#!/usr/bin/env bash
# Health check for the deployed torly relay.
# Runs from torly-monitor.timer every five minutes and writes a JSON status.
set -uo pipefail

DATA=${TORLY_DATA_DIR:-/home/mleku/.local/share/torly}
HTTP=${TORLY_HTTP:-http://127.0.0.1:3336}
STATUS=$DATA/monitor-status.json
FAILURES=$DATA/monitor-failures
TORLY_UNIT=${TORLY_UNIT:-torly.service}

mkdir -p "$DATA"

health=false
if curl -fsS --max-time 5 "$HTTP/health" 2>/dev/null |
  jq -e '.status == "ok"' >/dev/null 2>&1; then
  health=true
fi

nip11=false
if curl -fsS --max-time 5 -H 'Accept: application/nostr+json' "$HTTP/" 2>/dev/null |
  jq -e '.name and (.supported_nips | length > 0) and .pubkey' >/dev/null 2>&1; then
  nip11=true
fi

torly_active=false
systemctl is-active --quiet "$TORLY_UNIT" && torly_active=true

healthy=false
if [ "$health" = true ] && [ "$nip11" = true ] && [ "$torly_active" = true ]; then
  healthy=true
fi

# Restart a genuinely dead service after three consecutive unhealthy runs.
if [ "$health" = true ]; then
  rm -f "$FAILURES"
else
  n=0
  [ -f "$FAILURES" ] && n=$(cat "$FAILURES" 2>/dev/null || echo 0)
  n=$((n + 1))
  echo "$n" >"$FAILURES"
  if [ "$n" -ge 3 ]; then
    echo "torly-monitor: health failed $n times, restarting $TORLY_UNIT" >&2
    systemctl restart "$TORLY_UNIT" 2>/dev/null || true
    rm -f "$FAILURES"
  fi
fi

jq -n \
  --arg updated_at "$(date -u +%Y-%m-%dT%H:%M:%SZ)" \
  --argjson healthy "$healthy" \
  --argjson health "$health" \
  --argjson nip11 "$nip11" \
  --argjson torly_active "$torly_active" \
  '{updated_at: $updated_at, healthy: $healthy, health: $health, nip11: $nip11,
    torly_active: $torly_active}' >"$STATUS"

if [ "$healthy" = true ]; then
  echo "torly-monitor: ok"
  exit 0
fi
echo "torly-monitor: unhealthy health=$health nip11=$nip11 torly=$torly_active" >&2
exit 1