Cidr4 merge algorithm #5

Merged
PavelPatsey merged 91 commits from CIDR4_merge_algorithm into main 2025-01-27 22:05:39 +03:00
2 changed files with 621 additions and 0 deletions
Showing only changes of commit 2cab2c31a6 - Show all commits
+395
View File
@@ -0,0 +1,395 @@
import pytest
from cidr4_merger import (
Cidr4MergerError,
cidr4_to_node,
find_neighbours_singles,
get_group_with_max_mask_len,
get_net_addr,
get_parent_ip,
have_same_parent,
lift_lonely_node,
make_cidr4,
make_groups,
make_parent,
merge_neighbors,
merge_nodes_cycle,
merge_nodes_deprecated,
merge_nodes_recursion,
reduce_nodes,
sort_nodes,
)
from .cidr4_data_for_tests import test_cidr4_data
def test_true():
assert True
bin_a = "10011000000000001000010000010000"
assert len(bin_a) == 32
ip_a = int(bin_a, 2)
bin_b = "10011100000000000000000000101011"
assert len(bin_b) == 32
ip_b = int(bin_b, 2)
bin_c = "10011000000000000000000000000000"
assert len(bin_c) == 32
ip_c = int(bin_c, 2)
bin_d = "11111100000000000000000000000000"
assert len(bin_c) == 32
ip_d = int(bin_d, 2)
def test_cidr4_to_node():
assert cidr4_to_node("4.78.139.0/24") == (72256256, 24, 0, 72256000)
assert cidr4_to_node("0.0.0.0/32") == (0, 32, 0, 0)
assert cidr4_to_node("23.234.30.0/24") == (401219072, 24, 0, 401219072)
assert cidr4_to_node("172.217.0.0/19") == (2899902464, 19, 0, 2899902464)
assert cidr4_to_node("23.225.141.0/24") == (400657664, 24, 0, 400657408)
assert cidr4_to_node("31.13.94.0/23") == (520969728, 23, 0, 520969216)
assert cidr4_to_node("0.0.0.0/2") == (0, 2, 0, 0)
assert cidr4_to_node("64.0.0.0/2") == (1073741824, 2, 0, 0)
assert cidr4_to_node("128.0.0.0/2") == (2147483648, 2, 0, 2147483648)
assert cidr4_to_node("192.0.0.0/2") == (3221225472, 2, 0, 2147483648)
def test_make_cidr4():
assert make_cidr4(72256256, 24) == "4.78.139.0/24"
assert make_cidr4(0, 32) == "0.0.0.0/32"
assert make_cidr4(401219072, 24) == "23.234.30.0/24"
assert make_cidr4(2899902464, 19) == "172.217.0.0/19"
assert make_cidr4(400657664, 24) == "23.225.141.0/24"
assert make_cidr4(520969728, 23) == "31.13.94.0/23"
def test_get_net_addr():
assert get_net_addr(ip_a, 5) == ip_c
assert get_net_addr(ip_b, 5) == ip_c
assert get_net_addr(0, 1) == 0
assert get_net_addr(0, 0) == 0
def test_get_parent_mask():
assert get_parent_ip(ip_a, 6) == ip_c
assert get_parent_ip(ip_b, 6) == ip_c
assert get_parent_ip(0, 1) == 0
with pytest.raises(Exception) as exc_info:
get_parent_ip(0, 0)
assert str(exc_info.value) == "The top of the tree has no parent!"
assert exc_info.type is Cidr4MergerError
def test_have_same_parent():
assert have_same_parent(ip_c, 6, ip_c, 6) is True
assert have_same_parent(ip_c, 6, ip_c, 5) is False
assert have_same_parent(ip_a, 6, ip_d, 6) is False
assert have_same_parent(ip_a, 6, 0, 1) is False
assert have_same_parent(ip_a, 6, 0, 0) is False
def test_sort_nodes():
assert sort_nodes(
[
(401219072, 24, 0, 401219072),
(2899902464, 19, 0, 2899902464),
(400657664, 24, 0, 400657408),
(520969728, 23, 0, 520969216),
]
) == [
(2899902464, 19, 0, 2899902464),
(520969728, 23, 0, 520969216),
(400657664, 24, 0, 400657408),
(401219072, 24, 0, 401219072),
]
def test_get_group_with_max_mask_len():
assert get_group_with_max_mask_len(
[
(2899902464, 19, 0, 2899902464),
(520969728, 23, 0, 520969216),
(400657664, 24, 0, 400657408),
(401219072, 24, 0, 401219072),
]
) == [(400657664, 24, 0, 400657408), (401219072, 24, 0, 401219072)]
assert get_group_with_max_mask_len(
[
(401219072, 24, 0, 401219072),
(2899902464, 19, 0, 2899902464),
(520969728, 23, 0, 520969216),
]
) == [(401219072, 24, 0, 401219072)]
def test_make_parent():
assert make_parent((0, 2, 12, 0), (1073741824, 2, 3, 0)) == (0, 1, 15, 0)
assert make_parent(
(2147483648, 2, 1, 2147483648), (3221225472, 2, 2, 2147483648)
) == (2147483648, 1, 3, 0)
with pytest.raises(Exception) as exc_info:
make_parent((0, 2, 12, 0), (3221225472, 2, 2, 2147483648))
assert str(exc_info.value) == "Nodes must be neighbors!"
