Dir diff and removal

Question

Is there a simple way to (recursively) remove all files from a directory A that are identical (same name and same hash) to files in directory B while preserving the structure?

e.g. A/file1 gets removed if B/file1 exists; but A/file1 gets not removed although B/d/file1 exists

Answer 1

You could do something like this. Please note this is untested code. I assume $A and $B are your two directories:

(cd $B && find . -type f -print) | (cd $A && while read f; do /bin/rm -f "${f}"; done;)

I just tried it, but echoing the command to stdout instead of running it, and it appears to work ok:

$ (cd $B && find . -type f -print) | (cd $A && while read f; do echo /bin/rm -f "${f}";done;)
/bin/rm -f ./1
/bin/rm -f ./2
/bin/rm -f ./3
/bin/rm -f ./4

If that appears to do what you want, run it again but omit the word "echo" in the command.

Answer 2

If you want all files in A to exist in B then you could simply use rsync with the --remove-source-files option, to add / update the files in B, with those in A, and then delete the original files from A. Or with a bit more hacking, running rsync in --dry-run mode, and piping the output through grep, xargs and rm -f, only compare the two directory structures, and delete the identical files, see:

Linux / Unix rsync: Delete Source File After Transfer

rsync --remove-source-files deletes source files one by one or after rsync completes?