#!/usr/bin/env bash
# nexus installer — runs inside the LXC, called by ct/nexus.sh
#
# Sets up the deploy target for nexus (Family Knowledge Hub):
#   - Node.js (NodeSource) + git + rsync + sudo  (Node also needed by checkout action in host mode)
#   - act_runner in HOST mode (no Docker), registered to the Gitea instance,
#     running as the unprivileged `nexus` user and polling Gitea outbound
#   - a SKELETON systemd service (nexus.service) — the stack-introducing SDD
#     card in nexus-hub provides /opt/nexus/current/start.sh and may extend
#     the RUNTIME section below (DB clients, Python/uv, etc.)
#   - a narrow sudoers rule so the runner may only restart that one service
#
# The actual build/test/deploy logic lives in the nexus-hub repo workflows
# (.gitea/workflows/ci.yml + deploy.yml) — deploy-as-code.

set -euo pipefail

APP="nexus"
LIB_URL="${LIB_URL:-https://gitea.luki-net.org/luki-net/proxmox-scripts/raw/branch/main/lib}"
source <(curl -fsSL "$LIB_URL/install.func")

[[ "$EUID" -eq 0 ]] || { msg_err "Must run as root"; exit 1; }

# ── load deploy config pushed in by the host script ──────────────────────────
CONF="/root/nexus.deploy.env"
[[ -f "$CONF" ]] || { msg_err "$CONF not found (host bootstrap incomplete)"; exit 1; }
set -a; . "$CONF"; set +a

: "${GITEA_INSTANCE_URL:?missing GITEA_INSTANCE_URL}"
: "${RUNNER_TOKEN:?missing RUNNER_TOKEN}"
APP_PORT="${APP_PORT:-8080}"
NODE_MAJOR="${NODE_MAJOR:-22}"
RUNNER_VERSION="${RUNNER_VERSION:-0.2.13}"
RUNNER_LABELS="${RUNNER_LABELS:-nexus:host}"
RUNNER_NAME="${RUNNER_NAME:-$(hostname)}"

APP_USER="nexus"
APP_HOME="/opt/nexus"
RUNNER_DIR="$APP_HOME/runner"
CURRENT_DIR="$APP_HOME/current"
DATA_DIR="$APP_HOME/data"
CONF_DIR="/etc/nexus"
ENV_FILE="$CONF_DIR/env"

run_user() { runuser -u "$APP_USER" -- env HOME="$APP_HOME" "$@"; }

# ── packages: git + rsync + sudo ; Node.js via NodeSource ─────────────────────
setup_base_apt git rsync sudo
configure_ssh_root_login

NODE_HAVE="$(command -v node >/dev/null 2>&1 && node -v | sed -E 's/^v([0-9]+).*/\1/' || echo 0)"
if [[ "$NODE_HAVE" != "$NODE_MAJOR" ]]; then
  msg_info "Installing Node.js ${NODE_MAJOR}.x (NodeSource)..."
  curl -fsSL "https://deb.nodesource.com/setup_${NODE_MAJOR}.x" | bash - >/dev/null
  apt-get install -y -qq nodejs >/dev/null
  msg_ok "Node $(node -v) installed"
else
  msg_warn "Node $(node -v) already present, skipping"
fi

# ── RUNTIME ───────────────────────────────────────────────────────────────────
# Stack decided (nexus-hub ADR-0002, card K-101): Python 3.12 via uv, tesseract
# deu+eng, nexus-worker.service + sudoers extension. Provisioned by
# install/nexus-runtime.sh, invoked at the END of this script (it extends the
# sudoers rule and systemd units written below, so order matters). Commands are
# documented in nexus-hub docs/05_AGENT_RULES.md → Projekt-Kommandos.

# ── act_runner binary ─────────────────────────────────────────────────────────
if [[ ! -x /usr/local/bin/act_runner ]]; then
  ARCH="$(dpkg --print-architecture)"
  case "$ARCH" in amd64|arm64) ;; *) msg_err "unsupported arch: $ARCH"; exit 1 ;; esac
  msg_info "Downloading act_runner $RUNNER_VERSION ($ARCH)..."
  curl -fsSL "https://dl.gitea.com/act_runner/${RUNNER_VERSION}/act_runner-${RUNNER_VERSION}-linux-${ARCH}" \
    -o /usr/local/bin/act_runner
  chmod +x /usr/local/bin/act_runner
  msg_ok "act_runner $(/usr/local/bin/act_runner --version 2>/dev/null | head -n1)"
else
  msg_warn "act_runner already present, skipping download"
fi

# ── user + dirs ───────────────────────────────────────────────────────────────
create_system_user "$APP_USER" "$APP_HOME"
mkdir -p "$RUNNER_DIR" "$CURRENT_DIR" "$DATA_DIR" "$CONF_DIR"
chown -R "$APP_USER:$APP_USER" "$APP_HOME"

# ── env file (runtime config; populated further by the stack card) ────────────
cat >"$ENV_FILE" <<EOF
NEXUS_ENV=production
NEXUS_PORT=$APP_PORT
NEXUS_DATA_DIR=$DATA_DIR
EOF
chown root:"$APP_USER" "$ENV_FILE"
chmod 640 "$ENV_FILE"

