#!/usr/bin/env bash
# Runner disk cleanup script for Kindred Create CI/CD
#
# Designed to run as a cron job on the pubworker host:
#   */30 * * * * /path/to/cleanup.sh >> /var/log/runner-cleanup.log 2>&1
#
# Or install the systemd timer (see cleanup.timer / cleanup.service).
#
# What it cleans:
#   1. Docker: stopped containers, dangling images, build cache
#   2. act_runner action cache: keeps only the newest entry per key prefix
#   3. act_runner workspaces: removes leftover build workspaces
#   4. System: apt cache, old logs
#
# What it preserves:
#   - The current runner container and its image
#   - The most recent cache entry per prefix (so ccache hits still work)
#   - Everything outside of known CI paths

set -euo pipefail

# ---------------------------------------------------------------------------
# Configuration -- adjust these to match your runner setup
# ---------------------------------------------------------------------------

# Disk usage threshold (percent) -- only run aggressive cleanup above this
THRESHOLD=${CLEANUP_THRESHOLD:-85}

# act_runner cache directory (default location)
CACHE_DIR=${CACHE_DIR:-/root/.cache/actcache}

# act_runner workspace directories
WORKSPACES=(
    "/root/.cache/act"
    "/workspace"
)

# Maximum age (days) for cache entries before unconditional deletion
CACHE_MAX_AGE_DAYS=${CACHE_MAX_AGE_DAYS:-7}

# Maximum age (days) for Docker images not used by running containers
DOCKER_IMAGE_MAX_AGE=${DOCKER_IMAGE_MAX_AGE:-48h}

# Log prefix
LOG_PREFIX="[runner-cleanup]"

# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------

log() { echo "$(date '+%Y-%m-%d %H:%M:%S') ${LOG_PREFIX} $*"; }

disk_usage_pct() {
    df --output=pcent / | tail -1 | tr -dc '0-9'
}

bytes_to_human() {
    numfmt --to=iec-i --suffix=B "$1" 2>/dev/null || echo "${1}B"
}

# ---------------------------------------------------------------------------
# Phase 1: Check if cleanup is needed
# ---------------------------------------------------------------------------

usage=$(disk_usage_pct)
log "Disk usage: ${usage}%  (threshold: ${THRESHOLD}%)"

if [ "$usage" -lt "$THRESHOLD" ]; then
    log "Below threshold, running light cleanup only"
    AGGRESSIVE=false
else
    log "Above threshold, running aggressive cleanup"
    AGGRESSIVE=true
fi

# ---------------------------------------------------------------------------
# Phase 2: Docker cleanup (always runs, safe)
# ---------------------------------------------------------------------------

log "--- Docker cleanup ---"

# Remove stopped containers
stopped=$(docker ps -aq --filter status=exited --filter status=dead 2>/dev/null | wc -l)
if [ "$stopped" -gt 0 ]; then
    docker rm $(docker ps -aq --filter status=exited --filter status=dead) 2>/dev/null || true
    log "Removed ${stopped} stopped containers"
fi

# Remove dangling images (untagged layers)
dangling=$(docker images -q --filter dangling=true 2>/dev/null | wc -l)
if [ "$dangling" -gt 0 ]; then
    docker rmi $(docker images -q --filter dangling=true) 2>/dev/null || true
    log "Removed ${dangling} dangling images"
fi

# Prune build cache
docker builder prune -f --filter "until=${DOCKER_IMAGE_MAX_AGE}" 2>/dev/null || true
log "Pruned Docker build cache older than ${DOCKER_IMAGE_MAX_AGE}"

if [ "$AGGRESSIVE" = true ]; then
    # Remove all images not used by running containers
    running_images=$(docker ps -q 2>/dev/null | xargs -r docker inspect --format='{{.Image}}' | sort -u)
    all_images=$(docker images -q 2>/dev/null | sort -u)
    for img in $all_images; do
        if ! echo "$running_images" | grep -q "$img"; then
            docker rmi -f "$img" 2>/dev/null || true
        fi
    done
    log "Removed unused Docker images (aggressive)"

