[SOLVED] Bash if filesize is greater than 2GB

cober · 08-19-2011, 02:39 AM

Hello,

I have some problems with a bash script...

Here it is:

Code:

for FILENAME in /home/cober/downloads/*/*

    do
    FILESIZE=$(stat -c%s "$FILENAME")

        if [[ $FILESIZE > 2050000000 ]] ;then
        echo "$FILENAME is too large = $FILESIZE bytes."
        fi

    done

exit 0

I just want the output of the files which are greater than 2GB, but I get this right now..

Code:

/home/cober/downloads/x1/1.wmv is too large = 3555295269 bytes.
/home/cober/downloads/x2/2.wmv is too large = 2901575187 bytes.
/home/cober/downloads/x3/3.wmv is too large = 2580265584 bytes.
/home/cober/downloads/x4/4.mp4 is too large = 651414946 bytes.
/home/cober/downloads/x5/5.mp4 is too large = 783791599 bytes.
/home/cober/downloads/x6/6.mp4 is too large = 838259812 bytes.

As you see some info is incorrect... Don't know why am I getting the notice that 4.mp4, 5.mp4 and 6.mp4 are larger than 2050000000 bytes.

Anyway, how to deal with this?

Plus, how could I make that those files which are bigger than 2GB take an appropriate action (let say, to make a rar) - rar a -v2048000000b -m0 -r $file

Thanks for your help!

colucix · 08-19-2011, 03:01 AM

In bash the integer comparison is performed by the -eq, -ne, -gt, -lt, ... operators. If you use the =, >, < notation it will perform a string comparison. Change the operator > to -gt in the IF statement and the trick is done.

Regarding the action, you can just put the command inside the IF statement using the loop variable FILENAME in the command line (in place of $file in your rar statement).

cober · 08-19-2011, 03:56 AM

colucix, thank you very much! -gt did the trick.

about an action, i just found out that I have to go (cd) in every single dir and then perform this action with rar, otherwise it won't pack
as it should. rar has "w<path> Assign work directory" option that could fix my problem, but don't know how to integrate into if statement.

So, as we said, these 3 are above 2GB

Code:

/home/cober/downloads/x1/1.wmv is too large = 3555295269 bytes.
/home/cober/downloads/x2/2.wmv is too large = 2901575187 bytes.
/home/cober/downloads/x3/3.wmv is too large = 2580265584 bytes.

Now I have to "enter" workdir (x1, x2 and x3) and then I'll be able to perform rar action. Any clues on that?

Much appreciated!

colucix · 08-19-2011, 05:37 AM

Yes. Once you have the full path inside the loop (as in your example) stored in the variable FILENAME, you can either use parameter substitution to strip the filename from the end of the string:

Code:

cd ${FILENAME%/*}

or use the dirname command:

Code:

cd $(dirname $FILENAME)

Hope this helps.

ta0kira · 08-19-2011, 08:26 AM

You can also do this with find:

Code:

find /home/cober/downloads/*/ -maxdepth 1 -size +2G | while read file; do
  echo "$file is too large = $(stat -c%s "$file") bytes."
done

Kevin Barry

cober · 08-19-2011, 01:34 PM

Thanks! Now just works perfect!

David the H. · 08-19-2011, 10:14 PM

You should use ((..)) for numeric tests anyway. The <> characters are treated as integer comparators when used in an arithmetic context.

Code:

if (( FILESIZE > 2050000000 )) ;then
     echo "$FILENAME is too large = $FILESIZE bytes."
fi

http://mywiki.wooledge.org/ArithmeticExpression
http://www.tldp.org/LDP/abs/html/comparison-ops.html

cober · 08-20-2011, 12:35 AM

David, thanks for that explanation.

Just last question regarding bash.. let say that I want upload some files with various extensions through ftp via command line.

It goes just like this:

Code:

ncftpput -u $username -p $password $host / /*.mp4

Ok, how could I make this command correct to upload some files with different extensions like .mp4, .wmv, .avi etc... in one command.

Thanks guys for helping me out.

EDIT: i have also this tiny script to get direct urls of the movies on my server

Code:

for i in `ls -1 */* | grep jpg`; do echo http://domain.com/$i; done

but instead of */*, i'd like to specify this movie extensions to avoid any non-movie file around there. (*/*.mp4, */*.avi ... - in one command)

David the H. · 08-21-2011, 05:31 AM

Quite detailed filtering can be done with globbing, particularly with bash's extended glob patterns.

http://mywiki.wooledge.org/glob
http://www.linuxjournal.com/content/...ended-globbing

Note also though that globbing patterns are expanded into lists of files before the command itself is executed, so you should be able to simply put two or more separate globs in the same command.

Code:

ncftpput -u $username -p $password $host / /*.mp4 /*.wmv /*.avi

The actual command when executed will look something like this:

Code:

ncftpput -u username -p password host / /file1.mp4 /file2.mp4 /file1.wmv /file2.wmv /file1.avi /file2.avi

Assuming those files exist, of course. Stick an echo in front of the command to view what bash would actually run. See the nullglob shell option for what to do if a glob fails to match anything.

For more advanced matches, you'll generally want to use find.

http://mywiki.wooledge.org/UsingFind

Always use the -print0 option whenever possible, to avoid problems with spaces and special characters in names. But the receiving program has to have the ability to handle the null-delimited file list, like xargs.

There are lots of previous discussions on filename matching here and on the web. Try searching around a little.

As for this one:

Code:

for i in `ls -1 */* | grep jpg`; do echo http://domain.com/$i; done

1) $(..) is highly recommended over `..`

2) Parsing ls is both unsafe and generally unnecessary. The for loop will accept glob inputs directly, and without word splitting.

3) If globbing isn't up to the job, then use find, either alone with -exec, or run the output via process substitution into a while+read loop, instead of a for.