# ── register the runner (idempotent) ──────────────────────────────────────────
if [[ ! -f "$RUNNER_DIR/.runner" ]]; then
  msg_info "Registering runner '$RUNNER_NAME' [$RUNNER_LABELS] with $GITEA_INSTANCE_URL..."
  run_user bash -c "cd '$RUNNER_DIR' && /usr/local/bin/act_runner register \
    --no-interactive \
    --instance '$GITEA_INSTANCE_URL' \
    --token '$RUNNER_TOKEN' \
    --name '$RUNNER_NAME' \
    --labels '$RUNNER_LABELS'"
  chown -R "$APP_USER:$APP_USER" "$RUNNER_DIR"
  msg_ok "Runner registered"
else
  msg_warn "Runner already registered (.runner exists), skipping"
fi

# ── narrow sudoers: runner may ONLY (re)start/stop/status nexus.service ───────
cat >/etc/sudoers.d/nexus-deploy <<EOF
$APP_USER ALL=(root) NOPASSWD: /usr/bin/systemctl restart nexus.service, /usr/bin/systemctl start nexus.service, /usr/bin/systemctl stop nexus.service, /usr/bin/systemctl status nexus.service
EOF
chmod 440 /etc/sudoers.d/nexus-deploy
visudo -cf /etc/sudoers.d/nexus-deploy >/dev/null

# ── systemd units ─────────────────────────────────────────────────────────────
# nexus.service: SKELETON. The deploy workflow populates /opt/nexus/current and
# the stack card provides current/start.sh. Enabled (starts on boot) but not
# started now — first successful deploy starts it via the sudoers-allowed restart.
cat >/etc/systemd/system/nexus.service <<EOF
[Unit]
Description=nexus — Family Knowledge Hub
After=network-online.target
Wants=network-online.target
# Skeleton-safe: only startable once the first deploy shipped the artifact
# (otherwise a reboot before first deploy leaves the unit in failed state).
ConditionPathExists=$CURRENT_DIR/start.sh

[Service]
Type=simple
User=$APP_USER
Group=$APP_USER
WorkingDirectory=$CURRENT_DIR
EnvironmentFile=$ENV_FILE
ExecStart=$CURRENT_DIR/start.sh
Restart=on-failure
RestartSec=5
NoNewPrivileges=true
ProtectSystem=full
PrivateTmp=true

[Install]
WantedBy=multi-user.target
EOF

cat >/etc/systemd/system/nexus-runner.service <<EOF
[Unit]
Description=nexus — Gitea Actions runner (host mode, CI + deploy)
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
User=$APP_USER
Group=$APP_USER
WorkingDirectory=$RUNNER_DIR
Environment=HOME=$APP_HOME
Environment=PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
ExecStart=/usr/local/bin/act_runner daemon
Restart=on-failure
RestartSec=5

[Install]
WantedBy=multi-user.target
EOF

systemctl daemon-reload
systemctl enable nexus.service >/dev/null 2>&1
systemctl enable --now nexus-runner.service
msg_ok "systemd units installed (runner started; nexus.service enabled, starts on first deploy)"

# ── RUNTIME provisioning (K-101): uv/Python 3.12, tesseract, worker unit ──────
# Runs LAST on purpose: it extends the sudoers rule and unit set from above.
# Idempotent — the same script retrofits an existing LXC:
#   curl -fsSL .../install/nexus-runtime.sh | bash
bash <(curl -fsSL "${LIB_URL%/lib}/install/nexus-runtime.sh")

# ── notes / summary file ──────────────────────────────────────────────────────
CRED_FILE="/root/nexus.credentials"
cat >"$CRED_FILE" <<EOF
nexus — Family Knowledge Hub deployed via a self-hosted Gitea Actions runner

Repo (informational):  ${REPO_URL:-https://gitea.luki-net.org/l.kirchner/nexus-hub.git}
Gitea instance:        $GITEA_INSTANCE_URL
Runner name / labels:  $RUNNER_NAME  [$RUNNER_LABELS]   (act_runner $RUNNER_VERSION, host mode)
API (local):           http://127.0.0.1:$APP_PORT   (live after first deploy + stack card)
Live copy:             $CURRENT_DIR
Data dir:              $DATA_DIR
Runtime env:           $ENV_FILE

CI + deploy are driven by the repo workflows .gitea/workflows/{ci,deploy}.yml
nexus.service is a SKELETON — ExecStart expects $CURRENT_DIR/start.sh, which
the stack-introducing SDD card provides. Until then only ci.yml is functional.

Manual restart:        sudo systemctl restart nexus.service
Logs:                  journalctl -u nexus -u nexus-runner -f

Prerequisites on the Gitea side:
  - Actions enabled on the instance and on the repo (repo -> Settings -> Actions)
  - The LXC needs outbound HTTPS to $GITEA_INSTANCE_URL and to github.com
    (the latter only to fetch actions/checkout, unless you self-host actions)
EOF
chmod 600 "$CRED_FILE"

# registration token already consumed → drop the bootstrap env file
shred -u "$CONF" 2>/dev/null || rm -f "$CONF"

apt_cleanup
msg_ok "$APP installation finished"