2

I have an issue. I have a folder with many files and i need to execute a program on all combinations of 2 files among my files.

My linux bash script looks like this so far:

for ex in $(ls ${ex_folder}/${testset_folder} | sort -V ); do
   #ex is the name of my current file
   #I would like to do something like a nested loop where
   # ex2 goes from ex to the end of the list $(ls ${ex_folder}/${testset_folder} | sort -V )
done

I'm new to bash, in other languages this would look something like:

for i in [0,N]
  for j in [i,N]
    #the combination would be i,j

My list of files looks like the following:

ex_10.1 ex_10.2 ex_10.3 ex_10.4 ex_10.5

And i want to execute a python program on all combinations of 2 files among these (so i execute my program 10 times)

Thank you in advance for your help!

Roulbacha
  • 457
  • 7
  • 20
  • yes, `for` loops can be nested. Just make sure you have a closing `do/done` for each. If that's not enough to help you, then there is something about the files you aren't telling up. Update your Q to include a small set of filenames and show how you want them to be combined. Good luck. – shellter Oct 12 '17 at 21:27
  • (btw, not quite on-point for your question, but you should [avoid parsing `ls`](http://mywiki.wooledge.org/ParsingLs)). – Charles Duffy Oct 12 '17 at 21:31

1 Answers1

6

The logic you describe is easily transcribed if we use an array, and iterate by index:

files=( * )                                       # Assign your list of files to an array
max=${#files[@]}                                  # Take the length of that array

for ((idxA=0; idxA<max; idxA++)); do              # iterate idxA from 0 to length
  for ((idxB=idxA; idxB<max; idxB++)); do         # iterate idxB from idxA to length
    echo "A: ${files[$idxA]}; B: ${files[$idxB]}" # Do whatever you're here for.
  done
done

To safely implement sort -V (in a manner that doesn't allow malicious filenames or bugs to inject extra entries into the array), we'd want to replace that initial assignment line with logic akin to:

files=( )
while IFS= read -r -d '' file; do
  files+=( "$file" )
done < <(printf '%s\0' * | sort -V -z)

...which uses NUL delimiters (which, unlike newlines, cannot exist as a literal in UNIX filenames) to separate names in the stream to and from sort.

Charles Duffy
  • 280,126
  • 43
  • 390
  • 441