Simple? shell questions from a Perl programmer

tspang · 05-21-2009, 11:34 AM

Hello all, first post.

Due to various reasons, I'm stuck writing some fairly complex scripts in shell (ksh or bash, depending on the system). I have quite a bit of perl experience, but little experience using shell for more than really simple tasks.

If there is some excellent resource that I don't know about that maps common perl procedures to an equavelent shell procedure I'll happily go figure it out on my own. Since I haven't found such a thing, I'll just throw them out here.

Things I'm trying to do in shell, that would be easy in perl that I can't seem to figure out in shell:

1) Take a space deliminated value from awk and assign it to 2 variables.

In perl, I would do something like this:

Code:

$sys_return=`tar -tvf /dev/nst0 | awk '{print $3,$6}'`;
($fileName,$fileSize) = split (/\s/,$sys_return);

In bash, I get stuck doing something like this:

Code:

FNAME=`tar -tvf /dev/nst0 | awk '{print $6}'`
mt -f /dev/nst0 bsfm 1
FSIZE=`tar -tvf /dev/nst0 | awk '{print $3}'`
echo "restoring $FNAME with filesize of $FSIZE"

It must be easier, right?

2) Read a file and look for a value line by line
Perl:

Code:

open(FILE,"list.txt");
foreach $line (<FILE>) {
  $verified = 1 if ($line =~ /$pattern_to_match/);
}

shell:
I'm not even sure, at the moment. I was searching for that when I signed up here.

I see some references to this:

Code:

while read line
do
 if [[ $line == *$pattern_to_match* ]]
 then
  #do something
 fi
done

but none of the sample code I can find actually shows how to pick which file you want to read from.

Thanks for assistance in advance, I'm sure I'll have more questions later!

MensaWater · 05-21-2009, 12:44 PM

1)

Code:

set `tar -tvf /dev/nst0 | awk '{print $3,$6}'`
FSIZE=$1
FNAME=$2

The set command makes the input into positional parameters - since there are 2 items of input you get $1 and $2. (Not to be confused with the $1 and $2 awk itself would have seen in the prior step - $1 = $3 from awk and $2 = $6 from awk).

2)

Code:

FILE=yourfile
while read line
do
 if [[ $line == *$pattern_to_match* ]]
 then
  #do something
 fi
done <$FILE

Since it is a while loop you just redirect in from the file at the end of the loop (done). I set a variable before the loop with name of file so that variable can be used at end but that's not required - you could hard code the file name at the redirect.

chrism01 · 05-22-2009, 01:39 AM

Loop a bit more Perlish

Code:

IFS="
" # hard coded rtn inside quotes
for rec in `cat $file`
do
    if [[ "$rec" =~ "pattern" ]]   # needs bash >= v3  http://tldp.org/LDP/abs/html/bashver3.html
    then
        do_something
    fi
done

ghostdog74 · 05-22-2009, 02:04 AM

Quote:

Originally Posted by tspang

In perl, I would do something like this:

Code:

$sys_return=`tar -tvf /dev/nst0 | awk '{print $3,$6}'`;
($fileName,$fileSize) = split (/\s/,$sys_return);

you could do it inside awk

Code:

tar -tvf /dev/nst0 | awk '{
  filename=$6
  filesize=$3
}
/pattern_to_match/{
  #do something
}'

ntubski · 05-22-2009, 01:56 PM

Quote:

Originally Posted by tspang

2) Read a file and look for a value line by line
Perl:

Code:

open(FILE,"list.txt");
foreach $line (<FILE>) {
  $verified = 1 if ($line =~ /$pattern_to_match/);
}

I'm surprised nobody mentioned this:

Code:

if grep -q "$pattern_to_match" list.txt ; then
  verified=true # or verified=1 if you prefer
else
  verified=false
fi