perl chop numbers off os a string

casperdaghost · 08-12-2010, 11:24 PM

I am trying to strip the numbers from the 5th columns and just get letters - in perl - i tried using subsrt, chop. and a regex - anybody got any ideas?
casper@casper-laptop:~$ more myfile
procmonES snoc active 21943 -1 eustk100 21943 ibutil
procmonUT snoc active 23712 -1 tus100 23712 ibutil
procmonUS snoc active 23796 -1 topsrv1 23796 ibutil
procmonHK snoc active 26602 -1 hibmis100 26602 ibutil
procmonEU snoc active 30129 -1 stopsrv1 30129 ibutil
procmonUI snoc active 32013 -1 usint100 20026 ibutil
procmonUM snoc active 33231 -1 usmrg100 20007 ibutil
procmonFU snoc active 34436 -1 usfrx100 20015 ibutil
procmonEI snoc active 35387 -1 euint100 20019 ibutil
procmonFE snoc active 38268 -1 eufrx100 20024 ibutil
procmonUU snoc active 38377 -1 usstk102 20007 ibutil
procmonNY snoc active 40456 -1 nystk100 20016 ibutil
procmonIL snoc active 43301 -1 cib100 20010 ibutil
procmonSM snoc active 47569 -1 ussmm100 20016 ibutil

casper@casper-laptop:~$ more myfile | perl -ane 'print $F[5]."\n"'
eustk100
tus100
topsrv1
hibmis100
stopsrv1
usint100
usmrg100
usfrx100
euint100
eufrx100
usstk102
nystk100
cib100
ussmm100

what i want to get is
eustk
tus
topssrv
hibmis
stopsrv
usint
usmrg
usfrx
euint
eufrx
euint
eufrx
usstk
nystk
cib
ussmm

i tired appending a regex perl -nle 'print /(\w{1,20})\d{1-20} but it did not work.

casper@casper-laptop:~$ more myfile | perl -ane 'print $F[5]."\n"' | perl -nle 'print /(\w{1,20})\d{1,20}/'

Tinkster · 08-13-2010, 12:14 AM

Code:

$ cat casper 
procmonES snoc active 21943 -1 eustk100 21943 ibutil
procmonUT snoc active 23712 -1 tus100 23712 ibutil
procmonUS snoc active 23796 -1 topsrv1 23796 ibutil
procmonHK snoc active 26602 -1 hibmis100 26602 ibutil
procmonEU snoc active 30129 -1 stopsrv1 30129 ibutil
procmonUI snoc active 32013 -1 usint100 20026 ibutil
procmonUM snoc active 33231 -1 usmrg100 20007 ibutil
procmonFU snoc active 34436 -1 usfrx100 20015 ibutil
procmonEI snoc active 35387 -1 euint100 20019 ibutil
procmonFE snoc active 38268 -1 eufrx100 20024 ibutil
procmonUU snoc active 38377 -1 usstk102 20007 ibutil
procmonNY snoc active 40456 -1 nystk100 20016 ibutil
procmonIL snoc active 43301 -1 cib100 20010 ibutil
procmonSM snoc active 47569 -1 ussmm100 20016 ibutil

$ perl -ane '$F[5]=~s/[0-9]//g; print "$F[5]\n"'  casper
eustk
tus
topsrv
hibmis
stopsrv
usint
usmrg
usfrx
euint
eufrx
usstk
nystk
cib
ussmm

Cheers,
Tink

casperdaghost · 08-13-2010, 11:41 AM

Tink this is beautiful

so the -n loops through but does not print the output -a splits the string in the fith column into individual characters (ex cib100 = c-i-b-1-0-0) and if a character =~ (matches) a number it substitutes is with a nothing

$ perl -ane '$F[5]=~s/[0-9]//g; print "$F[5]\n"' casper

which part of the script acts on the file first - the print "$F[5]\n"' or the perl -ane '$F[5]=~s/[0-9]//g; ?\
i am just hazy on if the script seperates the string into single characters - how does the script puts the string back together.

PradeepKr · 08-13-2010, 03:53 PM

I am afraid this would remove the numbers from within the word also, if any.
eg, euf1rx100 would be printed as eufrx

Les Windoze · 08-13-2010, 05:46 PM

Generic algorithm:
I'd read each line into an array; grab the desired element of the array and
walk the string starting at the end, throwing away values which test true for numbers.
Hit a character and from there forward is the desired result.

Tinkster · 08-13-2010, 11:59 PM

Quote:

Originally Posted by casperdaghost

Tink this is beautiful

so the -n loops through but does not print the output -a splits the string in the fith column into individual characters (ex cib100 = c-i-b-1-0-0) and if a character =~ (matches) a number it substitutes is with a nothing

$ perl -ane '$F[5]=~s/[0-9]//g; print "$F[5]\n"' casper

which part of the script acts on the file first - the print "$F[5]\n"' or the perl -ane '$F[5]=~s/[0-9]//g; ?\
i am just hazy on if the script seperates the string into single characters - how does the script puts the string back together.

Nope regarding the -a ... the array gets split along the
whitespace, and the snippet takes each line, splits the
columns into the array element. $F[5] is column 6.
The regex 's/[0-9]//g' simply strips ANY number from
the 6th column (of every line), and the print then
prints it.

basically -an is shorthand for

Code:

while (<>) {
    @F = split(' ');
}

All the ops on the commandline (following -e) get
executed in sequence AFTER the split.

Cheers,
Tink