refactoring: represent binary as a string of vlsm length

cidr4_merger.py

@@ -9,77 +9,90 @@ def get_data(input_file):
     return data
 
 
-def cidr4_to_binary(cidr4: str) -> Tuple[str, int]:
+def cidr4_to_binary(cidr4: str) -> str:
     ip_str, vlsm = cidr4.strip().split("/")
     vlsm = int(vlsm)
     ipv4 = IPv4Address(ip_str)
     binary_ip = bin(int(ipv4))[2:]
     binary_ip = binary_ip.zfill(32)
-    return binary_ip, vlsm
+    binary = binary_ip[:vlsm]
+    return binary
 
 
-def binary_to_cidr4(binary_ip: str, vlsm: int) -> str:
+def binary_to_cidr4(binary: str) -> str:
+    vlsm = len(binary)
+    binary_ip = binary + "0" * (32 - vlsm)
     int_ip = int(binary_ip, 2)
-    ip_str = str(IPv4Address(int_ip))
+    ip = IPv4Address(int_ip)
-    return f"{ip_str}/{vlsm}"
+    return f"{ip}/{vlsm}"
 
 
-def reduce_bin_ip(bin_ip: str, vlsm: int) -> Tuple[str, int]:
+def reduce_binary(binary: str) -> str:
-    if vlsm == 0:
+    if len(binary) == 0:
-        return "0" * 32, vlsm
+        return ""
-    new_vlsm = vlsm - 1
+    new_binary = binary[:-1]
-    new_bin_ip = bin_ip[:new_vlsm] + "0" * (32 - new_vlsm)
+    if "1" not in new_binary:
-    if new_bin_ip == "0" * 32:
+        return ""
-        return "0" * 32, 0
+    return new_binary
-    return new_bin_ip, new_vlsm
 
 
-def rough_merge_ips(
+def rough_merge_binaries(binaries: List[str], req_len: int) -> List[str]:
-    bin_ips: List[Tuple[str, int]], req_len: int
+    ips = set(deepcopy(binaries))
-) -> List[Tuple[str, int]]:
+    non_reducible_ips = set()
-    ips = deepcopy(bin_ips)
     reduction_limit_reached = False
-    while len(ips) > req_len and not reduction_limit_reached:
+    while (
-        max_vlsm = max(ips, key=lambda x: x[1])[1]
+        len(ips) > 0
-        ips_to_reduce = []
+        and len(ips) + len(non_reducible_ips) > req_len
-        merged_ips = []
+        and not reduction_limit_reached
-        for ip_vlsm in ips:
+    ):
-            if ip_vlsm[1] == max_vlsm:
+        max_vlsm = len(max(ips, key=lambda x: len(x)))
-                ips_to_reduce.append(ip_vlsm)
+        reduced_ips = set()
+        merged_ips = set()
+        for ip in ips:
+            if len(ip) == max_vlsm:
+                reduced_ip = reduce_binary(ip)
+                if len(reduced_ip) > 0:
+                    reduced_ips.add(reduced_ip)
                 else:
-                merged_ips.append(ip_vlsm)
+                    non_reducible_ips.add(ip)
-        reduced_ips = map(lambda x: reduce_bin_ip(*x), ips_to_reduce)
+            else:
-        reduced_ips = set(reduced_ips)
+                merged_ips.add(ip)
-        reduced_ips = filter(lambda x: "1" in x[0], reduced_ips)
+        merged_ips.update(reduced_ips)
-        merged_ips.extend(list(reduced_ips))
+        if len(merged_ips) + len(non_reducible_ips) > req_len:
-        if len(merged_ips) > req_len:
             ips = merged_ips
         else:
             reduction_limit_reached = True
-    return ips
+
+    ips.update(non_reducible_ips)
+    return sorted(ips)
 
 
 def main():
     file = "cidr4.txt"
+    required_len = 20
+
     data = get_data(file)
     bin_ips = list(map(cidr4_to_binary, data))
-    rough_merged_bin_ips = rough_merge_ips(bin_ips, 20)
+    rough_merged_bin_ips = rough_merge_binaries(bin_ips, required_len)
     print(f"{len(rough_merged_bin_ips)=}")
-    sorted_ips = sorted(rough_merged_bin_ips)
+    for ip in rough_merged_bin_ips:
-    for ip in sorted_ips:
         print(ip)
 
 
 if __name__ == "__main__":
-    assert cidr4_to_binary("4.78.139.0/24") == ("00000100010011101000101100000000", 24)
+    assert cidr4_to_binary("4.78.139.0/24") == "000001000100111010001011"
-    assert binary_to_cidr4("00000100010011101000101100000000", 24) == "4.78.139.0/24"
+    assert cidr4_to_binary("128.0.0.0/1") == "1"
+    assert cidr4_to_binary("0.0.0.0/0") == ""
 
-    assert reduce_bin_ip("00000100010011101000101100000000", 24) == (
+    assert binary_to_cidr4("000001000100111010001011") == "4.78.139.0/24"
-        "00000100010011101000101000000000",
+    assert binary_to_cidr4("1") == "128.0.0.0/1"
-        23,
+    assert binary_to_cidr4("") == "0.0.0.0/0"
-    )
+
-    assert reduce_bin_ip("10000000000000000000000000000000", 1) == ("0" * 32, 0)
+    assert reduce_binary("000001000100111010001011") == "00000100010011101000101"
-    assert reduce_bin_ip("10000000000000000000000000000000", 0) == ("0" * 32, 0)
+    assert reduce_binary("1") == ""
-    assert reduce_bin_ip("0" * 32, 0) == ("0" * 32, 0)
+    assert reduce_binary("0") == ""
+    assert reduce_binary("") == ""
+    assert reduce_binary("0001") == ""
+    assert reduce_binary("0000") == ""
 
     main()