[SOLVED] replace grep -o with sed or awk

nicomoresi · 08-05-2011, 08:14 AM

I would like to remplace grep -o with sed or awk...

For example:

egrep -oi "Cisco|Motorola|Arris"

-o, --only-matching: Show only the part of a matching line that matches PATTERN.

-i, --ignore-case: Ignore case distinctions in both the PATTERN and the input files.

syg00 · 08-05-2011, 08:19 AM

Good for you.
What have you tried ?.

grail · 08-05-2011, 08:27 AM

Sounds plausible. Good luck

nicomoresi · 08-05-2011, 08:32 AM

Quote:

Originally Posted by syg00

Good for you.
What have you tried ?.

I have tried with

sed -e "s/.*\(Cisco\|Motorola\|Arris\).*/\1/"

But, if doesn't find a matching prints all the line. I would like if doesn't match return empty.

nicomoresi · 08-05-2011, 08:38 AM

syg00 · 08-05-2011, 08:55 AM

sed by default prints the pattern space - that basically means everything it sees.
You're on the right track - look at the manpage. You need "-n" to suppress the printing, but then you need to tell it to print your selected records. Takes some testing to figure it out.

crts · 08-05-2011, 09:34 AM

Hi,

sed is not the way to go for this. Consider this file:

Code:

line1
line2 maTch
line3 alternate
line4 mAtch me and alternatE eol
line5 alternate me and matCh reverse

Several possible matches in one line are hard to identify via RegEx. The Solaris version of 'sed' is also very limited and probably does not have switch for case-insensitive matching.
I don't know about the limitations of awk on solaris but this one worked:

Code:

awk 'BEGIN{IGNORECASE=1}{ for (i=1;i<=NF;i++){if ($i ~ "match|alternate"){ print $i}}}' file

The bold part are your possible matches. Adjust them accordingly.

Hope this helps.

nicomoresi · 08-05-2011, 09:47 AM

Doing very well, except for one thing:

Code:

[root@cpe-172-16-0-136:~]#  echo " aaa Motorolabbbb" | awk 'BEGIN{IGNORECASE=1}{ for (i=1;i<=NF;i++){if ($i ~ "Cisco|Motorola|Arris"){ print $i}}}'
Motorolabbbb
[root@cpe-172-16-0-136:~]#

In this case I need to print just the PATTERN, it's this posible??

nicomoresi · 08-05-2011, 09:54 AM

Anyway this does not work on Solaris:

Code:

 
[root@cpe-172-16-0-137:~]# uname
SunOS
[root@cpe-172-16-0-137:~]# echo " aaa Motorolabbbb" | awk 'BEGIN{IGNORECASE=1}{ for (i=1;i<=NF;i++){if ($i ~ "Cisco|Motorola|Arris"){ print $i}}}'
awk: syntax error near line 1
awk: illegal statement near line 1
awk: syntax error near line 1
awk: bailing out near line 1
[root@cpe-172-16-0-137:~]#

syg00 · 08-05-2011, 09:54 AM

Quote:

Originally Posted by crts

sed is not the way to go for this.

On the contrary, sed is (probably) a better tool for this. Works on my old OpenSolaris too.
But then so does grep -o ... :shrug:

nicomoresi · 08-05-2011, 09:57 AM

For Solaris most be:

Code:

 
[root@cpe-172-16-0-137:~]# uname
SunOS
[root@iway-cnr-01:/opt/iway/commands]# echo " aaa Motorolabbbb" | nawk 'BEGIN{IGNORECASE=1}{ for (i=1;i<=NF;i++){if ($i ~ "Cisco|Motorola|Arris"){ print $i}}}'
Motorolabbbb
[root@iway-cnr-01:/opt/iway/commands]#

nicomoresi · 08-05-2011, 10:01 AM

Quote:

Originally Posted by syg00

On the contrary, sed is (probably) a better tool for this. Works on my old OpenSolaris too.
But then so does grep -o ... :shrug:

Code:

[root@iway-cnr-01:/opt/iway/commands]# uname
SunOS
[root@iway-cnr-01:/opt/iway/commands]# echo " aaa Motorola bbbb" | egrep -oi "Cisco|Motorola|Arris"
egrep: illegal option -- o
usage: egrep [ -bchilnsv ] [ -e exp ] [ -f file ] [ strings ] [ file ] ...
[root@iway-cnr-01:/opt/iway/commands]#

nicomoresi · 08-05-2011, 10:03 AM

Quote:

Originally Posted by syg00

On the contrary, sed is (probably) a better tool for this. Works on my old OpenSolaris too.
But then so does grep -o ... :shrug:

syg00 Do you know how to do it with sed?

crts · 08-05-2011, 10:06 AM

Quote:

Originally Posted by syg00

On the contrary, sed is (probably) a better tool for this. Works on my old OpenSolaris too.
But then so does grep -o ... :shrug:

How would you make a case-insensitive match for the 's' command on Solaris? According to my documentation this is a GNU extension, so I assume that it is not available on Solaris. Is it possible that you have also GNU-sed installed on your Solaris?

anomie · 08-05-2011, 10:20 AM

Indeed, Solaris' standard grep sucks. (Sorry if that's offensive to any Solaris admins.)

Code:

$ uname -sr
SunOS 5.10

$ grep -o 'simple' /etc/passwd
grep: illegal option -- o
Usage: grep -hblcnsviw pattern file . . .

-------

@nicomoresi: Just use Perl.

Code:

#!/opt/csw/bin/perl

use warnings ;

while(<>) {

  @matches = /(Cisco|Motorola|Arris)/i ;

  print "@matches\n" if(@matches)  ;

}

That will read from stdin, and print case-insensitive matches. As written, it only prints one occurrence of cisco, motorola, or arris per line. (If one or more of those could occur several times on a line, the script needs a little tweaking.)

-------

P.S. It's a good idea to post your OS / version when starting a thread! This is not even a Linux question.

-------

BTW, should you want to match more than once per line, it's a simple matter of the following. (Just tested it.)

Code:

--- match.pl    Fri Aug  5 10:26:29 2011
+++ more-match.pl       Fri Aug  5 10:26:23 2011
@@ -4,7 +4,7 @@
 
 while(<>) {
 
-  @matches = /(Cisco|Motorola|Arris)/i ;
+  @matches = /(Cisco|Motorola|Arris)/ig ;
 
   print "@matches\n" if(@matches)  ;