first of all, some background...I am really new to writing BASH scrips...but I find them wonderful...I am trying to write some BASH scrip, but got some problem and no good reference on the web

I have a folder of mp3s, sorter in this way:
/home/user/mp3/artist/albumname/01.mp3
/home/user/mp3/artist/albumname/02.mp3
/home/user/mp3/artist/albumname/03.mp3
...

okay?
but now I want to go through the whole /home/usr/mp3 directory and rename each file like artist_albumname_track#.mp3...

I am wondering couple of things...
1) say I do an ls, it lists the files out...how do I get that into a for loop for that I get to rename them?

3) how does ls distinguish a file from a directory????

3) say I want a variable dirname to have the result of running an ls command (or command in general), how do I do that??
ls > cur writes the result to a FILE, but I want a variable (list) to have the value...

4) for dirname in *
do
cd dirname
done

doesn't work since * would gives directories but as well as files...

thanks

5) say I am in the loop
for name in *;

how do I check if name is a file or directory???

i wouldn't use ls for this, i'd use a find statement.

find /home/you/mp3 -iname *.mp3

and then you'll instalty get a very simple list of all mp3's. ls isn't the best for this really.

to test if something is a directory you'd use a test statement like

if [ $filename -d ]
then...

i think that's the right syntax, check the test manpage for confirmation.

watch out for spaces as well if you do soemthing like

for i in `ls /home`

then each time you have a newline OR space it will iterate with that as teh main variable. i saw something on here a while back for how to use entire spaces, have a search around.

persaonlyl i'd say use perl to do this, it's very very similar and no more involved, it'd just make it a lot simpler, as you can pull out the parts of the directory structure so much easier.

Ok, I'll try to write down an example to show you Perl ain't needed for this :-]

We assert all mp3's are under the root /home/user/mp3, and all names of mp3's are to be taken from the subdir's artist/albumname 01.mp3 to produce your "artist_albumname_track#.mp3".

What we need is a command that will find us all files that end on ".mp3". Written out it will look like:
root=/home/user/mp3; find ${root} -type f -name "*.mp3"
Execute this and you'll see the list. The "type -f" isn't really needed int his case, but will come in handy when there are other filetypes dispersed over the search area you don't want to pick up (sockets, for instance).

Now we want to use these names to rename the file. What we do is assign each result of the find query to a variable:
root=/home/user/mp3; find ${root} -type f -name "*.mp3" | while read raw; do
echo ${raw}; done
Execute this and you'll see the list. Not from using "find", but by echoing the contents of the variable.

Each variable looks like "/home/user/mp3/artist/albumname/01.mp3", so we can't use it to rename it to something else. What we need is to split the variable into chewable bits. I like to use arrays' for that:
root=/home/user/mp3; find ${root} -type f -name "*.mp3" | tr "/" " " | while read raw; do m=( ${raw} ); echo ${m[@]}; done
Execute this and you'll see the list. Not from using "find", but by echoing the contents of the array.

Arrays count from zero, and we need parts 3,4 and 5 to rename them mp3's:
root=/home/user/mp3; find ${root} -type f -name "*.mp3" | tr "/" " " | while read raw;
do m=( ${raw} )
echo mv ${raw} ${root}/${raw[3]}_${raw[4]}_${raw[5]}; done

That should do it. If you're convinced this works for you, just remove the last "echo" to commence renaming.

More? Look up the ABS or Bash Advanced Scripting Guide at tldp.org.

HTH.

dammit i really wish people would stop running rings around me... pah!

LOL, no, I just checked before I posted and saw you had ure answer ready first, so I added that 1st line :-]
Blame it on the brewskies

#!/bin/bash

# rename and copy mp3 files

sep=_

# make the destination directory
mkdir -p mymp3

for i in $(find . -type f -name '*.mp3' -print); do
if [ -r $i ]; then
artist=$(echo $i | awk -F/ '{ printf("%s",$(NF-2)) }')
album=$(echo $i | awk -F/ '{ printf("%s",$(NF-1)) }')
track=$(echo $i | awk -F/ '{ printf("%s",$NF) }')

cp $i ./mymp3/$artist$sep$album$sep$track
fi
done

exit 0

The script above will make a new directory mymp3 and copy all the files in there. I changed the 'mv' to a 'cp' because of 2 things.

1) I havent tested it, hehe.
2) I'm not sure if you want a flat directory with all mp3s in there.

You can make a li'l tweak and make every mp3 land on the directory it already is.

start off with 'man test' as acid suggested.. if you're not faint hearted, i'd suggest 'man bash'. Also there is ABS
http://www.tldp.org/LDP/abs/html/index.html

I've told myself a million times to refresh the goddamn thread before I write that up and hit submit. You guys are fast!

I would try something like:
cd /home/usr/mp3
for FILE in `find * -type f -name "*.mp3"`
do
mv $FILE `echo $FILE | sed 's/\//_/g'`
done

This should turn artist/album/track.mp3 into artist_album_track.mp3.

what happens if you do "mv $FILE" and the name contains spaces?

well you won't get there with that example, as any spaces will get screwed up in the find statement.

how do you deal with spaces in script files?

i have a directory /SERVER/example directory/

when the script tries to access the dir it just says

'/SERVER/example': No such file or directory
'directory/': No such file or directory

so what would i need to put in between the dir. name for it to work?

a backslash before the space.

/SERVER/example\ directory/