How to validate a IPv6 address format with shell? - bash

Question

In my shell, I need check if a string is a valid IPv6 address.
I find two ways, neither of them is ideal enough to me.
One is http://twobit.us/2011/07/validating-ip-addresses/, while I wonder if it must be such complex for such a common requirement.
The other is expand ipv6 address in shell script, this is simple, but for major distribution of Linux, sipcalc isn't a common default utility.
So my question, is there a simple way or utility to validate a IPv6 address with shell?
Thanks in advance.

Answer 1

The code in the first link isn't particularly elegant, but modulo stylistic fixes, I don't think you can simplify much beyond that (and as indicated in a comment, it may already be too simple). The spec is complex and mandates a number of optional features, which is nice for the end user, but cumbersome for the implementor.
You could probably find a library for a common scripting language which properly encapsulates this logic in a library. My thoughts would go to Python, where indeed Python 3.3 includes a standard module called ipaddress; for older versions, try something like
#!/usr/bin/env python
import socket
import sys
try:
socket.inet_pton(socket.AF_INET6, sys.argv[1])
result=0
except socket.error:
result=1
sys.exit(result)
See also Checking for IP addresses

Answer 2

Here is a solution in POSIX compatible shell script that handles IPv4 and IPv6 addresses with an optional subnet mask. To test an IP that should not have a subnet mask just pass it a dummy one when performing the test. It seems like a lot of code but it should be significantly faster than using external programs like grep or scripts that are likely to fork.
Single IPv6 zero groups compressed to :: will be treated as invalid. The use of such representation is strongly discouraged, but technically correct. There is a note in the code explaining how to alter this behaivour if you wish to allow such addresses.
#!/bin/sh
set -e
# return nonzero unless $1 contains only digits, leading zeroes not allowed
is_numeric() {
case "$1" in
"" | *[![:digit:]]* | 0[[:digit:]]* ) return 1;;
esac
}
# return nonzero unless $1 contains only hexadecimal digits
is_hex() {
case "$1" in
"" | *[![:xdigit:]]* ) return 1;;
esac
}
# return nonzero unless $1 is a valid IPv4 address with optional trailing subnet mask in the format /<bits>
is_ip4() {
# fail if $1 is not set, move it into a variable so we can mangle it
[ -n "$1" ] || return
IP4_ADDR="$1"
# handle subnet mask for any address containing a /
case "$IP4_ADDR" in
*"/"* ) # set $IP4_GROUP to the number of bits (the characters after the last /)
IP4_GROUP="${IP4_ADDR##*"/"}"
# return failure unless $IP4_GROUP is a positive integer less than or equal to 32
is_numeric "$IP4_GROUP" && [ "$IP4_GROUP" -le 32 ] || return
# remove the subnet mask from the address
IP4_ADDR="${IP4_ADDR%"/$IP4_GROUP"}";;
esac
# backup current $IFS, set $IFS to . as that's what separates digit groups (octets)
IP4_IFS="$IFS"; IFS="."
# initialize count
IP4_COUNT=0
# loop over digit groups
for IP4_GROUP in $IP4_ADDR ;do
# return failure if group is not numeric or if it is greater than 255
! is_numeric "$IP4_GROUP" || [ "$IP4_GROUP" -gt 255 ] && IFS="$IP4_IFS" && return 1
# increment count
IP4_COUNT=$(( IP4_COUNT + 1 ))
# the following line will prevent the loop continuing to run for invalid addresses with many occurrences of .
# this makes no difference to the result, but may improve performance when validating many such invalid strings
[ "$IP4_COUNT" -le 4 ] || break
done
# restore $IFS
IFS="$IP4_IFS"
# return success if there are 4 digit groups, otherwise return failure
[ "$IP4_COUNT" -eq 4 ]
}
# return nonzero unless $1 is a valid IPv6 address with optional trailing subnet mask in the format /<bits>
is_ip6() {
# fail if $1 is not set, move it into a variable so we can mangle it
[ -n "$1" ] || return
IP6_ADDR="$1"
# handle subnet mask for any address containing a /
case "$IP6_ADDR" in
*"/"* ) # set $IP6_GROUP to the number of bits (the characters after the last /)
IP6_GROUP="${IP6_ADDR##*"/"}"
# return failure unless $IP6_GROUP is a positive integer less than or equal to 128
is_numeric "$IP6_GROUP" && [ "$IP6_GROUP" -le 128 ] || return
# remove the subnet mask from the address
IP6_ADDR="${IP6_ADDR%"/$IP6_GROUP"}";;
esac
# perform some preliminary tests and check for the presence of ::
case "$IP6_ADDR" in
# failure cases
# *"::"*"::"* matches multiple occurrences of ::
# *":::"* matches three or more consecutive occurrences of :
# *[^:]":" matches trailing single :
# *"."*":"* matches : after .
*"::"*"::"* | *":::"* | *[^:]":" | *"."*":"* ) return 1;;
*"::"* ) # set flag $IP6_EXPANDED to true, to allow for a variable number of digit groups
IP6_EXPANDED=0
