mv command error "already existing file"

Question

I am using a simple script to create some directories. Finding some specific files by name and then moving them into the directories that I have created. The code that I am using is below:

---

mkdir ./3DBGB_run   
mkdir ./run_optimization   
mkdir ./run_optimization/project_mesh   
mkdir ./run_optimization/CFD_optimized    
find . -name '*.template' -exec mv -i {} ./3DBGB_run \;                 
find . -name '3dbgbinput.*.dat' -exec mv -i {} ./3DBGB_run \;           
find . -name 'controlinputs.*.dat' -exec mv -i {} ./3DBGB_run \;            
find . -name '*.iec' -exec mv -i {} ./run_optimization/project_mesh \;           
find . -name '*.trb' -exec mv -i {} ./run_optimization/project_mesh \;  
find . -name '*.py' -exec mv -i {} ./run_optimization \;**

---

Code runs and moves the files to the desired directories but then it also display an error message for each file that "cannot move file because it is the same file"

Answer 1

You should avoid search in target directory. So you can add ! -path $TARGET_DIR as for example:

find . -name '*.template' ! -path "*3DBGB_run/*" -exec mv -it ./3DBGB_run {} \+

Answer 2

find is finding your files twice: in the place where they started, and where you moved them to. That means it ends up running:

mv -i 3DBGB_run/x.template 3DBGB_run/x.template

and produces the error you see.

find sees files "live" as it goes - it doesn't build up a list in advance and then run the command for everything in the list. The order that it looks at files and directories is undefined (it's likely to come from the filesystem), so it will happen for every file that's examined before the directory you move it into.

Because find doesn't know what you're doing in the -exec, it can't compensate for it. You can do a couple of things about this:

1. Exclude the directories from the search space: find . -name '*.template' -path "./3DBGB_run" -prune -o -exec mv -i {} ./3DBGB_run \;. The -path part matches the directory you're moving things into, and -prune then excludes that whole tree from recursion. -o causes the rest of the command (your original -exec) to run the rest of the time. The command never runs on files inside the directory. (This is an explicit example in the manual of GNU find)

2. Build up the list in advance of the move. This may be better in some circumstances, but if your filenames have spaces or other IFS characters then it doesn't work out so well. If that doesn't apply, something like:

   files=$(find . -name '*.template')
   mv "${files[@]}" 3DBGB_run
   

   will also work.

---

When you're using find -exec, ending the command with + instead of ; will make find run the minimum number of commands, collecting multiple arguments into each execution, which is more efficient. Sometimes that doesn't work for a particular command, but it would be fine here. If this is a one-time command it doesn't really matter, but if you run it a lot then saving processes will save time.