#!/bin/bash

echo "Starting disk space report... This may take a few minutes." > /tmp/disk_report.txt

# Function to check if Docker is installed
is_docker_installed() {
    if ! command -v docker &>/dev/null; then
        echo "Docker is not installed. Skipping Docker-related checks." >> /tmp/disk_report.txt
        return 1
    fi
    return 0
}

# Function to estimate the size of the apt cache
estimate_apt_cache_size() {
    echo "Estimating the size of the apt cache:" >> /tmp/disk_report.txt
    sudo du -sh /var/cache/apt 2>/dev/null >> /tmp/disk_report.txt || echo "Error estimating apt cache size." >> /tmp/disk_report.txt
    echo >> /tmp/disk_report.txt
}

# Function to estimate the size of old installed packages
estimate_old_packages_size() {
    echo "Estimating the size of old installed packages:" >> /tmp/disk_report.txt
    sudo du -sh /var/lib/apt/lists /var/lib/apt/lists/partial 2>/dev/null >> /tmp/disk_report.txt || echo "Error estimating old packages size." >> /tmp/disk_report.txt
    echo >> /tmp/disk_report.txt
}

# Function to estimate the size of journal logs
estimate_journal_size() {
    echo "Estimating the size of journal logs:" >> /tmp/disk_report.txt
    sudo journalctl --disk-usage 2>/dev/null >> /tmp/disk_report.txt || echo "Error estimating journal log size." >> /tmp/disk_report.txt
    echo >> /tmp/disk_report.txt
}

# Function to estimate the size of temporary files
estimate_tmp_size() {
    echo "Estimating the size of temporary files:" >> /tmp/disk_report.txt
    sudo du -sh /tmp /var/tmp 2>/dev/null >> /tmp/disk_report.txt || echo "Error estimating temporary files size." >> /tmp/disk_report.txt
    echo >> /tmp/disk_report.txt
}

# Function to estimate the size of unused Docker volumes
estimate_docker_volumes_size() {
    if is_docker_installed; then
        echo "Estimating the size of unused Docker volumes:" >> /tmp/disk_report.txt
        docker volume ls -qf dangling=true | xargs -I {} docker volume inspect --format '{{ .Mountpoint }}' {} 2>/dev/null | xargs -I {} sudo du -sh {} 2>/dev/null | awk '{ sum += $1 } END { print sum "B" }' >> /tmp/disk_report.txt
        echo >> /tmp/disk_report.txt
    fi
}

# Function to check and suggest logrotate for large Docker logs
check_docker_logs() {
    if is_docker_installed; then
        echo "Checking Docker logs for large files..." >> /tmp/disk_report.txt
        large_logs=$(docker ps -q --filter "status=exited" | xargs -I {} docker inspect --format '{{.LogPath}}' {} 2>/dev/null | xargs -I {} sudo find {} -type f -size +1G 2>/dev/null)
        if [ -n "$large_logs" ]; then
            echo "The following Docker logs are larger than 1GB:" >> /tmp/disk_report.txt
            echo "$large_logs" >> /tmp/disk_report.txt
            echo >> /tmp/disk_report.txt
            echo "Consider setting up logrotate to manage Docker logs." >> /tmp/disk_report.txt
            echo "To truncate all Docker logs, run:" >> /tmp/disk_report.txt
            echo 'sudo find /var/lib/docker/containers/ -type f -name "*.log" -exec truncate -s 0 {} \;' >> /tmp/disk_report.txt
            echo >> /tmp/disk_report.txt
        else
            echo "No large Docker logs found." >> /tmp/disk_report.txt
            echo >> /tmp/disk_report.txt
        fi
    fi
}

# Function to list directories consuming more than a specified size
list_large_directories() {
    local directory=$1
    echo "Scanning $directory for large directories..." >> /tmp/disk_report.txt
    sudo du -ahx $directory --exclude=/home/virtfs 2>/dev/null | sort -hr > /tmp/du_output.txt || echo "Error scanning directories in $directory." >> /tmp/disk_report.txt
    cat /tmp/du_output.txt >> /tmp/disk_report.txt
}


# Function to parse and categorize directory sizes
categorize_large_directories() {
    echo "Categorizing large directories..." >> /tmp/disk_report.txt
    awk '
    {
        size=$1; path=$2;
        unit=substr(size, length(size));
        size_val=substr(size, 1, length(size)-1);
        if (unit == "G" || unit == "T") {
            if (unit == "T") size_val *= 1024;
            if (size_val > 50) print "> 50GB: " path;
            else if (size_val > 20) print "> 20GB: " path;
            else if (size_val > 10) print "> 10GB: " path;
            else if (size_val > 5) print "> 5GB: " path;
            else if (size_val > 1) print "> 1GB: " path;
        }
    }' /tmp/du_output.txt >> /tmp/disk_report.txt
    echo >> /tmp/disk_report.txt
}

# Estimate storage savings
echo "Estimating potential storage savings..." >> /tmp/disk_report.txt
estimate_apt_cache_size
estimate_old_packages_size
estimate_journal_size
estimate_tmp_size
estimate_docker_volumes_size

# Check Docker logs
check_docker_logs

# List large directories and categorize them
list_large_directories /
categorize_large_directories

list_large_directories /home
categorize_large_directories

# Scan and report specifically for /home/virtfs
echo "Scanning /home/virtfs for large directories..." >> /tmp/disk_report.txt
sudo du -ahx /home/virtfs 2>/dev/null | sort -hr > /tmp/du_output.txt || echo "Error scanning directories in /home/virtfs." >> /tmp/disk_report.txt
categorize_large_directories

echo "Storage savings estimation and large directory listing completed." >> /tmp/disk_report.txt

# Upload the report to hastebin
response=$(curl -s -X POST -T /tmp/disk_report.txt https://haste.nixc.us/documents)
if [[ $response == *"key"* ]]; then
    key=$(echo $response | jq -r '.key')
    echo "Report available at: https://haste.nixc.us/$key"
else
    echo "Failed to upload report to haste.nixc.us"
fi