Replace multiple underscores in filename using bash script

Question

Doing some reading here and here I found this solution to replace two underscores in filenames with only one using bash:

for file in *; do
  f=${file//__/_}
  echo $f
done;

However how do I most easily expand this expression to replace an arbitrary number of underscores with only one?

If you have extended globbing (enable it using `shopt -s extglob` if it's not already enabled) you could use `f="${f//+(_)/_}"`. E.g. for the file "test_____something__else.txt", `for f in test_*; do f="${f//+(_)/_}"; echo $f; done` returns `test_something_else.txt`. More details here: https://askubuntu.com/a/889746 — jared_mamrot, Nov 30 '21 at 21:27

score 4 · Answer 1 · answered Nov 30 '21 at 21:33

4

Typically, it's going to be faster to just put your original code in a loop than to do anything else.

for file in *; do
  f=$file
  while [[ $f = *__* ]]; do
    f=${f//__/_}
  done
  echo "$f"
done

Even better, if you're on a modern shell release, you can enable extended globs, which provide regex-like functionality:

shopt -s extglob
for file in *; do
  f=${file//+(_)/_}
  echo "$f"
done

answered Nov 30 '21 at 21:33

Charles Duffy

280,126
43
390
441

Great, number 2 really made things easier. Being a sporadic bash user I find it very hard to figure out these things just by searching manuals and descriptions. – Oortone Nov 30 '21 at 22:04

Guilherme Moreira · Answer 2 · 2021-11-30T21:43:14.823

0

You could use a simple regex using sed

for file in *; do
  f=$(echo "$file" | sed -e 's/_\+/_/')
  echo "$f"
done;

This regex matches one or more underscores (_\+) and substitutes them with only one (_)

edited Nov 30 '21 at 21:43

answered Nov 30 '21 at 21:29

Guilherme Moreira

13
4

1

`echo "$file"`, not `echo $file`; and `echo "$f"`, not `echo $f`. See [I just assigned a variable, but `echo $variable` shows something else!](https://stackoverflow.com/questions/29378566/i-just-assigned-a-variable-but-echo-variable-shows-something-else) – Charles Duffy Nov 30 '21 at 21:30
1

(also, this is going to be a lot slower than using the OP's original code in a loop; starting a pipeline for every file is expensive!) – Charles Duffy Nov 30 '21 at 21:31
Edited my answer to correct the variable usage, and you are absolutely correct about it being slower. I wouldn't use this in any directory with a lot of files – Guilherme Moreira Nov 30 '21 at 21:45

score 0 · Answer 3 · answered Nov 30 '21 at 22:06

0

GNU tr has --squeeze-repeats:

$ echo foo_______bar | tr --squeeze-repeats _
foo_bar

If you're using BSD tr you can use -s instead:

$ echo foo_______bar | tr -s _
foo_bar

answered Nov 30 '21 at 22:06

l0b0

55,365
30
138
223

score 0 · Answer 4 · answered Dec 01 '21 at 11:14

This Shellcheck-clean pure shell code should work with any POSIX-compliant shell, including bash and dash:

for file in *; do
    while :; do
        case $file in
            *__*)   file=${file%%__*}_${file#*__};;
            *)      break;;
        esac
    done
    printf '%s\n' "$file"
done

${file%%__*} expands to $file with the first __ and all characters after it removed (e.g. a__b__c produces a).
${file#*__} expands to $file with all characters up to and including the first __ removed (e.g. a__b__c produces b__c).
See the accepted, and excellent, answer to Why is printf better than echo? for an explanation of why printf '%s\n' "$file" is used instead of echo "$file".

Replace multiple underscores in filename using bash script

4 Answers4