Capturing a word in a line of text whose placement moves

[Staci Tribelhorn]

I'm trying to write a script and need assistance in figuring out how to capture a word in each line of text when that word changes position. Below are the result of my cat | grep 'CHECK: #'
What I want to do then is to extract just the check number from each line. The problem is that the placement of the check number varies depending on the length of the name, so just using the awk print function doesn't always capture the correct field. Is there a command that I can use to locate the field # so that I can then grab the proceeding check number?

OWNER: 59020 1949 INC CHECK: #185841 06/17/05

OWNER: 30252 1995 INCOME PROGRAM LIMITED CHECK: #185842 06/17/05
OWNER: 12129 A G ANDRIKOPOULOS RESOURCES INC CHECK: #185843 06/17/05
OWNER: 52984 A V M INC CHECK: #185844 06/17/05

[Pete Randall]

How about using awk and setting the FS (field separator) to # ?


Pete

[Mel Burslan]

I am not a PERL expert but I know this is a task for using PERL on the very basic level. The command you are looking for is 'index' but again I am the worst person to advise you when the subject is PERL.

HTH

[Mel Burslan]

well, just to give you an idea

create a file findpos.pl as follows

#!/usr/bin/perl
$a=;
$b=index $a, "word3";
printf $b

chmod 755 findpos.pl

echo "word1 word2 word3 word4 word5" | ./findpos.pl

12


hope this helps

[A. Clay Stephenson]

Okay, let's do it closer to the real Perl Cowboy Way:

perl -n -e 'if (/#(\d+)/) { print $1,"\n"; }' < myfile

You might want to make the pattern match a bit more foolproof by:
perl -n -e 'if (/CHECK:#(\d+)/) { print $1,"\n"; }' < myfile


The \d+ says match all the digits you can; expressions after the matching string in paren's are assigned $1 $2 ... but we are looking for only the number so there is just one paren'ed pattern. Perl is really pattern matching on steroids. I haven't tested this but if I didn't make no booboo's this here should work.

[RAC_1]

Fast and dirty.

cat file | tr " " "\n" | egrep -i "check|#[0-9]"

Anil

[H.Merijn Brand (procura]

Even shorter

# perl -nle'/CHECK:\s*#(\d+)/ and print$1' file

Enjoy, Have FUN! H.Merijn

[harry d brown jr]

sed:

cat filename | sed -e "s/\(.*\)CHECK: \#\([0-9]\{1,10\}\) \([0-9].*$\)/\2/"

live free or die
harry d brown jr

[Staci Tribelhorn]

Thanks for your suggestions. The perl commands worked best for me particular issue.

[Staci Tribelhorn]

Thank you all of your suggestions. The perl command worked best.