replace string with underscore and dots using sed or awk

Question

I have a bunch of files with filenames composed of underscore and dots, here is one example:

META_ALL_whrAdjBMI_GLOBAL_August2016.bed.nodup.sortedbed.roadmap.sort.fgwas.gz.r0-ADRL.GLND.FET-EnhA.out.params

I want to remove the part that contains .bed.nodup.sortedbed.roadmap.sort.fgwas.gz. so the expected filename output would be META_ALL_whrAdjBMI_GLOBAL_August2016.r0-ADRL.GLND.FET-EnhA.out.params

I am using these sed commands but neither one works:

stringZ=META_ALL_whrAdjBMI_GLOBAL_August2016.bed.nodup.sortedbed.roadmap.sort.fgwas.gz.r0-ADRL.GLND.FET-EnhA.out.params
echo $stringZ | sed -e 's/\([[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.\)//g'
echo $stringZ | sed -e 's/\[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.[[:lower:]]\.//g'

Any solution is sed or awk would help a lot

Answer 1

Don't use external utilities and regexes for such a simple task! Use parameter expansions instead.

stringZ=META_ALL_whrAdjBMI_GLOBAL_August2016.bed.nodup.sortedbed.roadmap.sort.fgwas.gz.r0-ADRL.GLND.FET-EnhA.out.params
echo "${stringZ/.bed.nodup.sortedbed.roadmap.sort.fgwas.gz}"

To perform the renaming of all the files containing .bed.nodup.sortedbed.roadmap.sort.fgwas.gz , use this:

shopt -s nullglob
substring=.bed.nodup.sortedbed.roadmap.sort.fgwas.gz
for file in *"$substring"*; do
    echo mv -- "$file" "${file/"$substring"}"
done

Note. I left echo in front of mv so that nothing is going to be renamed; the commands will only be displayed on your terminal. Remove echo if you're satisfied with what you see.

Answer 2

Your regex doesn't really feel too much more general than the fixed pattern would be, but if you want to make it work, you need to allow for more than one lower case character between each dot. Right now you're looking for exactly one, but you can fix it with \\+ after each [[:lower:]] like

printf '%s' "$stringZ" | sed -e 's/\([[:lower:]]\+\.[[:lower:]]\+\.[[:lower:]]\+\.[[:lower:]]\+\.[[:lower:]]\+\.[[:lower:]]\+\.[[:lower:]]\+\.\)//g'

which with

stringZ="META_ALL_whrAdjBMI_GLOBAL_August2016.bed.nodup.sortedbed.roadmap.sort.fgwas.gz.r0-ADRL.GLND.FET-EnhA.out.params"

give me the output

META_ALL_whrAdjBMI_GLOBAL_August2016.r0-ADRL.GLND.FET-EnhA.out.params

Answer 3

Try this:

#!/bin/bash
for line in $(ls -1 META*);
do
f2=$(echo $line | sed 's/.bed.nodup.sortedbed.roadmap.sort.fgwas.gz//')
mv $line $f2
done