A newbie function question...

fatsheep · 11-05-2006, 02:16 PM

I'm writing a very simple package management script for use with CLFS which I won't explain the details of it but I need to make some modifications to the output of "du" and then flip the output upside down. I have made my modifications, I just need to flip the output of du upside down (so the first line would be last and the last line would be first).

I decided to use an existing gnu script and insert it as a function named "reverse". When I call this function with the command "reverse <FILENAME>" it outputs the contents of the file flipped. This is what I want it to do. But if I use the command "cat <FILENAME> | reverse" it outputs "bash - file-name". How would I modify this script so that it can use a pipe correctly?

Any help would be appreciated. Also, feel free to explain how the below function works if you feel like it.

Code:

function reverse
{
	FILE="$1"
	if [ $# -eq 0 ]; then
	  echo "$(basename $0) - file-name"
	  exit 1
	fi

	textArray[0]="" # hold text
	c=0 # counter
	# read whole file in loop
	while read line
	do
	  textArray[c]=$line # store line
	  c=$(expr $c + 1) # increase counter by 1
	done < $FILE
	# get length of array
	len=$(expr $c - 1 )

	# use for loop to reverse the array
	for (( i=$len; i>=0; i-- ));
	do
	  echo ${textArray[$i]}
	done
}

soggycornflake · 11-05-2006, 02:42 PM

There is already a command to do this: rev.

As for your script, you are explicity testing for an argument (which will be absent when reading from stdin):

Code:

if [ $# -eq 0 ]; then
  echo "$(basename $0) - file-name"
  exit 1
fi

I don't use bash, so I'm not sure how best to differentiate between the filename and stdin. Zsh can reverse the file with a one-liner,

Code:

print -l ${(fOa)"$(<file)"}

fatsheep · 11-05-2006, 03:01 PM

Quote:

Originally Posted by soggycornflake

There is already a command to do this: rev.

As for your script, you are explicity testing for an argument (which will be absent when reading from stdin):

Code:

if [ $# -eq 0 ]; then
  echo "$(basename $0) - file-name"
  exit 1
fi

I don't use bash, so I'm not sure how best to differentiate between the filename and stdin. Zsh can reverse the file with a one-liner,

Code:

print -l ${(fOa)"$(<file)"}

Thanks for the reply but rev doesn't do what I want. I only want the order of the lines reversed, not the lines themselves. For example, if I have a file with these contents:

Quote:

Line 1
Line 2
Line 3

Then after I apply the reverse function I get

Quote:

Line 3
Line 2
Line 1

Which is what I want. The rev command would give me:

Quote:

3 eniL
2 eniL
1 eniL

soggycornflake · 11-05-2006, 03:34 PM

Oops, sorry, I mis-grepped the meaning of rev.

soggycornflake · 11-05-2006, 03:50 PM

Anyway, the zsh snippet I gave should work:

Code:

#!/bin/zsh

if [[ -n $1 ]]; then
        print -l ${(fOa)"$(<$1)"}
else
        array=()
        while read line; do
                array+=$line
        done
        print -l ${(Oa)array}
fi

If you must use bash, you might need to duplicate the while loop and wrap it in an if test:

e.g.

Code:

if [[ -n $1 ]]; then
        while read line
        do
          textArray[c]="$line" # store line
          let "c+=1"
        done < $1
else    
        while read line
        do
          textArray[c]="$line" # store line
          let "c+=1"
        done
fi

Note that I replaced 'c=$(expr $c - 1 )' with the simpler 'let "c+=1"'. And, of course, remove the if test on $# at the start.

keefaz · 11-05-2006, 03:56 PM

Or use sed, example given with "info sed" :

Code:

#!/usr/bin/sed -nf
    
# reverse all lines of input, i.e. first line became last, ...
     
# from the second line, the buffer (which contains all previous lines)
# is *appended* to current line, so, the order will be reversed
1! G
     
# on the last line we're done -- print everything
$ p
     
# store everything on the buffer again
h

soggycornflake · 11-05-2006, 04:05 PM

Yes, of course! I momentarily toyed with mentioning sed but couldn't remember how to do it offhand (and was too lazy to RTFM). That's nice, keefaz, much better than arsing around with a shell array.

fatsheep · 11-05-2006, 04:27 PM

Quote:

Originally Posted by keefaz

Or use sed, example given with "info sed" :

Code:

#!/usr/bin/sed -nf
    
# reverse all lines of input, i.e. first line became last, ...
     
# from the second line, the buffer (which contains all previous lines)
# is *appended* to current line, so, the order will be reversed
1! G
     
# on the last line we're done -- print everything
$ p
     
# store everything on the buffer again
h

I saved the above as a executable file named "sed_reverse" under ~/bin and called it like so:

Quote:

cat du1.txt | sed_reverse

And I got this error:

Quote:

bash: /home/ubuntu/bin/sed_reverse: /usr/bin/sed: bad interpreter: No such file or directory

I think it would be easier just to have a regular sed command inside my script. Could I do that instead?

soggycornflake · 11-05-2006, 04:37 PM

Maybe your sed is in /bin or /usr/local/bin? The "hash-bang" line must specify the correct path, the kernel doesn't search for the command. But yes, you could just call sed from the script, i.e.

Code:

#!/bin/sh
exec sed -nf ...

(The exec is optional, but prevents having a shell hanging around just waiting for sed to exit.)

One last point, redirection is a little more efficient than cat, e.g.

Code:

sed_reverse < du1.txt

(Though I realise it was just an example, you could of course, avoid the tmp file and pipe the output of du straight to the script.)

osor · 11-05-2006, 04:42 PM

There's also tac (cat backwards).

fatsheep · 11-05-2006, 04:44 PM

Quote:

Originally Posted by soggycornflake

Maybe your sed is in /bin or /usr/local/bin? The "hash-bang" line must specify the correct path, the kernel doesn't search for the command. But yes, you could just call sed from the script, i.e.

Code:

#!/bin/sh
exec sed -nf ...

(The exec is optional, but prevents having a shell hanging around just waiting for sed to exit.)

One last point, redirection is a little more efficient than cat, e.g.

Code:

sed_reverse < du1.txt

(Though I realise it was just an example, you could of course, avoid the tmp file and pipe the output of du straight to the script.)

When I pipe the output of du to

Code:

sed -nf 1! G p h

I get this error:

Quote:

sed: couldn't open file 1!: No such file or directory

soggycornflake · 11-05-2006, 04:46 PM

Quote:

There's also tac (cat backwards).

Damn, I'd completely forgotten about tac. Y'know, they should really supply manuals for these things. Maybe in a paginated format.

Quote:

When I pipe the output of du to
Code:

sed -nf 1! G p h

I get this error:
Quote:
sed: couldn't open file 1!: No such file or directory

Oops, forgot to check the options, you need to get rid of the -f and quote the sed 'script', and separate the commands with ; or newlines, e.g.

Code:

#!/bin/sh
sed -n '1!;G;$p;h;' $1

That should work now. But, I think osor has hit on the optimal solution.

fatsheep · 11-05-2006, 05:08 PM

Quote:

Originally Posted by soggycornflake

Damn, I'd completely forgotten about tac. Y'know, they should really supply manuals for these things. Maybe in a paginated format.

Oops, forgot to check the options, you need to get rid of the -f and quote the sed 'script', and separate the commands with ; or newlines, e.g.

Code:

#!/bin/sh
sed -n '1!;G;$p;h;' $1

That should work now. But, I think osor has hit on the optimal solution.

Yea I think I'll just use tac. I missed that post... Thanks guys.