[SOLVED] Extract Multiple Strings per Line

ckinninger · 03-19-2011, 09:34 PM

Hello Guys,

Not sure if sed, awk or maybe even grep would work at all...

I need to extract all 16 and 17 digit numbers from a multiline text file. They need to stay on their original lines.

Some lines may have multiple string matches

Input file:

d-1 1234567890123456 random_text 2543210987654321 random 3234567890123456 4543210987654321
d-2 1234567890123456 25432109876543210 random_text
d-4 1234567890123456 random 2543210987654321
d-5 12345
d-6 random_text

Output

d-1 1234567890123456 2543210987654321 3234567890123456 4543210987654321
d-2 1234567890123456 25432109876543210
d-4 1234567890123456 2543210987654321
d-5
d-6

First string of input file is always starts with "d-" and I want to keep that if possible.

Can you guys help me get a string going? I've spent time reading on sed but no luck. Just keep first or last match.

Thanks.

savona · 03-19-2011, 09:49 PM

You should be able to extract what you need with egrep. I haven't tested, this is just from my head.

egrep '([0-9]\{16\})|([0-9]\{17\})' filename

savona · 03-19-2011, 09:56 PM

Actually now that I think about it the first regular expression would catch both 16 and 17 digit strings. So this should find both:

grep '[0-9]\{16\}' file

ckinninger · 03-19-2011, 10:14 PM

Quote:

Originally Posted by savona

Actually now that I think about it the first regular expression would catch both 16 and 17 digit strings. So this should find both:

grep '[0-9]\{16\}' file

Thanks for the input...

What that's doing is finding the lines and saving the original lines if anything in that line is 16 digits.

I am trying to do is save the "d-123" number and follow that with any string matches on the same line. Strip out the garbage (non string matches).

Telengard · 03-19-2011, 10:22 PM

I think this does what you want. You should definitely review the code and carefully test it before trusting it. I am no expert and there is most likely a better solution to be found.

Code:

$ gawk --re-interval '{ printf $1 " " ; for ( i = 2 ; i <= NF ; i ++ ) { if ( $i ~ /^[0-9]{16,17}$/ ) { printf $i " " } } printf "\n" }' input.txt
d-1 1234567890123456 2543210987654321 3234567890123456 4543210987654321
d-2 1234567890123456 2543210987654321
d-4 1234567890123456 2543210987654321
d-5
d-6
$

HTH

Edit

http://www.gnu.org/software/gawk/man...ode/index.html

ckinninger · 03-19-2011, 10:29 PM

[QUOTE=Telengard;4296620]I think this does what you want. You should definitely review the code and carefully test it before trusting it. I am no expert and there is most likely a better solution to be found.

Code:

$ gawk --re-interval '{ printf $1 " " ; for ( i = 2 ; i <= NF ; i ++ ) { if ( $i ~ /^[0-9]{16,17}$/ ) { printf $i " " } } printf "\n" }' input.txt
d-1 1234567890123456 2543210987654321 3234567890123456 4543210987654321
d-2 1234567890123456 2543210987654321
d-4 1234567890123456 2543210987654321
d-5
d-6
$

Sweet! Worked great on my 6 line test file. Now going for the big test... will let you know. Thanks a lot.

ckinninger · 03-19-2011, 10:50 PM

[QUOTE=Telengard;4296620]

Code:

$ gawk --re-interval '{ printf $1 " " ; for ( i = 2 ; i <= NF ; i ++ ) { if ( $i ~ /^[0-9]{16,17}$/ ) { printf $i " " } } printf "\n" }' input.txt
$

Worked great. I really appreciate it. I thought I could handle it on my own but 6 hours later nothing. Now I can move forward.

Telengard · 03-20-2011, 12:38 AM

Quote:

Originally Posted by ckinninger

I really appreciate it.

I'm very pleased to know you found my post helpful. If this solution is acceptable, then please consider using the thread tools option to mark this thread solved.

Happy Linux-ing