# because :: should not be used for remove a single zero group we start the group count at 1 when :: exists
# NOTE This is a strict interpretation of the standard, applications should not generate such IP addresses but (I think)
# they are in fact technically valid. To allow addresses with single zero groups replaced by :: set $IP6_COUNT to
# zero after this case statement instead
IP6_COUNT=1;;
* ) # set flag $IP6_EXPANDED to false, to forbid a variable number of digit groups
IP6_EXPANDED=""
# initialize count
IP6_COUNT=0;;
esac
# backup current $IFS, set $IFS to : to delimit digit groups
IP6_IFS="$IFS"; IFS=":"
# loop over digit groups
for IP6_GROUP in $IP6_ADDR ;do
# if this is an empty group then increment count and process next group
[ -z "$IP6_GROUP" ] && IP6_COUNT=$(( IP6_COUNT + 1 )) && continue
# handle dotted quad notation groups
case "$IP6_GROUP" in
*"."* ) # return failure if group is not a valid IPv4 address
# NOTE a subnet mask is added to the group to ensure we are matching addresses only, not ranges
! is_ip4 "$IP6_GROUP/1" && IFS="$IP6_IFS" && return 1
# a dotted quad refers to 32 bits, the same as two 16 bit digit groups, so we increment the count by 2
IP6_COUNT=$(( IP6_COUNT + 2 ))
# we can stop processing groups now as we can be certain this is the last group, : after . was caught as a failure case earlier
break;;
esac
# if there are more than 4 characters or any character is not a hex digit then return failure
[ "${#IP6_GROUP}" -gt 4 ] || ! is_hex "$IP6_GROUP" && IFS="$IP6_IFS" && return 1
# increment count
IP6_COUNT=$(( IP6_COUNT + 1 ))
# the following line will prevent the loop continuing to run for invalid addresses with many occurrences of a single :
# this makes no difference to the result, but may improve performance when validating many such invalid strings
[ "$IP6_COUNT" -le 8 ] || break
done
# restore $IFS
IFS="$IP6_IFS"
# if this address contained a :: and it has less than or equal to 8 groups then return success
[ "$IP6_EXPANDED" = "0" ] && [ "$IP6_COUNT" -le 8 ] && return
# if this address contained exactly 8 groups then return success, otherwise return failure
[ "$IP6_COUNT" -eq 8 ]
}
Here are some tests.
# tests
TEST_PASSES=0
TEST_FAILURES=0
for TEST_IP in 0.0.0.0 255.255.255.255 1.2.3.4/1 1.2.3.4/32 12.12.12.12 123.123.123.123 101.201.201.109 ;do
! is_ip4 "$TEST_IP" && printf "IP4 test failed, test case '%s' returned invalid\n" "$TEST_IP" && TEST_FAILURES=$(( TEST_FAILURES + 1 )) || TEST_PASSES=$(( TEST_PASSES + 1 ))
done
for TEST_IP in ::1 ::1/128 ::1/0 ::1234 ::bad ::12 1:2:3:4:5:6:7:8 1234:5678:90ab:cdef:1234:5678:90ab:cdef \
1234:5678:90ab:cdef:1234:5678:90ab:cdef/127 1234:5678:90ab::5678:90ab:cdef/64 f:1234:c:ba:240::1 \
1:2:3:4:5:6:1.2.3.4 ::1.2.3.4 ::1.2.3.4/0 ::ffff:1.2.3.4 ;do
! is_ip6 "$TEST_IP" && printf "IP6 test failed, test case '%s' returned invalid\n" "$TEST_IP" && TEST_FAILURES=$(( TEST_FAILURES + 1 )) || TEST_PASSES=$(( TEST_PASSES + 1 ))
done
for TEST_IP in junk . / 0 -1.0.0.0 1.2.c.0 a.0.0.0 " 1.2.3.4" "1.2.3.4 " " " 01.0.0.0 09.0.0.0 0.0.0.01 \
0.0.0.09 0.09.0.0.0 0.01.0.0 0.0.01.0 0.0.0.a 0.0.0 .0.0.0.0 256.0.0.0 0.0.0.256 "" 0 1 12 \
123 1.2.3.4/s 1.2.3.4/33 1.2.3.4/1/1 ;do
is_ip4 "$TEST_IP" && printf "IP4 test failed, test case '%s' returned valid\n" "$TEST_IP" && TEST_FAILURES=$(( TEST_FAILURES + 1 )) || TEST_PASSES=$(( TEST_PASSES + 1 ))
done
for TEST_IP in junk "" : / :1 ::1/ ::1/1/1 :::1 ::1/129 ::12345 ::bog ::1234:345.234.0.0 ::sdf.d ::1g2 \
1:2:3:44444:5:6:7:8 1:2:3:4:5:6:7 1:2:3:4:5:6:7:8/1c1 1234:5678:90ab:cdef:1234:5678:90ab:cdef:1234/64 \
1234:5678:90ab:cdef:1234:5678::cdef/64 ::1.2.3.4:1 1.2.3.4:: ::1.2.3.4j ::1.2.3.4/ ::1.2.3.4:junk ::1.2.3.4.junk ;do
is_ip6 "$TEST_IP" && printf "IP6 test failed, test case '%s' returned valid\n" "$TEST_IP" && TEST_FAILURES=$(( TEST_FAILURES + 1 )) || TEST_PASSES=$(( TEST_PASSES + 1 ))
done
printf "test complete, %s passes and %s failures\n" "$TEST_PASSES" "$TEST_FAILURES"

Answer 3

Most distros come with package iproute2 (name may vary) preinstalled. So you can rely on the command ip for querying the routing table:
ip -6 route get <probe_addr>/128 >/dev/null 2>&1
Even on a machine without appropriate route this delivers rc=0 when the probe is in valid v6-syntax.

Answer 4

valid_ip(){
ip -6 route get "$1"/128 >/dev/null 2>&1
case "$?" in
0|2) return 0
1) return 1
esac
}
i took sgundlach's answer, but needed it on a machine which did not have ip6 connectivity so through testing and reading the manpage I found that I can trust exit code 1 to mean the syntax is invalid, meanwhile 0 is success and 2 is valid syntax but kernel error reported.
from the man page:
Exit status is 0 if command was successful, and 1 if there is a syntax error. If an error was reported by the kernel exit status is 2.