Disk Cleanup

cleanup.sh

#!/bin/bash
# Remove files from a folder up to certain accumulated size to free up space

folder_path="/var/www/vhosts/"  # Replace with the actual folder path
max_size_bytes=$((2 * 1024 * 1024 * 1024))  # 2GB in bytes
total_size=0
num_files=0
selected_files=()
max_date=""
min_date=""
total_date=0

# Find and calculate the size of largest files until total size matches 2GB
while IFS= read -r line; do
    file_size=$(echo "$line" | awk '{print $1}')
    file_path=$(echo "$line" | cut -d$'\t' -f2-)
    file_name=$(basename "$file_path")
    file_date=$(stat -c %Y "$file_path")

    total_size=$((total_size + file_size))

    if [[ $total_size -le $max_size_bytes ]]; then
        echo "$file_size / $total_size | Path: $file_name"
        selected_files+=("$file_path")

        if [[ -z $min_date || $file_date -lt $min_date ]]; then
            min_date=$file_date
        fi

        if [[ -z $max_date || $file_date -gt $max_date ]]; then
            max_date=$file_date
        fi

        total_date=$((total_date + file_date))
        ((num_files++))
    else
        break
    fi
done < <(find "$folder_path" -type f -exec du -b {} + | sort -rn)

if [[ $num_files -gt 0 ]]; then
    average_date=$(date -d "@$((total_date / num_files))" "+%Y-%m")
    min_date_f=$(date -d "@$min_date" "+%Y-%m")
    max_date_f=$(date -d "@$max_date" "+%Y-%m")
    echo "Date Range: $min_date_f to $max_date_f. Average: $average_date"
fi

# Prompt for user confirmation before removal
if [[ $num_files -gt 0 ]]; then
    read -p "Do you want to remove these $num_files files? (yes/no): " confirmation
    if [[ $confirmation == "yes" ]]; then
        for file_path in "${selected_files[@]}"; do
            echo "deleting $file_path ..."
            #rm -i "$file_path"
        done
        echo "Files removed."
    else
        echo "Files not removed."
    fi
else
    echo "No files found to remove."
fi