ATTENTION version 2.0.0 is a breaking change from previous versions for
handling IPv6 addresses (functions for IPv4 are unchanged). Calls that result
in arithmatic operations against IPv6 now use uint128.Uint128
instead of *big.Int
. Until now this library restricted itself to using the
standard library, but math/big
is sloooooooooow and the performance gains
from switching were too large to ignore:
Benchmark | *big.Int | uint128.Uint128 |
---|---|---|
Benchmark_DeltaIP6 | 79.27 ns/op | 2.809 ns/op |
BenchmarkDecrementIP6By | 50.54 ns/op | 13.88 ns/op |
BenchmarkIncrementIP6By | 50.48 ns/op | 13.92 ns/op |
BenchmarkNet_Count6 | 122.2 ns/op | 11.26 ns/op |
It would be fantastic to remove this external dependency in some future v3
that switched to a native uint128
but for that to happen this proposal
(or something similar) would need to be adopted.
Okay you can stop paying attention now
I really enjoy Python's ipaddress library and Ruby's ipaddr, I think you can write a lot of neat software if some of the little problems around manipulating IP addresses and netblocks are taken care of for you, so I set out to write something like them for my language of choice, Go. This is what I've come up with.
IPLib is a hopefully useful, aspirationally full-featured library built around and on top of the address primitives found in the net package, it seeks to make them more accessible and easier to manipulate.
It includes:
Some simple tools for performing common tasks against IP objects:
- compare two addresses
- make a copy of a net.IP address
- get the delta between two addresses
- sort
- decrement or increment addresses
- print addresses as binary or hexadecimal strings, or print their addr.ARPA DNS name
- print v6 in fully expanded form
- convert between net.IP and integer values
- get the version of a v4 address or force a IPv4-mapped IPv6address to be a v4 address
An enhancement of net.IPNet
, iplib.Net
is an interface with two, version-
specific implementations providing features such as:
- retrieve the first and last usable address
- retrieve the wildcard mask
- enumerate all or part of a netblock to
[]net.IP
- decrement or increment addresses within the boundaries of the netblock
- return the supernet of a netblock
- allocate subnets within the netblock
- return next- or previous-adjacent netblocks
The two address versions behave differently in both large and subtle ways, and the version-specific implementations seek to account for this. For example the Net4 implementation omits the network and broadcast addresses from consideration during enumeration; while the Net6 implementation introduces the concept of a HostMask, which blocks usable addresses off from the right in the same way that a netmask constrains them from the left
Additional version-specific considerations described in the Net4 and Net6 sections below.
- iana - a module for referencing IP netblocks against the Internet Assigned Numbers Authority's Special IP Address Registry
- iid - a module for generating and validating IPv6 Interface Identifiers, including RFC4291 modified EUI64 and RFC7217 Semantically Opaque addresses
go get -u github.com/c-robinson/iplib/v2
There are a series of functions for working with v4 or v6 net.IP
objects:
package main
import (
"fmt"
"net"
"sort"
"github.com/c-robinson/iplib/v2"
)
func main() {
ipa := net.ParseIP("192.168.1.1")
ipb := iplib.IncrementIPBy(ipa, 15) // ipb is 192.168.1.16
ipc := iplib.NextIP(ipa) // ipc is 192.168.1.2
fmt.Println(iplib.CompareIPs(ipa, ipb)) // -1
fmt.Println(iplib.DeltaIP(ipa, ipb)) // 15
fmt.Println(iplib.IPToHexString(ipc)) // "c0a80102"
iplist := []net.IP{ ipb, ipc, ipa }
sort.Sort(iplib.ByIP(iplist)) // []net.IP{ipa, ipc, ipb}
fmt.Println(iplib.IP4ToUint32(ipa)) // 3232235777
fmt.Println(iplib.IPToBinaryString(ipa)) // 11000000.10101000.00000001.00000001
fmt.Println(iplib.IP4ToARPA(ipa)) // 1.1.168.192.in-addr.arpa
}
Addresses that require or return a count default to using uint32
, which is
sufficient for working with the entire IPv4 space. As a rule these functions
are just lowest-common wrappers around IPv4- or IPv6-specific functions. The
IPv6-specific variants use uint128.Uint128
so they can access the entire v6
space.
