def test_different_n():
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] * 2
n = 5
actual = group(iterable, n)
expected = [[1, 2, 3, 4, 5], [6, 7, 8, 9, 10], [1, 2, 3, 4, 5],
[6, 7, 8, 9, 10]]
assert actual == expected
def testGroup():
items = [('I', 'A', 1), ('I', 'A', 2), ('I', 'B', 1), ('II', 'A', 1),
('II', 'B', 1)]
expected = {
'I': {
'A': {
1: [('I', 'A', 1)],
2: [('I', 'A', 2)]
},
'B': {
1: [('I', 'B', 1)]
}
},
'II': {
'A': {
1: [('II', 'A', 1)]
},
'B': {
1: [('II', 'B', 1)]
}
}
}
actual = group(items, lambda i: i[0], lambda i: i[1], lambda i: i[2])
assert actual == expected
def test_different_iterable_size():
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] * 2
n = 3
actual = group(iterable, n)
expected = [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10, 1, 2], [3, 4, 5],
[6, 7, 8], [9, 10]]
assert actual == expected
def test_passing_in_generator():
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9]
gen = (i for i in iterable)
assert isinstance(gen, types.GeneratorType)
n = 3
actual = group(gen, n)
expected = [[1, 2, 3], [4, 5, 6], [7, 8, 9]]
assert actual == expected
def print_hashes(args):
exclude_list = build_exclude_list()
results = []
for filepath in args["files_list"]:
result = telfhash_single(filepath,
debug=args["debug"],
exclude_list=exclude_list)
results.append(result)
# the fancy formatting is done so that we could properly vertically
# align the telfhashes in the second column. we're using the `max_len`
# value computed before in the get_args() function
#
# data is printed as soon as the data is obtained so that the user sees
# data right away, and it makes the console more active.
# only go this path if args['output']=None and args['format']=None
if args["output"] is None and args["format"] is None:
if result["telfhash"] is not None:
print("{:<{max_len}} {}".format(result["file"],
result["telfhash"],
max_len=args["max_len"]))
else:
print("{:<{max_len}} {msg}".format(filepath,
max_len=args["max_len"],
msg=result["msg"]))
if args["format"] == "tsv":
output_format_tsv(args, results)
elif args["format"] == "json":
output_format_json(args, results)
if args["group"] and len(results) > 1:
groups = grouping.group(results, threshold=args["threshold"])
print()
for i in range(len(groups["grouped"])):
print("Group {}:".format(i + 1))
for f in groups["grouped"][i]:
print(" {}".format(f))
if len(groups["nogroup"]) > 0:
print("Ungrouped:")
for f in groups["nogroup"]:
print(" {}".format(f))
print()
def group(telfhash_results, threshold=50):
"""Group the files according to the TLSH distances between the telfhashes
of the files
Args:
telfhash_results: The output of the telfhash.telfhash function call. List
of telfhash data of the files
threshold: [Optional] The minimum TLSH distance between telfhashes
for the files to be considered as related
Returns:
Tuple of tuples, each member tuple is one group
"""
groups = grouping.group(telfhash_results, threshold=threshold)
return groups
def test_passing_in_tuple():
iterable = (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
n = 3
actual = group(iterable, n)
expected = [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10]]
assert actual == expected
def test_split_10_ints_by_3():
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
n = 3
actual = group(iterable, n)
expected = [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10]]
assert actual == expected
iterable = list(iterable)
for i in range(0, len(iterable), n):
result.append(iterable[i:i + n])
return result
# another way
# iterator = iter(iterable)
# result = list(iter(lambda: list(islice(iterator, n)), []))
# return result
if __name__ == '__main__':
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
n = 3
ret = group(iterable, n)
print(ret)
# TESTS
import types
from grouping import group
def test_split_10_ints_by_3():
iterable = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
n = 3
actual = group(iterable, n)
expected = [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10]]
assert actual == expected
