I was reading a book on UNIX when I came across this question.
You have a directory structure
$HOME/a/a/b/c
where the first a is empty.
How do you remove it and move the lower directories up?
I have looked on the man pages of the commands rmdir and rm but could find nothing.
Please help me.
You should move subdirectory before remove parent directory. So..
mv $HOME/a/a $HOME
rmdir $HOME/a
But, this does not work because name a collide.
Rename the parent directory to avoid name collision.
mv $HOME/a $HOME/temp_a
mv $HOME/temp_a/a $HOME
rmdir $HOME/temp_a
Since you cannot move a/a into a, you could instead rename a.
mv a a_
mv a_/a .
rmdir a_
Related
I am trying to move 2 files from current directory (/base/level1/level2) to one directory below (/base/level1)
Is there any easier command other than mv file1 /base/level1 ? I'm trying to understand if we have some command that move it to a specific level up or down in the current folder structure.
TIA!
My solution to this one is navigating to the path where I want to move the file and run the following command.
cp ./level2/file1 .
or
cp ./file1 ./level2/file1
Please share other solutions as well
Thanks,
AMK
You could use wildcards if the 2 files have something uniquely in common.
ie. mv file*.ext path/to/new/dest/
This will move all files starting with "file" and ending with the extension ".ext" to the destination. Have a look at this and this which will explain wildcards a bit more
You can always use .. for "one directory up".
And you can give more than 2 arguments to mv, the last always being the destination.
So mv file1 file2 .. would move those 2 files a directory up.
Or mv * .. to move all files.
I have
A/1/a
A/1/b
B/1/a
B/1/b
all these are folders. I am trying to move the directories so that it may look like
A/a
A/b
B/a
B/b
I am sure I should use mv command but I am not sure to do it for all the directories at once.
Not in one go, but since we're not code golfing:
mv A/1/* A; rmdir A/1
mv B/1/* B; rmdir B/1
The canonical gotchas apply, like * not globbing dot files, depending on shell options. You'll know when you run into them when you see rmdir: directory not empty.
If I have a directory tree that looks like this:
parent/
child/
many files
What command/commands should I use to quickly make it like this:
parent/
many files
Effectively, I just want to remove one level of folder nesting. I don't care about any other files currently in parent.
This is what you do:
cd parent
mv child/* .
rmdir child
If you have files beginning with . (dot) in the child/ directory, you have to do
shopt -s dotglob
in order for the * to match them (there are other ways to force this, but setting dotglob is the easiest).
If child/ contains a file or directory named child, then you have to either rename that or the directory parent/child before the mv operation.
What about this?
mkdir new
find parent -type f -exec mv {} new/. \;
That would work for any depth of directory.
Hi this is a simple question but the solution eludes me at the moment..
I can find out the folder name that I want to change the name of, and I know the command to change the name of a folder is mv
so from the current directory if i go
ls ~/relevant.directory.containing.directory.name.i.want.to.change
to which i get the name of the directory is called say lorem-ipsum-v1-3
but the directory name may change in the future but it is the only directory in the directory:
~/relevant.directory.containing.directory.name.i.want.to.change
how to i programmatically change it to a specific name like correct-files
i can do it normally by just doing something like
mv lorem-ipsum-v1-3 correct-files
but I want to start automating this so that I don't need to keep copying and pasting the directory name....
any help would be appreciated...
Something like:
find . -depth -maxdepth 1 -type d | head -n 1 | xargs -I '{}' mv '{}' correct-files
should work fine as long as only one directory should be moved.
If you are absolutely certain that relevant.directory.containing.directory.name.i.want.to.change only contains the directory you want to rename, then you can simply use a wildcard:
mv ~/relevant.directory.containing.directory.name.i.want.to.change/*/ ~/relevant.directory.containing.directory.name.i.want.to.change/correct-files
This can can also be simplified further, using bash brace expansion, to:
mv ~/relevant.directory.containing.directory.name.i.want.to.change/{*/,correct-files}
cd ~/relevant.directory.containing.directory.name.i.want.to.change
find . -type d -print | while read a ;
do
mv $a correct-files ;
done
Caveats:
No error handling
There may be a way of reversing the parameters to mv so you can use xargs instead of a while loop, but that's not standard (as far as I'm aware)
Not parameterised
If there any any subdirectories it won't work. The depth parameters on the find command are (again, AFAIK) not standard. They do exist on GNU versions but seem to be missing on Solaris
Probably others...
I have a feeling that I already know the answer to this one, but I thought I'd check.
I have a number of different folders:
images_a/
images_b/
images_c/
Can I create some sort of symlink such that this new directory has the contents of all those directories? That is this new "images_all" would contain all the files in images_a, images_b and images_c?
No. You would have to symbolically link all the individual files.
What you could do is to create a job to run periodically which basically removed all of the existing symbolic links in images_all, then re-create the links for all files from the three other directories, but it's a bit of a kludge, something like this:
rm -f images_all/*
for i in images_[abc]/* ; do; ln -s $i images_all/$(basename $i) ; done
Note that, while this job is running, it may appear to other processes that the files have temporarily disappeared.
You will also need to watch out for the case where a single file name exists in two or more of the directories.
Having come back to this question after a while, it also occurs to me that you can minimise the time during which the files are not available.
If you link them to a different directory then do relatively fast mv operations that would minimise the time. Something like:
mkdir images_new
for i in images_[abc]/* ; do
ln -s $i images_new/$(basename $i)
done
# These next two commands are the minimal-time switchover.
mv images_all images_old
mv images_new images_all
rm -rf images_old
I haven't tested that so anyone implementing it will have to confirm the suitability or otherwise.
You could try a unioning file system like unionfs!
http://www.filesystems.org/project-unionfs.html
http://aufs.sourceforge.net/
to add on to paxdiablo 's great answer, i think you could use cp -s
(-s or --symbolic-link)
which makes symbolic links instead of literal copying
to maybe speed up or simplify the the bulk adding of symlinks to the "merge" folder A , of the files from folder B and C.
(i have not tested this though)
I cant recall of the top of my head, but im sure there is some option for CP to NOT overwrite existing, thus only symlinks of new files will be "cp -s" ed