Net
describes an iplib.Net
object, the exposed functions are those that
are required for comparison, sorting, generic initialization and for ancillary
functions such as those found in this package's submodules.
Net4
represents an IPv4 network. Since the first and last addresses of a v4
network are typically not allocated for use these will be omitted by
Enumerate()
, NextIP()
and PreviousIP()
; they wont show up in Count()
;
and FirstAddress()
and LastAddress()
show the 2nd and 2nd-to-the-last
addresses respectively. The v4-specific method NetworkAddress()
returns the
first address, while BroadcastAddress()
returns the last. There is an
exception made for Net4
networks defined with a 31-bit netmask, since these
are assumed to be for RFC3021
point-to-point links.
Additionally Net4
contains a Wildcard()
method which will return the
network's wildcard address.
n := iplib.NewNet4(net.ParseIP("192.168.0.0"), 16)
fmt.Println(n.Count()) // 65534 (note: not 65536)
fmt.Println(n.Enumerate(2, 1024)) // [192.168.4.1 192.168.4.2]
fmt.Println(n.IP()) // 192.168.0.0
fmt.Println(n.FirstAddress()) // 192.168.0.1
fmt.Println(n.LastAddress()) // 192.168.255.254
fmt.Println(n.BroadcastAddress()) // 192.168.255.255
fmt.Println(n.Wildcard()) // 0000ffff
fmt.Println(n.Subnet(0)) // [192.168.0.0/17 192.168.128.0/17] <nil>
fmt.Println(n.Supernet(0)) // 192.168.0.0/15 <nil>
Net6
represents an IPv6 network. In some ways v6 is simpler than v4, as
it does away with the special behavior of addresses at the front and back of
the netblock. For IPv6 the primary problem is the sheer size of the thing:
there are 2^128th addresses in IPv6, which translates to 340 undecillion!
n := iplib.NewNet6(net.ParseIP("2001:db8::"), 56, 0)
fmt.Println(n.Count()) // 4722366482869645213696
fmt.Println(n.Enumerate(2, 1024)) // [2001:db8::400 2001:db8::401]
fmt.Println(n.FirstAddress()) // 2001:db8::
fmt.Println(n.NextIP(n.FirstAddress())) // 2001:db8::1 <nil>
fmt.Println(n.LastAddress()) // 2001:db8:0:ff:ffff:ffff:ffff:ffff
fmt.Println(n.Subnet(0, 0)) // [2001:db8::/57 2001:db8:0:80::/57] <nil>
fmt.Println(n.Supernet(0, 0)) // 2001:db8::/55 <nil>
To manage the address space, Net6
introduces HostMask
. This optional
constraint can be used to block addresses on the right-side of a netblock
somewhat like Netmasks do on the left. Hostmask
must be specified at
initialization time and, if set, will affect the behavior of Count()
,
Enumerate()
, LastAddress()
, NextIP()
and PreviousIP()
. Subnet()
and
Supernet()
generate objects that inherit the hostmask of their parent, while
a hostmask must be specified for NextNet()
and PreviousNet()
.
// this is the same as the previous example, except with a hostmask set
n := NewNet6(net.ParseIP("2001:db8::"), 56, 60)
fmt.Println(n.Count()) // 4096
fmt.Println(n.Enumerate(2, 1024)) // [2001:db8:0:40:: 2001:db8:0:40:100::]
fmt.Println(n.FirstAddress()) // 2001:db8::
fmt.Println(n.NextIP(n.FirstAddress())) // 2001:db8:0:0:100:: <nil>
fmt.Println(n.LastAddress()) // 2001:db8:0:ff:f00::
fmt.Println(n.Mask().String()) // ffffffffffffff000000000000000000
fmt.Println(n.Hostmask.String()) // 0000000000000000f0ffffffffffffff
fmt.Println(n.Subnet(0, 60)) // [2001:db8::/57 2001:db8:0:80::/57] <nil>
fmt.Println(n.Supernet(0, 60)) // 2001:db8::/55 <nil>
IPfool is a simple command-line tool that I wrote to test many of the features within this library and might be useful in evaluating it.