[SOLVED] shell scripts: storing result of a command into a variable

newbiecolonopenparens · 05-22-2012, 10:49 PM

hi there guys, I have a question to ask.

Let's say I typed in the command: find $1 -type d | wc -l
to find the number of directories. This command, from what I discovered in previous threads, gives 1 extra directory. That's not what I want. I want to do subtract 1 from this result and output it to my terminal.
Let's assume I have a shell script called "checkD."
An example would be:
$ checkD ~/direct/here
5
// I want it to output 4, but how would I do that?
I would assume I store that result into a variable, such as
num="find $1 -type d | wc -l" , or something along the lines.

What would be the correct way to modify that '5' into a 4 in my bash shell script?

evo2 · 05-22-2012, 10:53 PM

Hi,

assuming bash (or similar), I think you are looking for something like the following

Code:

# Store the output in the variable $num
num=$(find $1 -type d | wc -l)
echo $num

# Decrease $num by one
((num--))
echo $num

Cheers,

Evo2.

divyashree · 05-23-2012, 12:46 AM

YOu can use like this also:

Quote:

num=$(expr `find $1 -type d | wc -l` - 1)
echo $num

newbiecolonopenparens · 05-23-2012, 03:51 PM

Thank you for the prompt responses, divayashree and evo2.
evo2's code worked just fine. It was exactly what I needed.
However, divayashree's did not work. It looks a bit cleaner though. I got the error "expr: non-numeric argument"
Is there a way to make it work with expr?

colucix · 05-23-2012, 04:12 PM

Just out of curiosity, what is this extra directory you mentioned? Maybe there is a way to exclude it from the result using the proper predicates of the find command. For example, if you want to find the sub-directories contained in the main directory and exclude the main directory itself from the output:

Code:

find $1 -mindepth 1 -type d

newbiecolonopenparens · 05-23-2012, 04:54 PM

The extra directory is the current directory. It includes itself in the count when I do:

Code:

find $i -type d | wc -l

colucix, that line of code definitely prints out all sub-directories, but I'm trying to get the number of folders, where the argument comes from command line arguments.

Another question: I want to iterate through each of the command line arguments. I understand that

Code:

argument=$1

will store my first argument in there. ((argument++)) won't work. I've tried googling for the answer, but nothing much has helped. An example would be:

Code:

$ checkD ~/Desktop ~/Music

How do I iterate 'argument' from $1 to $2, and from $2 to $3, and so on?
I'm expecting something like

Code:

argument=$1
count=1
end=$#
while( count != end )
do
    echo $argument
    ((count++))
    shift 1
done

a C++ equivalent would be

Code:

int main( int argc , char* argv[] ){
    int i = 1;
    while( i < argc )
        cout << argv[i] << ' ';
    i++;
}

colucix · 05-23-2012, 05:17 PM

Quote:

Originally Posted by newbiecolonopenparens

colucix, that line of code definitely prints out all sub-directories, but I'm trying to get the number of folders, where the argument comes from command line arguments.

So pipe the output to wc -l as you did in your example. Or am I missing something?

Code:

n=$(find $1 -mindepth 1 -type d | wc -l)

My remark was that a workaround is not necessary (subtracting 1 from the result) if the source of the problem can be fixed.

Quote:

Originally Posted by newbiecolonopenparens

Another question: I want to iterate through each of the command line arguments.

Using shift as you already tried. However you must not increase the argument number, since arguments are moved to the left and the current one is always $1. Example:

Code:

until [[ -z "$1" ]]
do
  echo "$1"
  shift
done

newbiecolonopenparens · 05-23-2012, 05:27 PM

haha I didn't realize that I could've piped

Code:

find $1 -mindepth 1 -type d

also, I iterated through the arguments differently.

Code:

for argument in $*
do
   ...
done

hooray for being new to scripting. Thanks for the help everyone!

TobiSGD · 05-23-2012, 05:36 PM

Bash Beginners Guide, Section 3.2.5:

Quote:

$* vs. $@

The implementation of "$*" has always been a problem and realistically should have been replaced with the behavior of "$@". In almost every case where coders use "$*", they mean "$@". "$*" Can cause bugs and even security holes in your software.

chrism01 · 05-23-2012, 06:26 PM

Nice quote, and so true

newbiecolonopenparens · 05-23-2012, 09:55 PM

nice book...

divyashree · 05-23-2012, 10:38 PM

Quote:

Originally Posted by newbiecolonopenparens

Thank you for the prompt responses, divayashree and evo2.
evo2's code worked just fine. It was exactly what I needed.
However, divayashree's did not work. It looks a bit cleaner though. I got the error "expr: non-numeric argument"
Is there a way to make it work with expr?

Have you used the backtick as I mentioned , without which it will give error ...
marked in red bold,

Quote:

num=$(expr `find $1 -type d | wc -l` - 1)

catkin · 05-23-2012, 11:41 PM

Or, less error prone and more consistent: num=$(expr $(find $1 -type d | wc -l) - 1)

evo2 · 05-23-2012, 11:53 PM

While we're at it:

Code:

num=$(( $(find $1 -type d | wc -l) - 1))

Evo2.

David the H. · 05-24-2012, 06:08 PM

Note that using wc to count the number of files is subject to some error. It's possible for file/directory names to contain newline characters themselves, and they would be counted as multiple files using this technique.

A more robust approach would be to output the files into an array using the -print0 option, and using that to get the count and to iterate over them.

Some appropriate links:
http://mywiki.wooledge.org/BashFAQ/004
http://mywiki.wooledge.org/BashFAQ/005
http://mywiki.wooledge.org/BashFAQ/020

bash v4+ has a new globstar option that, in simple cases, can replace find. (although I've found it to still be buggy when operating on certain large directory trees.)

http://wiki.bash-hackers.org/syntax/expansion/globs

Code:

#!/bin/bash

shopt -s globstar dotglob	#set dotglob to count hidden directories as well

dirs=( "$1"/**/ )

#or using find
#while IFS= read -r -d '' ; do
#	dirs+=( "$REPLY" )
#done < <( find "$1" -type d -print0 )


echo "The number of subdirectories in $1 is: $(( ${#dirs[@]} -1 ))"

for dir in "${dirs[@]}"; do

	[[ $dir -ef "$1" ]] && continue		#skip processing if it's the starting directory

	echo "This is a command being applied to $dir"

done

exit 0