assert exc_info.type is Cidr4MergerError
def test_reduce_nodes():
assert reduce_nodes(
[
(0, 2, 12, 0),
(1073741824, 2, 3, 0),
]
) == [
(0, 1, 15, 0),
]
assert reduce_nodes(
[
(0, 2, 12, 0),
(1073741824, 2, 3, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
]
) == [
(2147483648, 1, 3, 0),
(0, 2, 12, 0),
(1073741824, 2, 3, 0),
]
assert reduce_nodes(
[
(0, 2, 12, 0),
(2147483648, 1, 0, 0),
]
) == [
(0, 1, 12 + 2**30, 0),
(2147483648, 1, 0, 0),
]
with pytest.raises(Exception) as exc_info:
reduce_nodes(
[
(0, 1, 12 + 2**30, 0),
(2147483648, 1, 0, 0),
]
)
assert exc_info.type is Cidr4MergerError
assert str(exc_info.value) == "The top of the tree has no parent!"
def test_merge_nodes_deprecated():
assert merge_nodes_deprecated(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
2,
) == [
(2147483648, 1, 3, 0),
(0, 2, 12, 0),
]
with pytest.raises(Exception) as exc_info:
merge_nodes_deprecated(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
1,
)
assert exc_info.type is Cidr4MergerError
assert str(exc_info.value) == "The top of the tree has no parent!"
def test_make_groups():
nodes = [
(2398793728, 20, 0, 2398789632),
(2899943424, 20, 0, 2899943424),
(3627728896, 20, 0, 3627728896),
(520963072, 22, 0, 520962048),
(1089054720, 22, 0, 1089054720),
(2899902464, 19, 0, 2899902464),
(2915221504, 19, 0, 2915221504),
]
assert dict(make_groups(nodes)) == {
20: [
(2398793728, 20, 0, 2398789632),
(2899943424, 20, 0, 2899943424),
(3627728896, 20, 0, 3627728896),
],
22: [(520963072, 22, 0, 520962048), (1089054720, 22, 0, 1089054720)],
19: [(2899902464, 19, 0, 2899902464), (2915221504, 19, 0, 2915221504)],
}
@pytest.fixture
def nodes_only_neighbours():
return [
(0, 2, 12, 0),
(1073741824, 2, 3, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
]
@pytest.fixture
def nodes_only_singles():
return [
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
]
@pytest.fixture
def nodes_with_neighbours_n_singles():
return [
(0, 2, 12, 0),
(1073741824, 2, 3, 0),
(2147483648, 2, 1, 2147483648),
]
@pytest.fixture
def groups_only_neighbours(nodes_only_neighbours):
return make_groups(nodes_only_neighbours)
@pytest.fixture
def groups_only_singles(nodes_only_singles):
return make_groups(nodes_only_singles)
@pytest.fixture
def groups_with_neighbours_n_singles(nodes_with_neighbours_n_singles):
return make_groups(nodes_with_neighbours_n_singles)
def test_find_neighbours_singles(
groups_only_neighbours,
groups_only_singles,
groups_with_neighbours_n_singles,
):
assert find_neighbours_singles(groups_only_neighbours) == (
[
((0, 2, 12, 0), (1073741824, 2, 3, 0)),
((2147483648, 2, 1, 2147483648), (3221225472, 2, 2, 2147483648)),
],
[],
)
assert find_neighbours_singles(groups_only_singles) == (
[],
[(0, 2, 12, 0), (2147483648, 2, 1, 2147483648)],
)
assert find_neighbours_singles(groups_with_neighbours_n_singles) == (
[((0, 2, 12, 0), (1073741824, 2, 3, 0))],
[(2147483648, 2, 1, 2147483648)],
)
def test_merge_neighbors__only_neighbours(nodes_only_neighbours):
new_nodes = merge_neighbors(
nodes_only_neighbours,
[
((0, 2, 12, 0), (1073741824, 2, 3, 0)),
((2147483648, 2, 1, 2147483648), (3221225472, 2, 2, 2147483648)),
],
)
assert new_nodes == [
(0, 1, 15, 0),
(2147483648, 1, 3, 0),
]
def test_merge_neighbors__neighbours_n_singles(nodes_with_neighbours_n_singles):
new_nodes = merge_neighbors(
nodes_with_neighbours_n_singles,
[((0, 2, 12, 0), (1073741824, 2, 3, 0))],
)
assert new_nodes == [
(0, 1, 15, 0),
(2147483648, 2, 1, 2147483648),
]
def test_lift_lonely_node(nodes_only_singles):
singles = [(0, 2, 12, 0), (2147483648, 2, 1, 2147483648)]
new_nodes = lift_lonely_node(nodes_only_singles, singles)
assert new_nodes == [(2147483648, 1, 1073741825, 0), (0, 2, 12, 0)]
def test_merge_nodes_recursion():
assert merge_nodes_recursion(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
2,
) == [
(2147483648, 1, 3, 0),
(0, 2, 12, 0),
]
with pytest.raises(Exception) as exc_info:
merge_nodes_recursion(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
1,
)
assert exc_info.type is Cidr4MergerError
assert str(exc_info.value) == "The top of the tree has no parent!"
