UNIX for Dummies Questions & Answers

Hi, I am fairly new to unix, and am trying to copy all files with the name "*.cons" within a directory (and all of the many directories within it) to a new directory called "output". There are multiple nested directories, and I would like to just pull out the files with ".cons" and not the other stuff. Do I use the "find ./ -name" option? Can anyone help me?

You might want to read this thread for a short introduction to "find".

I hope this helps.

bakunin

Thank you. I tried: find ./ -iname "*.cons" -type f -exec mv {} /consensus \;
but it doesn't seem to work...

Not bad for the first try. You were almost there:

Lets start with the starting directory: you probably want to start in the current working directory, which is always named "." If this is the case you shouldn't obscure this pathname with additional slashes - in the best case this alters the path to something else, rendering it uninterpretable for the system in the worst case.

Btw., it is a good idea to "build gradually" towards the correct command. This way you easily find out where you have gone wrong. We start by simply printing what we find and use "more" to pagewise read output (read the mapage of "more" to see what it does):

Code:

find . -print | more

Ok, this seems to work. Now we want only the files, leaving out the directories:

Code:

find . -type f -print | more

The order of the clauses for "find" doesn't matter (not in your case), but it is a good idea to keep some sort of order to be able to read it easily. I made a habit of always put the "output clause" ("-print" in this case) at the end and the limiting clauses first. Whatever you do, be consistent with yourself.

As next step we limit the files (we already deal with files only, because of the "-type" clause) to the ones which match out name criteria. The clause for that is "-name" (not "-iname", as you wrote, probably a typo):

Code:

find . -type f -name "*.cons" -print | more

Just an observation: in DOS/Windows all files have a filename and an extension. These parts of the filename are divided by a dot. In UNIX there is no such thing as an extension and the dot (if there is one) is a simple part of the name. This makes not a big difference here, but is good to know, because sometimes people implicitly assume the naming convention of DOS/Windows to be in effect, which doesn't have to be the case.

We seem to have gotten all the files we want and left out all the files we don't want (yes? - if not, we will have to start over) and get to the hard part: moving the files. Again we first want to see what we might do to the files and only then do it in fact. There lets "mask" the command with an "echo" command to just print it instead of executing:

Code:

find . -type f -name "*.cons" -exec echo mv {} /consensus \; | more

The output should look like this:

Code:

user@host:/some/where $ find . -type f -name "*.cons" -exec echo mv {} /consensus \; | more
mv ./a.cons /consensus
mv ./b.cons /consensus
mv ./c.cons /consensus
mv ./d.cons /consensus
.
.
.
-More-

If the commands look like the ones you really want to execute remove the "echo" in the "-exec" part and execute the "find"-command once more to actually do it. You should be done. If you are not it will be things outside the scope of "find": does the directory "/consensus" already exist? Are your rights sufficient to write there? Is there enough space in the filesystem to move the files there? Are you allowed to access the files you want to move or only their directory information? Etc., etc.

I hope this helps.

bakunin

If you want to retain the directory tree structure in the target directory it is easier with "find" combined with "cpio -p". See "man find" and "man cpio".

This post is a start:

https://www.unix.com/shell-programmin...g-folders.html

Thank you very much! This is very helpful!!!