    # Prune volumes
    docker volume prune -f 2>/dev/null || true
    log "Pruned unused Docker volumes"
fi

# ---------------------------------------------------------------------------
# Phase 3: act_runner action cache cleanup
# ---------------------------------------------------------------------------

log "--- Action cache cleanup ---"

if [ -d "$CACHE_DIR" ]; then
    before=$(du -sb "$CACHE_DIR" 2>/dev/null | cut -f1)

    # Delete cache entries older than max age
    find "$CACHE_DIR" -type f -mtime "+${CACHE_MAX_AGE_DAYS}" -delete 2>/dev/null || true
    find "$CACHE_DIR" -type d -empty -delete 2>/dev/null || true

    after=$(du -sb "$CACHE_DIR" 2>/dev/null | cut -f1)
    freed=$((before - after))
    log "Cache cleanup freed $(bytes_to_human $freed) (entries older than ${CACHE_MAX_AGE_DAYS}d)"
else
    log "Cache directory not found: ${CACHE_DIR}"

    # Try common alternative locations
    for alt in /var/lib/act_runner/.cache/actcache /home/*/.cache/actcache; do
        if [ -d "$alt" ]; then
            log "Found cache at: $alt (update CACHE_DIR config)"
            CACHE_DIR="$alt"
            find "$CACHE_DIR" -type f -mtime "+${CACHE_MAX_AGE_DAYS}" -delete 2>/dev/null || true
            find "$CACHE_DIR" -type d -empty -delete 2>/dev/null || true
            break
        fi
    done
fi

# ---------------------------------------------------------------------------
# Phase 4: Workspace cleanup
# ---------------------------------------------------------------------------

log "--- Workspace cleanup ---"

for ws in "${WORKSPACES[@]}"; do
    if [ -d "$ws" ]; then
        # Remove workspace dirs not modified in the last 2 hours
        # (active builds should be touching files continuously)
        before=$(du -sb "$ws" 2>/dev/null | cut -f1)
        find "$ws" -mindepth 1 -maxdepth 1 -type d -mmin +120 -exec rm -rf {} + 2>/dev/null || true
        after=$(du -sb "$ws" 2>/dev/null | cut -f1)
        freed=$((before - after))
        if [ "$freed" -gt 0 ]; then
            log "Workspace $ws: freed $(bytes_to_human $freed)"
        fi
    fi
done

# ---------------------------------------------------------------------------
# Phase 5: System cleanup
# ---------------------------------------------------------------------------

log "--- System cleanup ---"

# apt cache
apt-get clean 2>/dev/null || true

# Truncate large log files (keep last 1000 lines)
for logfile in /var/log/syslog /var/log/daemon.log /var/log/kern.log; do
    if [ -f "$logfile" ] && [ "$(stat -c%s "$logfile" 2>/dev/null)" -gt 104857600 ]; then
        tail -1000 "$logfile" > "${logfile}.tmp" && mv "${logfile}.tmp" "$logfile"
        log "Truncated $logfile (was >100MB)"
    fi
done

# Journal logs older than 3 days
journalctl --vacuum-time=3d 2>/dev/null || true

# ---------------------------------------------------------------------------
# Phase 6: Emergency cleanup (only if still critical)
# ---------------------------------------------------------------------------

usage=$(disk_usage_pct)
if [ "$usage" -gt 95 ]; then
    log "CRITICAL: Still at ${usage}% after cleanup"

    # Nuclear option: remove ALL docker data except running containers
    docker system prune -af --volumes 2>/dev/null || true
    log "Ran docker system prune -af --volumes"

    # Clear entire action cache
    if [ -d "$CACHE_DIR" ]; then
        rm -rf "${CACHE_DIR:?}/"*
        log "Cleared entire action cache"
    fi

    usage=$(disk_usage_pct)
    log "After emergency cleanup: ${usage}%"
fi

# ---------------------------------------------------------------------------
# Summary
# ---------------------------------------------------------------------------

usage=$(disk_usage_pct)
log "Cleanup complete. Disk usage: ${usage}%"

# Report top space consumers for diagnostics
log "Top 10 directories under /var:"
du -sh /var/*/ 2>/dev/null | sort -rh | head -10 | while read -r line; do
    log "  $line"
done