def test_merge_nodes_cycle():
assert merge_nodes_cycle(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
2,
) == [
(2147483648, 1, 3, 0),
(0, 2, 12, 0),
]
with pytest.raises(Exception) as exc_info:
merge_nodes_cycle(
[
(0, 2, 12, 0),
(2147483648, 2, 1, 2147483648),
(3221225472, 2, 2, 2147483648),
],
1,
)
assert exc_info.type is Cidr4MergerError
assert str(exc_info.value) == "The top of the tree has no parent!"
def test_merge_nodes_recursion_vs_cycle():
required_len = 20
test_data = test_cidr4_data.strip().splitlines()
test_nodes = list(map(cidr4_to_node, test_data))
test_nodes = sort_nodes(test_nodes)
merged_nodes_recursion = merge_nodes_recursion(test_nodes, required_len)
merged_nodes_cycle = merge_nodes_cycle(test_nodes, required_len)
assert merged_nodes_recursion == merged_nodes_cycle
+226
View File
@@ -0,0 +1,226 @@
import cProfile
import sys
from collections import defaultdict
sys.setrecursionlimit(10_000)
Node = tuple[int, int, int, int]
class Cidr4MergerError(Exception):
pass
def get_data(input_file):
with open(input_file, "r") as file:
data = file.read().splitlines()
return data
def cidr4_to_node(cidr4: str) -> Node:
ip_address, mask_len = cidr4.strip().split("/")
mask_len = int(mask_len)
a, b, c, d = list(map(int, ip_address.split(".")))
ip = a * 256**3 + b * 256**2 + c * 256**1 + d * 256**0
added_ips_number = 0
parent_ip = get_parent_ip(ip, mask_len)
return ip, mask_len, added_ips_number, parent_ip
def sort_nodes(nodes: list[Node]) -> list[Node]:
return sorted(nodes, key=lambda x: (x[1], x[0]))
def get_net_addr(ip: int, mask_len: int) -> int:
mask = ((1 << mask_len) - 1) << (32 - mask_len)
net_addr = ip & mask
return net_addr
def get_parent_ip(ip: int, mask_len: int) -> int:
if mask_len == 0:
raise Cidr4MergerError("The top of the tree has no parent!")
return get_net_addr(ip, mask_len - 1)
def have_same_parent(mask_len_a, parent_ip_a, mask_len_b, parent_ip_b) -> bool:
return mask_len_a == mask_len_b and parent_ip_a == parent_ip_b
def get_group_with_max_mask_len(nodes: list[Node]) -> list[Node]:
max_mask_len = max(nodes, key=lambda x: x[1])[1]
return list(filter(lambda x: x[1] == max_mask_len, nodes))
def make_parent(a: Node, b: Node | None = None) -> Node:
ip_a, mask_len_a, added_ips_number_a, parent_ip_a = a
if b:
ip_b, mask_len_b, added_ips_b, parent_ip_b = b
if not have_same_parent(mask_len_a, parent_ip_a, mask_len_b, parent_ip_b):
raise Cidr4MergerError("Nodes must be neighbors!")
