Undo copy (cp) command action

If I understand well, the following is the case:

  • You copied a (presumably large) number of files on to an existing directory, and you need / want to reverse the action.
  • The targeted directory contains a number of other files, that you need to keep there, which makes it impossible to simply remove all files from the directory

The script below looks in the original (source) directory and lists those files. Then it looks into the directory you copied the files to, and removes only the listed files, as they exist in the source directory.

The try element is added to prevent errors, for example in case you might have removed some files manually already, or if not all files from the source directory were copied to the destination. If you need sudo privileges, simply run the script with "sudo" (see below).

The script

#!/usr/bin/env python

import os

source_dir = "/path/to/source" # the folder you copied the files from
target_folder = "/path/to/destination" # the folder you copied the files to

for root, dirs, files in os.walk(source_dir):
    for name in files:
        try:
            os.remove(target_folder+"/"+name)
        except FileNotFoundError:
            pass

How to use

  • Paste the script in an empty file, save it as reverse.py,
  • Insert the correct paths for source and target folder,
  • Make it executable for convenience reasons,
  • Run it by the command:

    [sudo] /path/to/reverse.py
    

Warning

First try on a test directory if I understood well what you need to achieve!


If the sourcedirectory is "flat"

In case the source directory has no sub-directories, the script can even be simpler:

#!/usr/bin/env python

import os

source_dir = "/path/to/source" # the folder you copied the files from
target_folder = "/path/to/destination" # the folder you copied the files to

for file in os.listdir(source_dir):
    try:
        os.remove(target_folder+"/"+file)
    except FileNotFoundError:
        pass

Note

If the copy action overwrote (replaced) a similarly named file in the destination, the file will be removed, but the original file will (of course) not be brought back by the script. The assumption is that there are no name clashes.


TL;DR:

All files that are present in both src and dest can be removed from dest like this:

find . -maxdepth 1 -type f -exec cmp -s '{}' "$destdir/{}" \; -exec mv -n "$destdir/{}" "$toDelete"/ \;

For a step-by-step, explanation, see below.

Simplifying the problem:

To understand what the command we want to undo actually did, we start by simpifying it:

The command we want to undo is

sudo cp From_SOURCE/* To_DESTINATION/

For understanding how to undo, sudo is not relevant.

I'll use the directory names src for From_SOURCE and dest for To_DESTINATION.

Now, our command is:

cp src/* dest/

If src contains the files f1, f2 and f3, and the directories d1 and d2, the shell expands that command to:

cp src/f1 src/f2 src/f3 src/d1 src/d2 dest/

Without options like -r, -R or -a, the command cp does not copy directories.
That means, the command will leave them out, showing an error for each of them:

$ cp src/f1 src/f2 src/f3 src/d1 src/d2 dest/
cp: omitting directory 'src/d1'
cp: omitting directory 'src/d2'

That means, we have only copied simple files, and no directories, to dest.

Deciding which files to remove:

Possibly there were files in dest of the same name as files in src. In this case, the files were overwritten (1). It's too late for them, sorry. Get them back from the latest backup.

Regarding the files that are there, we want to remove only files that have been copied over. These files exist in both directories, with the same name, and the same content.

So we look for these files:

First, we cd into src, because it makes the following find commands much simpler, and set a variable to the absolute path of dest:

$ cd src
$ destdir="$(readlink -f dest)"

Then, we list all files in src:

$ find . -maxdepth 1 -type f

and, for each file found, use cmp to check whether there is a file with the same content in dest:

$ find . -maxdepth 1 -type f -exec cmp -s '{}' "$destdir/{}" \; -print

Removing the files, carefully:

These files are the ones we want to remove. But to be sure, we move them into a different directory first - and take a look at the commands before running them:

$ toDelete=/tmp/toDelete ; mkdir -p "$toDelete"
$ find . -maxdepth 1 -type f -exec cmp -s '{}' "$destdir/{}" \; -exec echo mv -n "$destdir/{}" "$toDelete"/ \;
mv -n /path/to/dest/./f1 /tmp/toDelete/`
mv -n /path/to/dest/./f2 /tmp/toDelete/`
mv -n /path/to/dest/./f3 /tmp/toDelete/`

Looks good! Now we can leave out the echo to run the real mv commands:

find . -maxdepth 1 -type f -exec cmp -s '{}' "$destdir/{}" \; -exec mv -n "$destdir/{}" "$toDelete"/ \;

All the files from dest that were copied from src, and still actually the same in src and dest, are now in /tmp/toDelete/, and can be removed after taking a last look.


Notes:
(1) Check whether cp is an alias to cp -i or so, that would have prevented overwriting files by asking first.


This is old, but I just wanted to post a pure bash answer:

First change to the directory where you copied the files.

cd dest

Then, ls the source directory and pipe the output into rm

ls source | xargs rm