UNIX for Dummies Questions & Answers

Hi all,

I am currently writing a find-and-remove kind of script that is to be used for Solaris and Linux.

Currently am using the find command below that I is in find only current directory (universal) | commandlinefu.com

This however gives me the "ksh: /bin/find: arg list too long" error on Solaris and "ksh: find: /usr/bin/find: cannot execute [Argument list too long]" error in Linux.

Using the command below works but it is doing a recursive find which is not what I wanted to have

How do I make "find /u01/tmp -name '*.tmp' -mtime +10 -exec ls {} \;" non-recursive? I can't used maxdepth as the find in Solaris does not recognize that option.

Going back to the first find that I am currently using. I noted that find only current directory (universal) | commandlinefu.com says "you must be in the current directory".

So I tried that, now I am doing

Now am not getting the list too long anymore but problem is the list is now coming up with having ./ in the beginning. Sample list below:

Code:

./arc_maint_20147_5.tmp
./arc_maint_20150_5.tmp
./arc_maint_20167_8.tmp
./arc_maint_20168_2.tmp
./arc_maint_2019_8.tmp
./arc_maint_20262_1.tmp
./arc_maint_20264_7.tmp
./arc_maint_20272_c.tmp
./arc_maint_20338_2.tmp
./arc_maint_20388_6.tmp
./arc_maint_20398_2.tmp
./arc_maint_20425_6.tmp
./arc_maint_20480_9.tmp
./arc_maint_20509_7.tmp

Now I change my find to do

Code:

find . \( ! -name . -prune \) \( -type f -name "*.tmp" -mtime +10 \) | sed 's#^./##'

It gets rids of the ./ ... but ... I want to be able to have the listing to include the present working directory. How do I do that? Ideally, I want the listing to look like the one below:

Code:

/u01/tmp/arc_maint_20147_5.tmp
/u01/tmp/arc_maint_20150_5.tmp
/u01/tmp/arc_maint_20167_8.tmp
/u01/tmp/arc_maint_20168_2.tmp
/u01/tmp/arc_maint_2019_8.tmp
/u01/tmp/arc_maint_20262_1.tmp
/u01/tmp/arc_maint_20264_7.tmp
/u01/tmp/arc_maint_20272_c.tmp
/u01/tmp/arc_maint_20338_2.tmp
/u01/tmp/arc_maint_20388_6.tmp
/u01/tmp/arc_maint_20398_2.tmp
/u01/tmp/arc_maint_20425_6.tmp
/u01/tmp/arc_maint_20480_9.tmp
/u01/tmp/arc_maint_20509_7.tmp

At the moment, I re-direct the output to a file, then use a for-loop to read each line and append the directory path. Painful and a long process but suits me for the time being.

I tried using awk below and it fails

Code:

find . \( ! -name . -prune \) \( -type f -name "*.aud" -mtime +10 \) | sed 's#^./##' | awk '{sub(/^/, "`pwd`")};1'
awk: syntax error near line 1
awk: illegal statement near line 1
awk: syntax error near line 1
awk: bailing out near line 1

Tried using sed below

Code:

find . \( ! -name . -prune \) \( -type f -name "*.aud" -mtime +10 \) | sed 's#^./##' | sed 's#^#`pwd`#'

But am getting the string `pwd` instead of the output from running the pwd command.

FYI, reason am wanting to have the full path is to be safe that am doing the rm of the right file since I can't undo the rm.

Any advice will be much appreciated. Thanks in advance.

maybe I'm missing your point, but if you don't want to recurse, then don't use 'find'. If your favorite shell's filename expansion is too weak to let you use complex regular expressions, then use 'echo *' or 'ls' and pipe to your favorite 'grep'?

Hi,

Thanks for your response.

I need to use find 'coz I need to work on files that are x-number of day's old? Can I achieve the same with just using echo, ls and pipe?

Hi,

Try this one,

Code:

find . \( ! -name . -prune \) \( -type f -name "*.aud" -mtime +10 \) | awk -v base=`pwd` '{gsub(/^./,base);}1'

Note: I am tested your find command. If it works well, awk command do the rest.

Cheers,
Ranga

Using sed :

Code:

find . \( ! -name . -prune \) \( -type f -name "*.aud" -mtime +10 \) | sed "s#^./#$PWD/#g"

Or (avoiding the usual Solaris issues with the default awk and sed).

Code:

DIR="/u01/tmp"
find . \( ! -name . -prune \) \( -type f -name "*.tmp" -mtime +10 \) |\
     while read filename1
     do
          filename2=$(echo "${filename1}"|cut -c3-)
          echo "${DIR}/${filename2}"
     done

Up find/sed works on solaris 10 using /usr/bin/sed

Regards
Peasant.