Calling Unixguy and the gods of Bash! a Cut, Sed, Awk query

ally_uk

One for the gods of Centos / Bash probably are a few ways of resolving this issue.

ip addr show - shows IP Address and loads of information

I am purely just interested in having the IP adress to be used in a script.

So if I narrow results down further:

ip addr show | grep 192

Produces:

inet 192.168.1.5/24 brd 192.168.1.255 scope global enp03s

Now I could run further greps to narrow it down but how do I just get the line 192.168.1.5

Am I right in thinking Cut? Sed? Awk? please elaborate and help a brother out with your wisdom
smile emoticon
Like · Comment · Share

networker050184

You can do grep -o pretty easy. Probably want to do -Eo and have a regex match any IP in case the IP doesn't start with 192.

ipchain

Try this: echo "inet 192.168.1.5/24 brd 192.168.1.255 scope global enp03s" | cut -f 2 -d " " | cut -f 1 -d "/"

It should give you what you need.

Expect

a really quick and dirty code, there are better ways than this but it works.

ip addr show | grep global | awk -F' ' '{print $2}'  | awk -F'/'  '{print $1}'

ChickenNuggetz

Depends on what you're trying to accomplish. If you just want to set a variable with the IP address of the machine (i.e. this could be a generic script being used on any number of hosts, so the IP is variable), then I'd probably use a combination of awk and sed.

Awk can let you filter by "column," so you can use awk to isolate the "column" of your output to just the IP.
Sed will let you trim off the "/24" on the end.

I wont tell you how exactly to do this, because half the fun is in figuring it out on your own! Good luck!

ChickenNuggetz

Expect wrote: »

a really quick and dirty code, there are better ways than this but it works.

ip addr show | grep global | awk -F' ' '{print $2}'  | awk -F'/'  '{print $1}'

That's the beauty of Linux, there's so many ways to get things done!

ally_uk

Thanks I got this aswell:

ip addr show | awk '/192/ {print $2}'

networker050184

What happens if the address is a 10. or 172. though? Probably want that to match on a variable length [1-9]{1,3}\. or something along those lines.

ally_uk

Yeah you are right if the I.P address isn't 192 I am not sure about how to go about what you are saying though.

networker050184

ip addr show | grep -Eo '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}'

That will pull out the IPs. Now you just need to get just the IP from the interface you want. The ip addr show command will give you all interface IPs.

ally_uk

That's way above my level

I need to go back to basics starting with the cut command, Like fields and rows I don't really understand it i'm reading tutorials but it's hurting my head lol

ally_uk

I need to take a baby steps approach to solidify concepts do you mind guiding me?


The first thing is say if I have lot's of text on the screen from a command output How would I go about just printing the 1st line of output and nothing else?


I now have a line of text say for example:


1: lo: <LOOPBACK,UP,LOWER,_UP> mtu 65536 qdisc noqueue state UNKNOWN


Using the cut utility I only want to display lo: my question is how the hell do you go about working out fields, rows, tables, sections when the formatting is off like that and how the hell do you know what to cut? god i'm so confused.....

is the 1: classed as field1 or is the : field2?


Thank you for help guys

DoubleNNs

ally_uk wrote: »

The first thing is say if I have lot's of text on the screen from a command output How would I go about just printing the 1st line of output and nothing else?

head -n 1

prints the 1st line. So does

awk 'NR==1 {print;exit}'

head would be slightly faster, but the benefit of the awk snippet is you can set that "1" to any line number you want. Change it to 25 and it'll print just the 25th line.

Using the cut utility I only want to display lo: my question is how the hell do you go about working out fields, rows, tables, sections when the formatting is off like that and how the hell do you know what to cut?

You can set your field delimiter to whatever you want. I think by default it's simply whitespace. So field 1 is the beginning of the line towards the first whitepace char (tab/space). Field 2 would be the text directly following the first space but before the 2nd space.

If you're working in a comma-deliminated file then you can make the field every time a new comma comes. You can even set the field to a number or letter. But I'm pretty sure w/ cut, the delimiter has to be a single char. If your text file has a string w/ multiple-char-length you want to set as your delimiter, you can use a sed substitution to turn that multiple-char-string into a single char - something that won't show up elsewhere on the line.

DoubleNNs

You can also set the entire line to a variable and cut off the beginning and front of the variable strings. I think the expansion is something similar to ${,,varname} and ${varname##}.

To be honest I don't remember the exact syntax off head. Maybe someone can correct it.

ally_uk

Wow thank you for the response you really do know your stuff you are i'm getting it I think

using this line as a example and working with cut:

(example1)

1: lo: <LOOPBACK,UP,LOWER,_UP> mtu 65536 qdisc noqueue state UNKNOWN

What would the cut command look like if I wanted to display just: ( alternative I could do this with sed? )

(example 2)

1: lo:

<LOOPBACK,UP,LOWER,_UP> mtu 65536 qdisc noqueue state UNKNOWN

(example 3)

Or display a specific part of the line say: This could be done with a grep?:

mtu 65546


Thanks for the guidance and excuse the amount of examples I find it helps me to cement it into my grey matter

DoubleNNs

grep doesn't print only a part of a line. The grep command will always grab the entire line. You can make grep grab multiple lines surrounding the specified string/pattern, but you can't (to my knowledge) make grep grab only part of the line. You would usually use cut/awk to break up lines and grab specific output from it.

cut would be faster - which is important when working w/ large files. awk is much more flexible.


Try to edit/reformat your examples. They didn't seem to come out right and I can't quite make out what your questions are.

DoubleNNs

I'm double posting a lot (procrastinating on last min studying) but I just understood what the "grep -o" option does. It DOES allow you to print just the matching part of the line. You'd want to use regex w/ the grep -o option to make it useful. (I'm simply echoing the very 1st response in this thread fromnetworker050184unfortunately.)

Additionally, found out Linux has the hostname -I command. Try it.
Depending on what you're trying to accomplish (and what platform you're coding on) that might be all you need.

ally_uk

Good morning thank you for your replies. I am going to be taking things back to basics and will be spending some time learning the basics of cut, sed, awk, grep.

I gave myself a migraine yesterday from over thinking haha I was getting really confused with the cut command, Like I understand that it can specifically remove characters and fields and you can use a delimiter that is all good if you are working with a nice neatly arranged file such as

one : two : three : four : five

but what if you output a command and it's a bit all over the place such as

one twothreee fourfivvvvvvvesixxxxxx

How does cut know to remove remove specific parts say for example if I wanted to output nothing else apart from the threee in the above example. or would I use a more suitable tool such as awk or sed or something.

UnixGuy

why confuse yourself with 'cut' ? I'd stick with "awk".

Use the solution by @Expect, it works. Try other variations with awk as well. Stick with one tool, master it, then move to another tool IF necessary. I don't use 'cut' at all, awk does everything just fine

ally_uk

Well that was AWKward! see what I did there?

I was thinking that aswell I will use cut when files are neatly organized i.e one: two: three:
actually your right I could just use awk why am I even having nightmares about cut....


Thank you UnixGuy, I have another brain teaser for yourself on Centos 7 I want to find out my I.P address for a script. A guy posted me a solution for centos 6.



Find the list of active ethernet interfaces.....

route -n|grep -v Destination|awk '{print $8}'|sort -u

Then for each of these (eth0 shown for example):

ifconfig eth0|sed -n -e '/inet /p'|awk '{print $2}'

---

I haven't tested on Centos 7 but I assume you would just change the ifconfig? or is there a better way to achieve the above

Many Thanks