How to get a file size from the shell

Numeromancer

There are many ways to get the file size from the shell in Unix. I like LEN=cat file | wc -c. Others do it this way:

IINFO=`ls -l filename`
I=0
for FI in ${IINFO}
do
    case "$I" in
    4)
        LEN=$FI
        ;;
    esac
I=$((I+1))

done

corey

TRWTF is `cat file | wc -c`. At least the above method doesn't have to pipe the entire contents of a file just to get the size. Your hard drive would cry out in pain every time you wanted to just get the size of a few video files.

@corey said:

TRWTF is `cat file | wc -c

Yeah. What's wrong with <font face="courier new,courier">ls -l | cut -d" " -f5</font>?

snoofle

@corey said:

TRWTF is `cat file | wc -c`.

I have to agree there... stat dwh | grep "Size:" | awk '{ printf("%s\n",$2); }'

skippy

@snoofle said:

stat dwh | grep "Size:" | awk '{ printf("%s\n",$2); }'

Don't want to be pedantic, and most likely this is COMPLETELY dependant on what system you are running, but:

stat dwh -c %s

Saving two extra external command invocations is always nice.

I'm sure bash has a hidden built-in somewhere too.

viraptor

@snoofle said:

@corey said:
TRWTF is `cat file | wc -c`.

I have to agree there... stat dwh | grep "Size:" | awk '{ printf("%s\n",$2); }'

But do they return the same value? `wc` returns number of bytes, `stat` returs file size - with sparse files, those numbers may differ. There's of course also `du` and `ls` - just to keep people entertained ;)

Renan

All these are still better than having a user take a screenshot from the file properties in windows and then using some OCR program to parse the file size out.

morbiuswilters

@corey said:

TRWTF is `cat file | wc -c`. At least the above method doesn't have to pipe the entire contents of a file just to get the size.

The compiler will just optimize that out.

snoofle

@Zecc said:

@corey said:
TRWTF is `cat file | wc -c
Yeah. What's wrong with <FONT face="courier new,courier">ls -l | cut -d" " -f5</FONT>?

I tried all of the following:

cat file | wc -c
ls -l file | cut -d" " -f6 (yes, it's the sixth field)
stat file  | grep "Size:" | awk '{ printf("%s\n",$2);}'

and they all give the same results on tiny and large files that I tried (Linux). 'ls | cut' is the least expensive...

shepd

Let me join in!

SIZE=$(du -b filename)

SIZE=${SIZE/[[:space:]]*}

danixdefcon5

du -k file (most Unices)

du file (Linux uses 1024-blocks by default)

Of course, these return 1024-block counts, so it is kind of an approximate.

omega0

It's too inefficient to let the shell copy STDOUT into a variable:

[code]perl -e "exit -s shift" FILE[/code]

No need for clunky backtick commands, that sticks it right into your $? variable (which has the nice memonic "What's the file's size? " unlike LEN which is just a nonsense word).

You just need to watch out for those disk-wasting persons. 256-bytes should be good enough for anything.