How to awk interval expression and grouping together? - awk

Question

I am trying to get the measure of cpu time spent on user tasks, system tasks, interrupt handling, io wait etc by parsing the the below output of /proc/stat.
My intent is to retrieve the numerical value in the first line{the one that starts with "cpu " into seperate array elements indexed from 1 through N
[kcube#myPc ~]$ cat /proc/stat
cpu 70508209 48325 12341967 18807644987 671141 0 11736 0 0 0
cpu0 4350458 319 868828 1175271469 23047 0 2397 0 0 0
cpu1 3944197 277 857728 1175822236 16462 0 1025 0 0 0
cpu2 3919468 538 924717 1175628294 136617 0 2270 0 0 0
cpu3 3763268 441 855219 1175968114 43631 0 733 0 0 0
cpu4 3551196 147 856029 1176198902 18392 0 851 0 0 0
cpu5 5394823 1806 997806 1174089493 120122 0 2056 0 0 0
cpu6 3425023 656 839042 1176324091 58718 0 3 0 0 0
cpu7 3167959 189 811389 1176654383 19218 0 2 0 0 0
cpu8 4454976 5046 625657 1175714502 10447 0 26 0 0 0
cpu9 5049813 5365 655732 1175082394 10511 0 30 0 0 0
cpu10 4746872 4727 630042 1175408141 10959 0 28 0 0 0
cpu11 5367186 4684 659408 1174759103 9992 0 23 0 0 0
cpu12 4744405 5940 704282 1175177246 149934 0 714 0 0 0
cpu13 4689816 5954 650193 1175439255 13494 0 5 0 0 0
cpu14 4872185 5479 699429 1175126266 16945 0 898 0 0 0
cpu15 5066558 6748 706459 1174981089 12643 0 669 0 0 0
I have below awk script.
[kcube#myPc ~]$ cat test.awk
{
if ( match($0,/cpu\s(\s[[:digit:]]+){10}$/, ary) ) {
print ary[0]
print ary[1]
}
}
This always gives me the last numeric value in the first line into ary[1].
What I am looking for is to have like :
ary[1] = 70508209
ary[2] = 48325
.
.
so on
I never used interval expression and grouping together. I tried searching for answers but couldn't find one. Can someone help me out?
I'm using GNU Awk 4.0.2

Answer 1

$ cat tst.awk
match($0,/^cpu\s(\s[[:digit:]]+){10}$/,ary) {
print "bad match:", ary[1]
print "bad match:", ary[2]
}
match($0,/^cpu\s+([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)\s([[:digit:]]+)$/,ary) {
print "good match:", ary[1]
print "good match:", ary[2]
}
/^cpu\s/ && split($0,tmp,/[[:digit:]]+/,ary) {
print "good split:", ary[1]
print "good split:", ary[2]
}
$ awk -f tst.awk file
bad match: 0
bad match:
good match: 70508209
good match: 48325
good split: 70508209
good split: 48325
An interval expression defines how many repetitions of the previous expression must exist for the regexp to match, that is all. It is not part of populating capture groups - that is entirely up to use of round brackets enclosing regexp segments. To do what you want you need to either define explicit capture groups for each number, or use split() or similar to create the array based on a regexp that describes each entity you want to be captured.
All of the above uses GNU awk - for the 3rd arg to match() and the 4th arg to split(). Note that you can also just do this with GNU awk for FPAT:
$ awk -v FPAT='[0-9]+' '/^cpu /{for (i=1; i<=NF; i++) print i, $i}' file
1 70508209
2 48325
3 12341967
4 18807644987
5 671141
6 0
7 11736
8 0
9 0
10 0