Backup adding numbers when using mv command

kilee · 05-24-2016, 11:03 AM

Hello
I'm moving all zip files I have into a single folder with this:

find / -iname "*.zip" -type f -exec mv {} /Volumes/500/zip/ \;

Works perfect, but I'd like to make backups in case of name conflicts. -b looks like the answer, but the backups add a ~ symbol to the extension.

file.zip
file.zip~ and so on.

Is it possible to number the files before the extension?

file.zip
file1.zip
file2.zip

Thanks!

hazel · 05-24-2016, 11:28 AM

Yes, it is possible. You would need to write a script containing a for loop. I'm not quite sure of the syntax but it would be something like

Code:

number=1
for file in `find / -iname "*.zip" -type f`
do
mv $file `echo ${file%%".zip"}$number.zip`
number=$number+1
done

The echo command should strip the suffix off, insert a number, and put the suffix back on again.

dab1414 · 05-24-2016, 11:29 AM

http://linux.die.net/man/1/mv

add "--backup=t" for numbered backups

EDIT: sorry this still puts number behind extension. "mv --suffix=.bak" will get rid of the ~

grail · 05-24-2016, 11:39 AM

You should not use a for loop if there is any possibility of word splitting occurring. Simply change to a while loop and you should be fine.

You also need to include some sort of test for the repeated file names as simply adding an ever increasing number will also give a numbered copy to something that only exists once.

HMW · 05-24-2016, 12:44 PM

^grail is right (as usual). A while loop is better:

Code:

#!/bin/bash

i=1
while read filename; do
    echo ${filename%%".zip"}$i.zip
    ((i++))
done < <(find . -iname "*.zip")

exit 0

Also:

Quote:

Originally Posted by grail

You also need to include some sort of test for the repeated file names as simply adding an ever increasing number will also give a numbered copy to something that only exists once.

syg00 · 05-24-2016, 06:59 PM

Me, I'd just run over the final target dir and change the trailing tilde to say "_$(date +%F)".
KISS.

jpollard · 05-26-2016, 02:21 AM

Quote:

Originally Posted by HMW

^grail is right (as usual). A while loop is better:

Code:

#!/bin/bash

i=1
while read filename; do
    echo ${filename%%".zip"}$i.zip
    ((i++))
done < <(find . -iname "*.zip")

exit 0

I think it is cleaner as:

Code:

i=0
find . -iname "file*.zip" | while read filename; do
   ((i++))
done
if ((i)) ; then
    target=${file%%".zip"}$number.zip
else
    target=file.zip
fi

But then that is just me.

szboardstretcher · 05-26-2016, 03:31 AM

It should be mentioned that using 'mv' is ballsy because many many things can go wrong and cause you headaches. You shouldn't be using 'mv' unless you already have a backup, or don't care much about the data. Making a backup, copying files from one place to another, verifying, double-checking, and only then deleting, is a much more reliable route.

grail · 05-26-2016, 03:33 AM

@jpollard - I am not quite understanding your code? You are counting all the files that are duplicates and then seeing if positive you set a variable ... at what point are you going to do the copy/move?

JJJCR · 05-26-2016, 04:26 AM

Quote:

Originally Posted by szboardstretcher

It should be mentioned that using 'mv' is ballsy because many many things can go wrong and cause you headaches. You shouldn't be using 'mv' unless you already have a backup, or don't care much about the data. Making a backup, copying files from one place to another, verifying, double-checking, and only then deleting, is a much more reliable route.

Agree with this one, do an rsync to the destination folder once done then delete the source. Safer way. KISS method.

jpollard · 05-26-2016, 05:43 AM

Quote:

Originally Posted by grail

@jpollard - I am not quite understanding your code? You are counting all the files that are duplicates and then seeing if positive you set a variable ... at what point are you going to do the copy/move?

Afterwards. The snippit is more just to identify the possible name.

grail · 05-26-2016, 09:22 AM

ahhh .. cool .. I was just a bit lost

jpollard · 05-26-2016, 01:22 PM

Ran across another way... using shell array variables.

Instead of counting the list, if the list of names is put into an array you can get the size of the array...

keefaz · 05-26-2016, 04:15 PM

Yes or even better using a bash 4's hash

Code:

declare -A found
destdir=/Volumes/500/zip

find . -iname "file*.zip" | while read filename; do
  name=$(basename $filename)
  target=${name%%".zip"}${found["$name"]}.zip
  ((found["$name"]++))
  echo "mv $filename $destdir/$target"
done

jpollard · 05-26-2016, 05:03 PM

Quote:

Originally Posted by keefaz

Yes or even better using a bash 4's hash

Code:

declare -A found
destdir=/Volumes/500/zip

find . -iname "file*.zip" | while read filename; do
  name=$(basename $filename)
  target=${name%%".zip"}${found["$name"]}.zip
  ((found["$name"]++))
  echo "mv $filename $destdir/$target"
done

Well... that would work - depending on the total number of files. Using perl would be unbounded (since the hash can be a disk file instead of memory resident). Not sure about python, but I believe it can be unbounded as well.

But bash would have a memory limit...