTO hashes (sha1_hash, md5_hash) VALUES (?,?)", ("hash1", "hash2")),
mocker.call("INSERT OR IGNORE INTO files (hash_id, full_path, filename, mimetype, size) VALUES (?,?,?,?,?)",
(1, '~/test.txt', 'test.txt', '', 0)),
])
assert connection.commit.called is True
def test_hashes_parts_of_big_files(self, mocker):
if six.PY3:
mocked_open = mocker.patch("builtins.open")
else:
mocked_open = mocker.patch("__builtin__.open")
file_mock = mocker.MagicMock()
mocked_open.return_value = file_mock
mocked_stat = mocker.patch("os.stat")
hasher = Hasher()
hasher.BYTE_COUNT = 6
mocked_stat.return_value = DummyStatResult(hasher.BYTE_COUNT + 1)
# mock a file with "abcdefg" as content
file_mock.__enter__.return_value = file_mock
file_mock.read.side_effect = [b"abc", b"efg"]
f = File('~/test.txt', 'test.txt')
hashes = hasher.get_hashes(f)
assert hashes == (sha1(b"abcefg").hexdigest(),
md5(b"abcefg").hexdigest())
def test_delete_file_fails(self, mocker):
connection = mocker.MagicMock()
connection \
.cursor.return_value \
.execute.side_effect = sqlite3.Error()
indexer = Indexer(connection, mocker.Mock())
assert indexer.delete(File("/full/path.log", "path.log")) is False
def test_cleanup_old_files(mocker):
runner = CliRunner()
i = mocker.MagicMock()
i.get_files.return_value = [
File("./existing.txt", "existing.txt"),
File("./non-existing.txt", "non-existing.txt")
]
mocked_indexer = mocker.patch('hashdex.cli.Indexer')
mocked_indexer.return_value = i
with runner.isolated_filesystem():
with open('./existing.txt', 'w') as df:
df.write("a" * 10000)
result = runner.invoke(cli, ['cleanup', '--index', './index.db'])
assert 'Deleted ./non-existing.txt' in result.output
def test_fetch_non_existing_file(self, mocker):
connection = mocker.MagicMock()
connection.cursor.return_value.execute.return_value.fetchone.return_value = None
hasher = mocker.Mock()
hasher.get_hashes.return_value = ("hash1", "hash2")
f = File("x", "y")
indexer = Indexer(connection, hasher)
assert indexer.fetch_indexed_file(f) is None
def test_in_index(self, mocker):
connection = mocker.Mock()
connection.execute.return_value.fetchone.return_value = ("x",)
hasher = mocker.Mock()
hasher.get_hashes.return_value = ("hash1", "hash2")
f = File("x", "y")
indexer = Indexer(connection, hasher)
assert indexer.in_index(f) is True
def test_get_files(self, mocker):
connection = mocker.MagicMock()
connection \
.cursor.return_value \
.execute.return_value \
.fetchmany.side_effect = [
[("/full/path.log", "path.log"), ("/full/path2.log", "path2.log")],
None
]
indexer = Indexer(connection, mocker.Mock())
assert list(indexer.get_files()) == [File("/full/path.log", "path.log"), ("/full/path2.log", "path2.log")]
def test_logging_of_db_exception_on_file_add(self, mocker):
connection = mocker.Mock()
connection.execute.return_value = mocker.Mock()
cursor = mocker.MagicMock()
cursor.execute.side_effect = sqlite3.Error()
connection.cursor.return_value = cursor
hasher = mocker.Mock()
hasher.get_hashes.return_value = ("hash1", "hash2")
indexer = Indexer(connection, hasher)
f = File('~/test.txt', 'test.txt')
indexer.add_file(f)
assert connection.rollback.called is True
def test_check_without_rm(mocker):
f = File("./x.txt", 'x.txt')
i = mocker.MagicMock()
i.fetch_indexed_file.return_value = f
mocked_indexer = mocker.patch('hashdex.cli.Indexer')
mocked_indexer.return_value = i
runner = CliRunner()
with runner.isolated_filesystem():
os.mkdir("output")
os.mkdir("input")
with open('./input/x.txt', 'w') as df:
df.write("a" * 10000)
result = runner.invoke(cli, ['check', './input', '--index', './index.db'])
assert os.path.exists('./input/x.txt') is True
assert f.full_path in result.output
def test_hasher_hashes_file_content(self, mocker):
if six.PY3:
mocked_open = mocker.patch("builtins.open")
else:
mocked_open = mocker.patch("__builtin__.open")
file_mock = mocker.MagicMock()
mocked_open.return_value = file_mock
mocked_stat = mocker.patch("os.stat")
mocked_stat.return_value = DummyStatResult(Hasher.BYTE_COUNT - 1)
file_mock.__enter__.return_value = file_mock
file_mock.read.return_value = b"content"
f = File('~/test.txt', 'test.txt')
hasher = Hasher()
hashes = hasher.get_hashes(f)
assert hashes == (sha1(b"content").hexdigest(), md5(b"content").hexdigest())
def test_delete_file_from_index(self, mocker):
connection = mocker.MagicMock()
indexer = Indexer(connection, mocker.Mock())
assert indexer.delete(File("/full/path.log", "path.log")) is True
