How to remove all but the last 3 parts of FQDN? - bash

Question

I have a list of IP lookups and I wish to remove all but the last 3 parts, so:
98.254.237.114.broad.lyg.js.dynamic.163data.com.cn
would become
163data.com.cn
I have spent hours searching for clues, including parameter substitution, but the closest I got was:
$ string="98.254.237.114.broad.lyg.js.dynamic.163data.com.cn"
$ string1=${string%.*.*.*}
$ echo $string1
Which gives me the inverted answer of:
98.254.237.114.broad.lyg.js.dynamic
which is everything but the last 3 parts.
A script to do a list would be better than just the static example I have here.
Using CentOS 6, I don't mind if it by using sed, cut, awk, whatever.
Any help appreciated.
Thanks, now that I have working answers, may I ask as a follow up to then process the resulting list and if the last part (after last '.') is 3 characters - eg .com .net etc, then to just keep the last 2 parts.
If this is against protocol, please advise how to do a follow up question.

Answer 1

if parameter expansion inside another parameter expansion is supported, you can use this:
$ s='98.254.237.114.broad.lyg.js.dynamic.163data.com.cn'
$ # removing last three fields
$ echo "${s%.*.*.*}"
98.254.237.114.broad.lyg.js.dynamic
$ # pass output of ${s%.*.*.*} plus the extra . to be removed
$ echo "${s#${s%.*.*.*}.}"
163data.com.cn
can also reverse the line, get required fields and then reverse again.. this makes it easier to use change numbers
$ echo "$s" | rev | cut -d. -f1-3 | rev
163data.com.cn
$ echo "$s" | rev | cut -d. -f1-4 | rev
dynamic.163data.com.cn
$ # and easy to use with file input
$ cat ip.txt
98.254.237.114.broad.lyg.js.dynamic.163data.com.cn
foo.bar.123.baz.xyz
a.b.c.d.e.f
$ rev ip.txt | cut -d. -f1-3 | rev
163data.com.cn
123.baz.xyz
d.e.f

Answer 2

echo $string | awk -F. '{ if (NF == 2) { print $0 } else { print $(NF-2)"."$(NF-1)"."$NF } }'
NF signifies the total number of field separated by "." and so we want the last piece (NF), last but 1 (NF-1) and last but 2 (NF-2)

Answer 3

$ echo $string | awk -F'.' '{printf "%s.%s.%s\n",$(NF-2),$(NF-1),$NF}'
163data.com.cn
Brief explanation,
Set the field separator to .
Print only last 3 field using the awk parameter $(NF-2), $(NF-1),and $NF.
And there's also another option you may try,
$ echo $string | awk -v FPAT='[^.]+.[^.]+.[^.]+$' '{print $NF}'
163data.com.cn

Answer 4

It sounds like this is what you need:
awk -F'.' '{sub("([^.]+[.]){"NF-3"}","")}1'
e.g.
$ echo "$string" | awk -F'.' '{sub("([^.]+[.]){"NF-3"}","")}1'
163data.com.cn
but with just 1 sample input/output it's just a guess.
wrt your followup question, this might be what you're asking for:
$ echo "$string" | awk -F'.' '{n=(length($NF)==3?2:3); sub("([^.]+[.]){"NF-n"}","")}1'
163data.com.cn
$ echo 'www.google.com' | awk -F'.' '{n=(length($NF)==3?2:3); sub("([^.]+[.]){"NF-n"}","")}1'
google.com

Answer 5

Version which uses only bash:
echo $(expr "$string" : '.*\.\(.*\..*\..*\)')
To use it with a file you can iterate with xargs:
File:
head list.dat
98.254.237.114.broad.lyg.js.dynamic.163data.com.cn
98.254.34.56.broad.kkk.76onepi.co.cn
98.254.237.114.polst.a65dal.com.cn
iterating the whole file:
cat list.dat | xargs -I^ -L1 expr "^" : '.*\.\(.*\..*\..*\)'
Notice: it won't be very efficient in large scale, so you need to consider by your own whether it is good enough for you.
Regexp explanation:
.* \. \( .* \. .* \. .* \)
\___| | | | |
| \------------------------/> brakets shows which part we extract
| | |
| \-------/> the \. indicates the dots to separate specific number of words
|
|
-> the rest and the final dot which we are not interested in (out of brakets)
details:
http://tldp.org/LDP/abs/html/string-manipulation.html -> Substring Extraction