added_ips_number = added_ips_number_a + added_ips_b
else:
added_ips_number = added_ips_number_a + 2 ** (32 - mask_len_a)
ip = parent_ip_a
mask_len = mask_len_a - 1
parent_ip = get_parent_ip(ip, mask_len)
return ip, mask_len, added_ips_number, parent_ip
def reduce_nodes(nodes: list[Node]) -> list[Node]:
group = get_group_with_max_mask_len(nodes)
neighbours = []
loners = []
i = 0
while i < len(group) - 1:
a, b = group[i], group[i + 1]
ip_a, mask_len_a, _, parent_ip_a = a
ip_b, mask_len_b, _, parent_ip_b = b
if have_same_parent(mask_len_a, parent_ip_a, mask_len_b, parent_ip_b):
neighbours.append((a, b))
i += 2
else:
loners.append(a)
i += 1
if i == len(group) - 1:
loners.append(group[i])
if neighbours:
zipped = zip(neighbours, map(lambda x: make_parent(x[0], x[1]), neighbours))
min_zipped = min(zipped, key=lambda x: x[1][2])
(a, b), parent = min_zipped
nodes.remove(a)
nodes.remove(b)
nodes.append(parent)
elif loners:
zipped = zip(loners, map(make_parent, loners))
min_zipped = min(zipped, key=lambda x: x[1][2])
a, parent = min_zipped
nodes.remove(a)
nodes.append(parent)
else:
assert False, "Error"
return sort_nodes(nodes)
def merge_nodes_deprecated(nodes: list[Node], required_len: int) -> list[Node]:
while len(nodes) > required_len:
nodes = reduce_nodes(nodes)
return nodes
def make_cidr4(ip, mask_len) -> str:
lst = [str(ip >> (i << 3) & 0xFF) for i in reversed(range(4))]
ip_address = ".".join(lst)
return f"{ip_address}/{mask_len}"
def lift_lonely_node(nodes: list[Node], singles: list[Node]) -> list[Node]:
# find single whose parent has the least added addresses
min_single, min_parent = singles[0], make_parent(singles[0])
for node in singles[1:]:
parent = make_parent(node)
if parent[2] < min_parent[2]:
min_single, min_parent = node, parent
nodes.remove(min_single)
nodes.append(min_parent)
nodes = sort_nodes(nodes)
return nodes
def merge_neighbors(
nodes: list[Node], neighbours: list[tuple[Node, Node]]
) -> list[Node]:
for a, b in neighbours:
parent = make_parent(a, b)
nodes.remove(a)
nodes.remove(b)
nodes.append(parent)
return sort_nodes(nodes)
def find_neighbours_singles(groups: defaultdict) -> tuple[list, list]:
neighbours = []
singles = []
for group in groups.values():
i = 0
while i < len(group) - 1:
a, b = group[i], group[i + 1]
ip_a, mask_len_a, _, parent_ip_a = a
ip_b, mask_len_b, _, parent_ip_b = b
if have_same_parent(mask_len_a, parent_ip_a, mask_len_b, parent_ip_b):
neighbours.append((a, b))
i += 2
else:
singles.append(a)
i += 1
if i == len(group) - 1:
singles.append(group[i])
return neighbours, singles
def make_groups(nodes: list[Node]) -> defaultdict:
groups = defaultdict(list)
for n in nodes:
groups[n[1]].append(n)
return groups
def merge_nodes_recursion(nodes: list[Node], required_len: int) -> list[Node]:
if len(nodes) <= required_len:
return nodes
groups = make_groups(nodes)
neighbours, singles = find_neighbours_singles(groups)
if neighbours:
new_nodes = merge_neighbors(nodes, neighbours)
return merge_nodes_recursion(new_nodes, required_len)
new_nodes = lift_lonely_node(nodes, singles)
return merge_nodes_recursion(new_nodes, required_len)
def merge_nodes_cycle(nodes_to_merge: list[Node], required_len: int) -> list[Node]:
nodes = [x for x in nodes_to_merge]
while not len(nodes) <= required_len:
groups = make_groups(nodes)
neighbours, singles = find_neighbours_singles(groups)
if neighbours:
nodes = merge_neighbors(nodes, neighbours)
elif singles:
nodes = lift_lonely_node(nodes, singles)
else:
raise Cidr4MergerError("Invalid case!")
return nodes
def main():
file = "cidr4.txt"
required_len = 20
data = get_data(file)
nodes = list(map(cidr4_to_node, data))
nodes = sort_nodes(nodes)
# merged_nodes = merge_nodes_deprecated(nodes, required_len)
# merged_nodes = merge_nodes_recursion(nodes, required_len)
merged_nodes = merge_nodes_cycle(nodes, required_len)
cidr4s = []
sum_added_ips = 0
for ip_value, mask_len, added_ips, _ in merged_nodes:
cidr4s.append(make_cidr4(ip_value, mask_len))
sum_added_ips += added_ips
cidr4s_str = "\n".join(cidr4s)
print(
f"Исходный список длины {len(nodes)} сокращен до {len(cidr4s)}\n"
f"Количество добавленных ip адресов: {sum_added_ips:_}\n"
f"Список объединенных cidr4:\n"
f"{cidr4s_str}"
)
if __name__ == "__main__":
cProfile.run("main()")