script to clean old files from the mass spec computer

clean_files.bash

#!/bin/bash

# To use, save this script in the directory you want to clean,
# name it "clean_files.bash"
# open bash in the directory
# run $ bash clean_files.bash
# This will generate a list of all the deleted files.



## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## THIS PROCESS IS NOT REVERSIBLE !!!!!!!!!!!!!!!!!!!!
## Run only if you are sure ...

current_time=$(date "+%Y.%m.%d")
echo "Current Time : $current_time"
 
new_fileName=deleted_files.${current_time}.txt
echo "The list of the deleted files will be in: ${new_fileName}"

echo "Calculating file sizes to be deleted, this might be slow ... go for a coffee..."
echo "Size to be deleted on Files"
find -mmin +1051200 \
    -regextype posix-extended -regex \
    '.*\.(pdResult|byspec2|mzML|blib|pep.xml|pin|pout)$' \
    -type f -exec du -bc {} + | \
    grep total$ | \
    cut -f1 | \
    awk '{ total += $1 }; END { print total }' | \
    numfmt --to=iec-i --suffix=B --padding=7

echo "Size to be deleted on folders"
find -mmin +1051200 \
    -regextype posix-extended -regex \
    '.*/(p0|combined/search|combined/andromeda|.Rproj.user|.RData)$' \
    -type d -exec du -bc {} + | \
    grep total$ | \
    cut -f1 | \
    awk '{ total += $1 }; END { print total }' | \
    numfmt --to=iec-i --suffix=B --padding=7


# Prompt for confirmation before actually deleting stuff ...
read -p "Are you sure? (y/n)" -n 1 -r

echo ""   # (optional) move to a new line
if [[ ! $REPLY =~ ^[Yy]$ ]]
then
    exit 1
fi

set -x
set -e

# This secttion deleted partial searches from PD
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*.pdResult$"
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.pdResult$" -delete -print |& tee --append ${new_fileName}

# This for byonic results
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.byspec2$" -delete -print |& tee --append ${new_fileName}

# This secttion deleted partial searches from open source search engines
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.mzML$" -delete -print |& tee --append ${new_fileName}
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.blib$" -delete -print |& tee --append ${new_fileName}
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.pep.xml$" -delete -print |& tee --append ${new_fileName}
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.pin$" -delete -print |& tee --append ${new_fileName}
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*\.pout$" -delete -print |& tee --append ${new_fileName}

# -type d finds only directories
# This section will delete old partial maxquant search results
find . -type d -mmin +1051200 -regextype posix-extended -regex ".*/p0$" -prune -exec rm -vr {} \; |& tee --append ${new_fileName}
find . -type d -mmin +1051200 -regextype posix-extended -regex ".*/combined/search$" -prune -exec rm -vr {} \; |& tee --append ${new_fileName}
find . -type d -mmin +1051200 -regextype posix-extended -regex ".*/combined/andromeda$" -prune -exec rm -vr {} \; |& tee --append ${new_fileName}

# This section deletes R projects
find . -type d -mmin +1051200 -regextype posix-extended -regex ".*/\.Rproj.user$" -prune -exec rm -vr {} \; |& tee --append ${new_fileName}
find . -type f -mmin +1051200 -regextype posix-extended -regex ".*/\.RData$" -delete -print |& tee --append ${new_fileName}

# This removes empty directories
find . -type d -empty -mmin +1051200 -delete -print |& tee --append ${new_fileName}

echo "Done deleting stuff"
wc -l ${new_fileName}