Sum all file sizes

Question

I'm new to bash and I need to do a little script to sum all file sizes, excluding subdirectories. My first idea was to keep the columns when you do ls -l. I cannot use grep, du or other advanced commands I've seen around here.

$9 corresponds to the 9th column where the name is shown.

$5 is the size of the file.

ls -l | awk '{if(-f $9) { total +=$5 } }; END { print total }

[Why *not* parse `ls`?](http://unix.stackexchange.com/questions/128985/why-not-parse-ls) — Cyrus, Feb 24 '18 at 02:37
[Parsing the output from ``ls`` is a very bad idea.](https://mywiki.wooledge.org/ParsingLs) Is this homework? Can you use ``stat``? — G-Man Says 'Reinstate Monica', Feb 24 '18 at 03:14
why can't you use `du`? is this homework? (homework is OK here, but it's best to mention that it is because most people here prefer to give hints and point you in the right direction rather than give a complete answer if it's for homework). — cas, Feb 24 '18 at 04:26
Column 9 is _not_ the filename if the filename contains whitespace, as is allowed and fairly common. In addition to the badness of using `ls`, tests like `-f name` work in the standard-and-traditional test-aka-[ utility, the similar ksh-bash-etc-only [[ builtin, and perl, but not awk. — dave_thompson_085, Feb 24 '18 at 07:26

score 6 · Accepted Answer · edited Apr 20 '19 at 20:06

6

With GNU find and awk:

find . -maxdepth 1 -type f -printf "%s\n" | awk '{sum+=$1} END{print sum+0}'

Output is file size in bytes. The final statement is print sum+0 rather than just print sum to handle the case where there are no files (i.e., to correctly print 0 in that case). This is an alternative to doing BEGIN {sum=0}.

edited Apr 20 '19 at 20:06

G-Man Says 'Reinstate Monica'

22,130
27
68
117

answered Feb 24 '18 at 02:39

Cyrus

12,059
3
29
53

score 1 · Answer 2 · answered Feb 24 '18 at 07:55

1

Using wc:

wc -c * 2> /dev/null

If all that's needed is the total, do:

wc -c * 2> /dev/null | tail -1

answered Feb 24 '18 at 07:55

agc

7,045
3
23
53

score 1 · Answer 3 · answered Feb 25 '18 at 02:29

If you're looking for a bash-centric, shell-script way to do it, here's a shell loop that gathers all of the files (dot-files included), then uses the GNU coreutils stat utility to print their size into a summation variable.

shopt -s dotglob
sum=0
for f in *
do 
  [[ -f "$f" && ! -h "$f" ]] || continue
  sum=$(( sum + $(stat -c "%s" "$f") ))
done
echo $sum

Bash considers symlinks to be "regular files", so we must skip them with the -h test.

score 0 · Answer 4 · answered Feb 24 '18 at 09:07

0

We can use below command too

find . -maxdepth 1 -type f -exec ls -ltr {} \;| awk 'BEGIN{sum=0} {sum=sum + $5} END {print sum}'

answered Feb 24 '18 at 09:07

Praveen Kumar BS

5,139
2
9
14

[Why *not* parse `ls`?](http://unix.stackexchange.com/questions/128985/why-not-parse-ls) – Cyrus Feb 24 '18 at 11:46
What benefit is there to doing a `-tr` in the `ls` when all you're doing is counting bytes one file at a time? It is indeed dangerous to parse ls, although you should be safe(r) counting only the file size. – Jeff Schaller Feb 25 '18 at 02:02

score 0 · Answer 5 · answered Mar 15 '18 at 12:10

As it was a very basic exercise, teacher demanded the exercises using basic commands that requiere a little bit more development and can be replaced by more powerful commands like find or stat later on. But I got the answer and it was this:

dir=$1
if [ ! -d $dir ]
then 
   exit 1
else

sum=0
cd $dir

(ls -l $dir) > fitxers.txt

C=($(awk '{print $5}' fitxers.txt))

len=${#C[*]}

i=0

while [ $i -lt $len ]
do 
    for element in $(ls $dir)
    do
        if [ -f $element ]
            then
            let "sum = $sum + ${C[$i]}"
            fi 
        (( i++ ))
    done
done
echo $sum
rm -r fitxers.txt
exit 0
fi

Hope it's a little bit helpfull for other beginners.

Sum all file sizes

5 Answers5