[SOLVED] How to rename files in a folder with specific conditions?

Issue

S_1004_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_1_fastqc.html
S_1004_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_1_fastqc.zip
S_1004_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_2_fastqc.html
S_1004_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_2_fastqc.zip
S_1006_DKDL220006298-1A_HKFTLDSX3_L1_cleaned_1_fastqc.html
S_1006_DKDL220006298-1A_HKFTLDSX3_L1_cleaned_1_fastqc.zip
S_1006_DKDL220006298-1A_HKFTLDSX3_L1_cleaned_2_fastqc.html

above are the name of the files in a folder. I want to remove between second _ from right and second _ from left. So that the output looks like

S_1004__1_fastqc.html
S_1004__1_fastqc.zip
S_1004__2_fastqc.html
S_1004__2_fastqc.zip
S_1006__1_fastqc.html
S_1006__1_fastqc.zip
S_1006__2_fastqc.html

How do I do this using bash?

I tried the following code:

for file in *.html *.zip; do
  new_name=$(echo "$file" | sed 's/_[^_]*_/_/')
  mv "$file" "$new_name"
done

but it did not work the way I want.

Solution

You don't need sed for this (but see at the end for an explanation why your attempt fails, and for a working sed command).

With a recent enough bash (at least 3.0 for [[ string =~ regexp ]] and BASH_REMATCH):

for f in *.html *.zip; do
  [[ "$f" =~ ^(([^_]*_){2}).+((_[^_]*){2})$ ]] && mv "$f" "${BASH_REMATCH[1]}${BASH_REMATCH[3]}"
done

With an older bash:

for f in *.html *.zip; do
  set -f; ifs="$IFS"; IFS=_ a=( $f ); IFS="$ifs"; set +f; n="${#a[@]}"
  (( n > 4 )) && mv "$f" "${a[0]}_${a[1]}__${a[n-2]}_${a[n-1]}"
done

Note: set -f; ...; set +f to temporarily suppress pathname expansion because your file names could contain glob operators (*, ?, [...]).

The reason why your attempt fails is that sed attempts to match regular expressions from left to right (and is greedy). In sed 's/_[^_]*_/_/' the leftmost match is thus substituted. If the file name is S_1004_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_1_fastqc.html, the matched part is _1004_ (leftmost) and the result is S_DKDL220006264-1A_HLGLFDSX3_L4_cleaned_1_fastqc.html.

If you really want to use sed for this you could try:

sed 's/^\([^_]*_[^_]*_\).\+\(_[^_]*_[^_]*\)$/\1\2/'

Answered By - Renaud Pacalet

Answer Checked By - Marie Seifert (WPSolving Admin)