Sed label question

Shell Programming and Scripting
1

Hi,

I'm trying to run the below sed command someone showed me to remove comments, blank lines and line continuations from a file but I get the following label error: Label too long: :loop;/\\$/N;s/\\\n//;t loop

sed '/^ *#/d' <$1 | sed '/^$/d' | sed ':loop;/\\$/N;s/\\\n//;t loop' | sed 's/[ \t]\+/ /g' > $workfile

From what I can figure out its due to the label length only being allowed to have 8 characters.

Is there another way to write the above sed command so that I don't get a label error?

Thanks,
Jaz

2

change to

-e ':loop' -e '/\\$/N' -e 's/\\\n//;t loop'

or

-e ':loop' -e ';/\\$/N;s/\\\n//;t loop'

solaris sed wants to labels as separate expression..

3

Thanks for the reply ygemici..

Guess I'm going to have to test the sed bit by bit.. The whole command is just giving me a blank output.. In fact just even trying to remove lines with comments gives me a blank output..

sed '/^ *#/d' <$filename > $workfile

**EDIT**
Eventually got this working but now have an awk related question..

Anyone know if there's an -v alternative for Solaris? Doesn't seem to be supported in my version..

By the way its SunOS 5.10

4

what command you tried?
why do you search the alternative?

5

Hi ygemici,

I'm trying to run the below awk command but I'm getting the following... -v: not found

echo " *** split out comma separated 'commands' into separate rows *** "
awk -F| -v OFS='\t' '
{n=split($3,a,/,/)
 for(i=1;i<=n;i++){
   $3=a
   print  
 }
}' <${workfile}_2 | sort > ${workfile}_3

I'm thinking nawk might be whats needed instead but I can't figure out what the nawk command should be.. The man pages aren't that user friendly...

6

You need to escape the pipe character in the -F option because it is a Shell special character. If this is SunOS, most posters advise not using awk and choosing one of the alternatives instead.

7

Should I open a new thread or continue with this one? This all part of one script but they are different issues...

8

try this

# awk 'BEGIN{FS="|";OFS="\t "}{n=split($3,a,",");
for(i=1;i<=n;i++){$3=a;print}
}' <${workfile}_2 | sort > ${workfile}_3

and you can use nawk (or /usr/xpg4/bin/awk) instead of awk in solaris..

regards
ygemici