def test_encode_images_verbose_false(capsys):
phasher = PHash(verbose=False)
phasher.encode_images(image_dir=PATH_IMAGE_DIR)
out, err = capsys.readouterr()
assert '' == err
assert '' == out
def test_find_duplicates_to_remove_dir():
phasher = PHash()
removal_list = phasher.find_duplicates_to_remove(image_dir=PATH_IMAGE_DIR,
max_distance_threshold=10)
assert isinstance(removal_list, list)
assert (removal_list == ['ukbench00120.jpg']
or removal_list == ['ukbench00120_resize.jpg'])
def test__find_duplicates_dict_outfile_none(mocker):
encoding_map = {'1.jpg': '123456'}
threshold = 10
scores = True
outfile = None
verbose = False
myhasher = PHash(verbose=verbose)
hasheval_mocker = mocker.patch('imagededup.methods.hashing.HashEval')
save_json_mocker = mocker.patch('imagededup.methods.hashing.save_json')
myhasher._find_duplicates_dict(
encoding_map=encoding_map,
max_distance_threshold=threshold,
scores=scores,
outfile=outfile,
)
hasheval_mocker.assert_called_with(
test=encoding_map,
queries=encoding_map,
distance_function=Hashing.hamming_distance,
verbose=verbose,
threshold=threshold,
search_method='brute_force_cython',
)
hasheval_mocker.return_value.retrieve_results.assert_called_once_with(
scores=scores)
save_json_mocker.assert_not_called()
def test_find_duplicates_correctness():
phasher = PHash()
duplicate_dict = phasher.find_duplicates(image_dir=PATH_IMAGE_DIR,
max_distance_threshold=10)
assert isinstance(duplicate_dict, dict)
assert isinstance(list(duplicate_dict.values())[0], list)
assert len(duplicate_dict['ukbench09268.jpg']) == 0
assert duplicate_dict['ukbench00120.jpg'] == ['ukbench00120_resize.jpg']
def test_find_duplicates_verbose_false(capsys):
phasher = PHash(verbose=False)
phasher.find_duplicates(
image_dir=PATH_IMAGE_DIR, max_distance_threshold=10, scores=False, outfile=False
)
out, err = capsys.readouterr()
assert '' == out
assert '' == err
def test_find_duplicates_correctness_score():
phasher = PHash()
duplicate_dict = phasher.find_duplicates(image_dir=PATH_IMAGE_DIR,
max_distance_threshold=10,
scores=True)
assert isinstance(duplicate_dict, dict)
duplicates = list(duplicate_dict.values())
assert isinstance(duplicates[0], list)
assert duplicate_dict['ukbench09268.jpg'] == []
assert duplicate_dict['ukbench00120.jpg'] == [('ukbench00120_resize.jpg',
0)]
def test_find_duplicates_to_remove_encoding():
encoding = {
'ukbench00120_resize.jpg': '9fee256239984d71',
'ukbench00120_rotation.jpg': '850d513c4fdcbb72',
'ukbench00120.jpg': '9fee256239984d71',
'ukbench00120_hflip.jpg': 'cabb7237e8cd3824',
'ukbench09268.jpg': 'c73c36c2da2f29c9',
}
phasher = PHash()
removal_list = phasher.find_duplicates_to_remove(encoding_map=encoding,
max_distance_threshold=10)
assert isinstance(removal_list, list)
assert (removal_list == ['ukbench00120.jpg']
or removal_list == ['ukbench00120_resize.jpg'])
def test_find_duplicates_encoding_map_input():
encoding = {
'ukbench00120_resize.jpg': '9fee256239984d71',
'ukbench00120_rotation.jpg': '850d513c4fdcbb72',
'ukbench00120.jpg': '9fee256239984d71',
'ukbench00120_hflip.jpg': 'cabb7237e8cd3824',
'ukbench09268.jpg': 'c73c36c2da2f29c9',
}
phasher = PHash()
duplicate_dict = phasher.find_duplicates(encoding_map=encoding,
max_distance_threshold=10)
assert isinstance(duplicate_dict, dict)
assert isinstance(list(duplicate_dict.values())[0], list)
assert len(duplicate_dict['ukbench09268.jpg']) == 0
assert duplicate_dict['ukbench00120.jpg'] == ['ukbench00120_resize.jpg']
def test_find_duplicates_clearing():
phasher = PHash()
duplicate_dict = phasher.find_duplicates(
image_dir=PATH_IMAGE_DIR,
max_distance_threshold=10,
scores=True,
search_method='brute_force_cython',
)
duplicate_dict = phasher.find_duplicates(
image_dir=PATH_IMAGE_DIR,
max_distance_threshold=10,
scores=True,
search_method='brute_force_cython',
)
assert isinstance(duplicate_dict, dict)
duplicates = list(duplicate_dict.values())
assert isinstance(duplicates[0], list)
assert duplicate_dict['ukbench09268.jpg'] == []
assert duplicate_dict['ukbench00120.jpg'] == [('ukbench00120_resize.jpg',
0)]
hasher.find_duplicates_to_remove(encoding_map=encoding_map,
max_distance_threshold=threshold,
outfile=outfile)
find_duplicates_mocker.assert_called_once_with(
encoding_map=encoding_map,
image_dir=None,
max_distance_threshold=threshold,
scores=False,
)
get_files_to_remove_mocker.assert_called_once_with(ret_val_find_dup_dict)
save_json_mocker.assert_not_called()
# Integration tests
phasher = PHash()
dhasher = DHash()
ahasher = AHash()
whasher = WHash()
common_test_parameters = [
phasher.encode_image,
dhasher.encode_image,
ahasher.encode_image,
whasher.encode_image,
]
@pytest.mark.parametrize('hash_function', common_test_parameters)
class TestCommon:
def test_len_hash(self, hash_function):