Dear friends,

I have created a script to rename all files in a directory by appending the file name with username (who created the file), the date it was created. For example, "apple.doc" should be renamed to "johnFeb23apple.doc" where "john" is the owner and "Feb23" is file created date. It works fine with file names without spaces. but there are few issues I don't know how to fix.

First, Here's my script:

#!/bin/bash
for i in *
do
newname=$(ls -l ${i} | awk '{ print $3 $6 $7 $9 }')

    mv $i $newname
done

ls: cannot access fff: No such file or directory
ls: cannot access eee.xls: No such file or directory
mv: cannot stat `fff': No such file or directory

You didn't quote the ${i} variable, so it split. Quote it "${i}"

Also, awk can't tell the difference between the spaces splitting the columns, and the spaces splitting the filename, causing problems.

Also, there's also no point running awk 9,000 times to process 9,000 lines. That's like making 9,000 phonecalls to say 9,000 words. awk can do it all in one go if it can do it at all.

But, I think shell read is better suited, since you can tell it exactly how many columns you want, and it won't split beyond that. Try and put 12 values into 9 columns, the last few columns will all get piled into the last variable unmodified. I really wish you could do that in awk, sometimes.

#!/bin/sh

cd /path/where/I/want/it/to/go

# The filename will get all put into the last variable, I
ls -l | while read A B C D E F G H I
do
        # Remove the 'echo' once you've tested it and found it does what you want.
        echo mv "$I" "$C$F$G$I"
done

If you do not quote the white space, "$name" every reference, you need to use line feed as your delimiter, or if you have embedded linefeeds, more extreme tricks, like keying off the 'ls -l' non-name printout to pile up the right number of lines, tricky at EOF where there is no following line with an 'ls -l' prefix. C, PERL and such can handle each file name as one string without looking inside.

$ mkdir emb-ws
$ cd emb-ws
$ echo 'a space' >'a space'
$ echo 'two  space' >'two  space'
$ echo 'a   tab' >'a   tab'
$ echo 'a^Mcarriage-return' >'a^Mcarriage-return'
$  echo 'a
linefeed' >'a
linefeed'
$ ls | while read f; do   echo ">$f<"; done | cat -vte
> )
>a^Itab<$
>a<$
>linefeed<$
>a^Mcarriage-return<$
>a space<$
>two  space<$
$ ls -l |cat -vte
total 20$
-rw-rw-r-- 1 dpickett dpickett  6 Feb 24 15:09 a^Itab$
-rw-rw-r-- 1 dpickett dpickett 11 Feb 24 15:09 a$
linefeed$
-rw-rw-r-- 1 dpickett dpickett 18 Feb 24 15:09 a^Mcarriage-return$
-rw-rw-r-- 1 dpickett dpickett  8 Feb 24 15:09 a space$
-rw-rw-r-- 1 dpickett dpickett 11 Feb 24 15:09 two  space$
$

Incorrect, * doesn't split on spaces by itself. * would be pretty useless if it did shell expansion inside itself. The right filenames end up in the $i variable.

But he doesn't quote the $i variable, causing it to split later.

Yes, I misspoke, and edited it back a bit. Life can get crazy if you have recursive evaluations, like with

for i in $(ssh -n hostx 'ls *')

ssh -n hostx 'ls *' | while read i