shell scripting full directory listing

sqn · 04-13-2005, 06:39 AM

hi all,

I have to write an shell script that starts from a directory that was given as a parameter, and makes a list of all names (files and folders) that appear, modifying their names by adding a number that sad how many times the file/folder appears
ex:

file2 - meas that this file appears 2 times
folder4 - means this folder appears 4 times

And i have no ideea on how to start it

thx in advanced

tangle · 04-13-2005, 08:57 AM

I am not a bash scripting expert or for that matter a novice. But the ls -R / will list all directories and their sub diresctories.

You would need to take the out put of that and match the findings. There are a few tutorials on www.tldp.org. You might want to read them. There is a beginning and an advanced guide.

ahwkong · 04-13-2005, 09:35 AM

Basically what you have to do is to find a way to traverse through all subdirectories and collect the dir and file name along the way. If a name already seen before, a counter associated with that name should be incremented.

Well, here is how it can be done:

Code:

#!/bin/bash

declare -a fns
declare -a fns_cnt
declare -a dirs
declare -a dirs_cnt

if [ "$1" == "" ]
then
    target="."
else
    target=$1
fi

for f in $(find $target)
do
    if [ -f $f ]
    then
        # echo "File $f"
        b=`basename $f`
        i=0
        find=0
        # check if file exists
        while [[ $i -lt `expr ${#fns[@]} - 1` ]]
        do
            if [ "x${fns[$i]}" == "x" ]
            then
                let "i += 1"
            else
                find=1
                break
            fi
        done
        if [ "$find" == "1"  ]
        then
            fns_cnt[$i]=$(( ${fns_cnt[$i]} + 1 ))
        else
            dim=${#fns[@]}
            fns[$dim]=$b
            fns_cnt[$dim]=1
        fi
     fi
done

echo ${fns[*]}
echo ${fns_cnt[*]}

OK,
1) it only checks file (not dir. But you can expand it very easily if you understand it)
2) the print out is not exactly in your stated format. But it is a very easy to adopt
3) command 'find' is expected to be available

ahh · 04-13-2005, 11:33 AM

Another method would be:-

Code:

ls -lC1R | sed -e '/.*:\|^$/d' | sort | uniq -c

This will provide the information you require in this format:-

Code:

          
      2 dir_a
      4 dir_b
      2 dir_c
      1 dir_d
      1 dir_f
      2 file_a
      3 file_b
      1 file_c

where the first field is the number of occurences and the second field is the file/dir name.

It will not differentiate between files and directories, so if you want to do this check out "man ls" for file listing options. This is probably a good idea, otherwise if you have a file called "dog" and a directory called "dog" it will count them both in the same line.

When you format the output I would suggest putting something between the file name and the number of occurences, otherwise if you have a listing with "pic123", for example, how would you know if it is 3 x "pic12" or 23 x "pic1" or even 123 x "pic"?

ahwkong · 04-13-2005, 04:33 PM

I love oneliner! :-)

sqn, ahh's is the right way to do it.