def get_duplicates(self, fcomp=True):
cursor = self.connection.cursor()
dupes = cursor.execute("""
SELECT GROUP_CONCAT(full_path , '|') FROM files f
JOIN hashes h ON h.hash_id = f.hash_id
GROUP BY h.hash_id
HAVING COUNT(h.hash_id) > 1
""").fetchall()
for (dupe, ) in dupes:
real_dupes = dupe.split("|")
result = DuplicateFileResult()
first = real_dupes[0]
result.add_duplicate(first)
if fcomp:
for next in real_dupes[1:]:
same = filecmp.cmp(first, next)
if not same:
result.add_diff(next)
else:
result.add_duplicate(next)
else:
for next in real_dupes[1:]:
result.add_duplicate(next)
yield result
def test_add_diff(self):
d = DuplicateFileResult()
d.add_diff("x")
assert d.get_files() == ["x"]
d.add_diff("y")
assert d.get_files() == ["x", "y"]
def test_duplicates_with_non_equal_duplicates(mocker):
runner = CliRunner()
r1 = DuplicateFileResult()
r1.add_diff("x")
i = mocker.MagicMock()
i.get_duplicates.return_value = [r1]
mocked_indexer = mocker.patch('hashdex.cli.Indexer')
mocked_indexer.return_value = i
with runner.isolated_filesystem():
result = runner.invoke(cli, ['duplicates', '--index', './index.db'])
assert 'x' in result.output
assert 'NOT EQUAL' in result.output
def test_get_duplicates(self, mocker):
connection = mocker.MagicMock()
connection\
.cursor.return_value \
.execute.return_value \
.fetchall.return_value = [("path1|path2",), (("path3|path4|path5",))]
mocker.patch("filecmp.cmp").return_value = True
r1 = DuplicateFileResult()
r1.add_duplicate("path1")
r1.add_duplicate("path2")
r2 = DuplicateFileResult()
r2.add_duplicate("path3")
r2.add_duplicate("path4")
r2.add_duplicate("path5")
indexer = Indexer(connection, mocker.Mock())
assert list(indexer.get_duplicates()) == [r1, r2]
def test_compare_duplicate_results(self):
d1 = DuplicateFileResult()
d1.add_duplicate("x")
d1.add_diff("y")
d2 = DuplicateFileResult()
d2.add_diff("x")
d2.add_duplicate("y")
assert d1 != d2
d3 = DuplicateFileResult()
d3.add_duplicate("x")
d3.add_diff("y")
d4 = DuplicateFileResult()
d4.add_diff("y")
d4.add_duplicate("x")
assert d3 == d4
def test_is_not_equal(self):
d = DuplicateFileResult()
d.add_diff("x")
assert d.is_equal() is False
def test_is_equal(self):
d = DuplicateFileResult()
d.add_duplicate("x")
assert d.is_equal() is True