Use regex or unix to select specific columns

Question

I am using Runtime.getRuntime().exec("df") to get the space details for my partitions.

Filesystem     1K-blocks    Used Available Use% Mounted on
/dev/sdb12      41022792 4219168  34713128  11% /
udev             8103980       4   8103976   1% /dev
tmpfs            3245332     924   3244408   1% /run
none                5120       0      5120   0% /run/lock
none             8113328     160   8113168   1% /run/shm
/dev/sdb2         262144   28584    233560  11% /boot/efi

How can I use regex or UNIX commands to get as the output just the 2nd,3rd and 4th columns - changing the headers as follows - Available Memory, Used Memory and Total memory exactly in this order.
So the desired output is :

Available    Used     Total
34713128   4219168    41022792
8103976       4       8103980
3244408      924      3245332
             ...

Thank you!

This is most definitely a job for awk, not regex. Would that be OK? — Floris
– Floris, Commented Nov 3, 2013 at 14:51
as long as I can execute it all in getRuntime().exec(...), Yes! — Iulian Rosca
– Iulian Rosca, Commented Nov 3, 2013 at 14:53
Do you want the text centered like that for the middle column, or is right-aligned preferred (as in the original command) — Floris
– Floris, Commented Nov 3, 2013 at 15:04

Floris · Accepted Answer · 2013-11-03 15:02:50Z

2

The following rather long string does what you ask:

df | awk 'BEGIN {print "   Available        Used       Total";} {if (NR>1)  printf("%12d%12d%12d\n", $4, $3, $2);}'

Explanation:

| awk          take the output of the df command as input to awk
BEGIN          do this first: print headers. Note the use of spaces to align things
if (NR > 1)    skip "record 1" = do not do anything with the headers
printf()       do formatted printing
%12d           print integer in fixed width of 12 characters (maintains alignment

Result on my machine:

   Available        Used       Total
    71644456   551306776   623463232
           0         372         372
           0           0           0
           0           0           0
   407510704   377391216   784901920
   618400760   549288520  1167689280

Not terribly useful without the name of the disks, but this is what you were asking...

answered Nov 3, 2013 at 15:02

Floris

46.6k7 gold badges73 silver badges128 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Floris Over a year ago

Happy to help - clearly the above can be modified in many ways and I hope you will… awk is really the perfect tool for this kind of thing, well worth becoming familiar with it. It can be obscure, but is surprisingly powerful. Best reference I found is the grymoire (unusual word meaning "book of spells") - written by Bruce Barnett. Google grymoire awk.

asthasr · Accepted Answer · 2013-11-03 14:53:11Z

0

You can use awk to do this:

df | awk '{print ""$4" "$3" "$2""}'

answered Nov 3, 2013 at 14:53

asthasr

9,4872 gold badges31 silver badges44 bronze badges

1 Comment

asthasr Over a year ago

I modified the order. I'd not try to do a modification of the headers in a Unix command; just do the substitution against the resulting Java string.

maaartinus · Accepted Answer · 2013-11-03 14:59:34Z

0

For me this works

df | perl -pe \
  's!^(?:\S+\s+){2}(\S+)\s+(\S+).*!"$1\t$2\t" . ($1 eq "Used" ? "Total" : $1 + $2)!e'

It's not nice: It finds the columns as non-blanks separated by blanks, which is a bit fragile. It computes the last column as the sum or "Total" based on the column name. Rather fragile, I'd say.

answered Nov 3, 2013 at 14:59

maaartinus

46.8k40 gold badges176 silver badges343 bronze badges

Collectives™ on Stack Overflow

Use regex or unix to select specific columns

3 Answers 3

1 Comment

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related