Recursively move specific files in nested directories to a name derived from the path

Question

I have some files in different nested directories. All the files are named Analysis.mzXML. I would like to have them all in one place, while changing the name of each one to something different, depending on the directories it was previously saved on.

I know that moving a file and giving it the directory name has been shown earlier in other questions such as this, this and this.

However, they only use the programs to move them one directory out and only change the name based on that one directory. In my case, I would like the files to be renamed using the entire route of nested directories

The files are all saved as follows:

/dir1/dir2/dir3/dir4/dir5/dir6/Analysis.mzXML

Each Analysis.mzXML is nested in folders of different names. However, dir6 happens to have the same name for all files, and since it no offers valuable information I would like to skip it.

My expectation is that by running some command or script I can get all these files named with their specific route as follows:

/dir1_dir2_dir3_dir4_dir5.mzXML

I have been just getting started with bash and I am not sure how to come around this issue.

Answer 1

Your question, as written, suggests that you want to search the entire filesystem for Analysis.mzXML files, even if they appear in /lib/perl5/5.14/Unicode/Collate/Locale, /proc/sys/net/ipv4/conf/all, /sys/devices/system/cpu/cpu0/cache, /usr/share/doc/cmake/html/_sources, or similarly unlikely / inappropriate places — and that you want to move the files into the root directory, which also seems unlikely / inappropriate.  I assumed that you are working from some other directory.

Run this command:

find . -mindepth 7 -maxdepth 7 -type f -name Analysis.mzXML -exec sh -c 'for arg do echo mv -i -- "$arg" "$(echo "$arg" | sed -e "s|^\./||" -e "s|/dir6/Analysis\.mzXML|.mzXML|" -e "s|/|_|g")"; done' sh {} +

You can leave out the -mindepth 7 -maxdepth 7 if you’re sure that all the Analysis.mzXML files are at the seventh level.  This command finds all Analysis.mzXML files at the seventh level and passes them to a mini-shell script.  That takes each pathname, strips off the ./ at the beginning and the dir6 at the end, along with the Analysis part of the filename, and changes all the remaining / characters to _.  This should produce an output something like this:

mv -i -- ./dir1/dir2/dir3/dir4/dir5/dir6/Analysis.mzXML dir1_dir2_dir3_dir4_dir5.mzXML
mv -i -- ./the/quick/brown/fox/jumps/dir6/Analysis.mzXML the_quick_brown_fox_jumps.mzXML
mv -i -- ./over/the/lazy/dog/foo bar/dir6/Analysis.mzXML over_the_lazy_dog_foo bar.mzXML

If this looks right, run the command again, but delete the first echo (i.e., change echo mv to just mv).

Notes:

• This doesn’t verify that the sixth level directory is actually called dir6.  If it finds an Analysis.mzXML file in dir11/dir12/dir13/dir14/dir15/dir16, it will rename that file to dir11_dir12_dir13_dir14_dir15_dir16_Analysis.mzXML rather than dir11_dir12_dir13_dir14_dir15.mzXML.
• The -i will cause mv to ask for confirmation if it tries to move a file to a name that already exists.
• The -- should protect you against arguments that begin with -.  This shouldn’t be an issue, since $arg should always begin with ./.
• If you really want to do this from the filesystem root, that should work.  Just cd / and follow the above instructions.

Afterthought: This might fail if you have a directory named Analysis.mzXML.  So don’t do that.

Answer 2

If you're moving only one file with mv, you specify the source filename and then the target filename OR the target directory if you want it to have the same name in a different directory. e.g.

mv /path/to/Analysis.mzXML /new/directory/new-filename.mzXML

or (keeping the same filename in the new directory):

mv /path/to/Analysis.mzXML /new/directory/

If you're moving multiple files, the target MUST be a directory. e.g.

 mv /path/to/*.mzXML /new/directory/

The same is true if you want to copy a file or files, with cp rather than mv.