awk help needed

Chris F.A. Johnson cfaj-uVmiyxGBW52XDw4h08c5KA at
Sun Jul 19 20:04:48 UTC 2009

On Sun, 19 Jul 2009, Jamon Camisso wrote:

> Chris F.A. Johnson wrote:
>> On Sun, 19 Jul 2009, Jamon Camisso wrote:
>>> Giles Orr wrote:
>>>> 2009/7/18 Alex Beamish <talexb-Re5JQEeQqe8AvxtiuMwx3w at>:
>>>>> On Sat, Jul 18, 2009 at 11:46 AM, Giles Orr<gilesorr-Re5JQEeQqe8AvxtiuMwx3w at> wrote:
>>>>>> By way of introduction: I'm finally, finally trying to get a new
>>>>>> edition of the Bashprompt HOWTO out there.  This will probably result
>>>>>> in me posting a lot of detailed and mildly weird questions to this
>>>>>> list.  This is the first.
>>>>>> The intention of this script is to figure out how much space the files
>>>>>> in the current directory take up.  There are about a million ways to
>>>>>> do this - and yes, I know that "ls -l" spits out a "total" line: I
>>>>>> don't know what it's totaling, but my math has never agreed with it
>>>>>> ... feel free to explain though.  I decided that I'd like to do this
>>>>>> as much in awk as possible since it does decimal math (unlike bash)
>>>>>> and it's certainly the easiest way to do the text parsing.  I've tried
>>>>>> bc as well, but you have to use other utility programs to parse and
>>>>>> split the input for it.
>>>>> Sorry, but I usually use the du command for this.
>>>>>  root at music:/etc/init.d# du -h .
>>>>>  504K    .
>>>>> Sometimes I want to know how heavy an entire tree is, so I use
>>>>>  $ du -sh foo
>>>>> The 'h' argument does intelligent size management, so shows K, M and G.
>>>> I would love to use "du" because initially it seems like precisely the
>>>> right tool ... but I want only the sum of the sizes of the files in
>>>> the current directory, and "du" is by default recursive (which also
>>>> makes it painfully slow to return, not a good thing for something
>>>> incorporated into a Bash prompt).  If it's possible to stop "du" from
>>>> recursing, I'll use it immediately - but that looks difficult to
>>>> impossible.  Any thoughts?
>>>> Thanks to everyone else who answered too: it all helped.  It certainly
>>>> sounds like piping into "awk" is the way to go rather than trying to
>>>> write a self-contained "awk" script.
>>> Try this:
>>   DON'T.
>>> #!/bin/bash
>>> i=0 j=0 k=0
>>> echo -n "Enter directory: "
>>> read dir
>>> for i in `ls $dir`
>>     Not only is ls unnecessary, but it will cause the script to fail
>>     if any filenames contain spaces or other pathological characters.
>>     Use:
>> for i in "$dir"/*
>>>  do
>>>    if [ -f $i ]; then
>>>      j=`du -s $i |awk '{print $1}'`

    Always quote variable references unless there is a good reason not
    to: "$i". Without them, it fails if the value contains spaces.

>>>      k=`expr $k + $j`

     Why are you using an external command for integer arithmetic?

>>>    fi
>>>  done
>>> echo $k

du -s "$dir"/* | awk '{ total += $1 } END { print total }'

> Interesting, I started with for i in * but that barfs on stuff too.

    There are other problems, too. I just reacted to the horrible
    "for i in `ls dir`". There is no point to using ls without any
    options; shell expansion can do it much more efficiently.

> Suppose since I don't have an files I can see with spaces in their
> names (pathological hatred for such names), it appears to work
> correctly for me :)

    I feel the same way of filenames with spaces, but they are hard to
    avoid. I never create any, but I expect that every distro contains
    many, and downloads often do. Unless you get into the habit of
    handling them, your scripts will fail sooner or later.

    Chris F.A. Johnson                      <>
    Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)
The Toronto Linux Users Group.      Meetings:
TLUG requests: Linux topics, No HTML, wrap text below 80 columns

More information about the Legacy mailing list