How to recursively delete all files with extensions *.a *.b *.c *.d etc

Grabby · 10-24-2017, 02:47 PM

Hi
I need some help with the command line.

I need to recursively delete all files with given extensions, starting from the current directory and including all its sub-directories.

Say I want to delete all *.a *.b *.c *.d etc files recursively, what command should I type into the terminal emulator?

Thanks

Grabby

suicidaleggroll · 10-24-2017, 02:50 PM

Code:

find . -iname '*.a' -or -iname '*.b' -or -iname '*.c' -exec rm {} \;

Of course you should run it without the "-exec rm {} \;" at first to make sure the files it finds are what you actually want to delete.

Grabby · 10-24-2017, 03:14 PM

Quote:

Originally Posted by suicidaleggroll

Code:

find . -iname '*.a' -or -iname '*.b' -or -iname '*.c' -exec rm {} \;

Of course you should run it without the "-exec rm {} \;" at first to make sure the files it finds are what you actually want to delete.

Does this work recursively for all the nested sub-directories?

MadeInGermany · 10-24-2017, 03:52 PM

The default -and has higher priority than -or ==> there should be brackets.
The brackets have special meaning in the shell ==> need to be escaped.
Further, -exec rm {} + is faster than -exec rm {} \; and -delete is even faster.

Code:

find . -type f \( -name '*.a' -o -name '*.b' -o -name '*.c' \) -delete

rknichols · 10-24-2017, 05:17 PM

It's a lot simpler with patterns:

Code:

find . -type f -name '*.[abcd]' -delete
# or
find . -type f -name '*.[a-d]' -delete

Use "-iname" in place of "-name" if you want a case-insensitive match.

!!! · 10-24-2017, 06:06 PM

A safer idea I use is: put (> or >>) the names in a file.
Review the file, to ensure nothing 'improper' will be removed.
Then, use one of several techniques to rm the list of files.
Safety first

I'd hate to see an errant `find` destroy things!!!

p.s. notice the "Similar Threads" at very bottom.

AnanthaP · 10-26-2017, 09:09 AM

rm -r *.ext ..
?

suicidaleggroll · 10-26-2017, 10:10 AM

Quote:

Originally Posted by AnanthaP

rm -r *.ext ..
?

That would delete any file or directory in the current directory named *.ext. The -r flag just means if you give it a directory, it will delete all of its contents as well. That's different than what the OP had asked for because it won't search out any file named *.ext located inside subdirectories, it will only match files/dirs named *.ext in the current directory. For example, if you had a directory named "blah", and inside a file named "file.ext", "rm -r *.ext" would NOT delete it, while "find . -name '*.ext' -delete" would.

allend · 10-26-2017, 05:20 PM

Setting a shell option in bash allows recursion into subdirectories.

Code:

shopt -s globstar; rm ./**/*.[a-d]; shopt -u globstar

TobiSGD · 10-27-2017, 06:48 AM

Quote:

Originally Posted by suicidaleggroll

Code:

find . -iname '*.a' -or -iname '*.b' -or -iname '*.c' -exec rm {} \;

Of course you should run it without the "-exec rm {} \;" at first to make sure the files it finds are what you actually want to delete.

This is inefficient. The -exec command at the end will call the rm command for every file it finds separately, which produces a lot of overhead. A better way to do it is to either use the inbult -delete command, or to use

Code:

-exec rm {} +

instead of

Code:

-exec rm {} \;

. The + sign signals the find command that you expect to find more than one file and that it should accumulate filenames until you reach the maximal allowed commandline length and then to call the rm command with all the accumulated files.
This may not be relevant if you only have a few files to delete, but if you traverse over large directories with thousands of files to delete you will see significant performance improvements.