claude-code-hub/scripts/cch

#!/usr/bin/env bash
# cch - Claude Code Hub Kubernetes Management CLI
# 兼容 k3s 与标准 Kubernetes,通过 env / ~/.config/cch/config 可覆盖默认值
# Reference: docs/k8s-deployment.md
set -euo pipefail

VERSION="1.0.0"

###############################################################################
# Colors (非 TTY / NO_COLOR 自动禁用)
###############################################################################
if [[ -t 1 ]] && [[ -z "${NO_COLOR:-}" ]]; then
    RED=$'\033[0;31m'
    GREEN=$'\033[0;32m'
    YELLOW=$'\033[1;33m'
    CYAN=$'\033[0;36m'
    NC=$'\033[0m'
else
    RED=""; GREEN=""; YELLOW=""; CYAN=""; NC=""
fi

info()  { echo -e "${CYAN}[INFO]${NC}  $*"; }
ok()    { echo -e "${GREEN}[OK]${NC}    $*"; }
warn()  { echo -e "${YELLOW}[WARN]${NC}  $*"; }
err()   { echo -e "${RED}[ERROR]${NC} $*" >&2; }

# 跨平台 base64 decode:GNU coreutils 用 -d,旧版 macOS BSD 只认 -D
b64d() {
    if base64 -d </dev/null >/dev/null 2>&1; then
        base64 -d
    elif base64 -D </dev/null >/dev/null 2>&1; then
        base64 -D
    else
        # 兜底:openssl 几乎所有平台都有
        openssl base64 -d
    fi
}

###############################################################################
# 配置解析 (优先级: env > config file > 默认)
###############################################################################
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"

parse_config_value() {
    local raw="$1"
    raw="${raw#"${raw%%[![:space:]]*}"}"
    raw="${raw%"${raw##*[![:space:]]}"}"

    if [[ "$raw" =~ ^\"(.*)\"$ ]]; then
        printf '%s' "${BASH_REMATCH[1]}"
        return
    fi
    if [[ "$raw" =~ ^\'(.*)\'$ ]]; then
        printf '%s' "${BASH_REMATCH[1]}"
        return
    fi
    printf '%s' "$raw"
}

load_config_file() {
    [[ -r "$CCH_CONFIG_FILE" ]] || return 0

    local line key raw value
    while IFS= read -r line || [[ -n "$line" ]]; do
        [[ "$line" =~ ^[[:space:]]*# ]] && continue
        [[ "$line" =~ ^[[:space:]]*$ ]] && continue
        if [[ ! "$line" =~ ^[[:space:]]*([A-Z0-9_]+)[[:space:]]*=(.*)$ ]]; then
            warn "跳过无法解析的配置行: $line"
            continue
        fi

        key="${BASH_REMATCH[1]}"
        raw="${BASH_REMATCH[2]}"
        case "$key" in
            CCH_NAMESPACE|CCH_IMAGE|CCH_DEPLOY_DIR|CCH_RUNTIME|CCH_INGRESS_HOST|CCH_INGRESS_VARIANT|CCH_BACKUP_DIR|CCH_BACKUP_KEEP)
                if [[ -z "${!key:-}" ]]; then
                    value="$(parse_config_value "$raw")"
                    printf -v "$key" '%s' "$value"
                fi
                ;;
        esac
    done < "$CCH_CONFIG_FILE"
}

# 1. 加载可选配置文件
CCH_CONFIG_FILE="${CCH_CONFIG_FILE:-${XDG_CONFIG_HOME:-$HOME/.config}/cch/config}"
load_config_file

# 2. 合并默认值
NAMESPACE="${CCH_NAMESPACE:-claude-code-hub}"
IMAGE="${CCH_IMAGE:-ghcr.io/ding113/claude-code-hub:latest}"
RUNTIME_OVERRIDE="${CCH_RUNTIME:-}"
INGRESS_HOST="${CCH_INGRESS_HOST:-}"
INGRESS_VARIANT="${CCH_INGRESS_VARIANT:-}"

# Manifest 目录查找顺序
resolve_deploy_dir() {
    if [[ -n "${CCH_DEPLOY_DIR:-}" ]] && [[ -d "$CCH_DEPLOY_DIR/k8s" ]]; then
        echo "$CCH_DEPLOY_DIR"; return
    fi
    local cfg_default="${XDG_CONFIG_HOME:-$HOME/.config}/cch"
    if [[ -d "$cfg_default/k8s" ]]; then echo "$cfg_default"; return; fi
    if [[ -d "/opt/claude-code-hub/k8s" ]]; then echo "/opt/claude-code-hub"; return; fi
    if [[ -d "$REPO_ROOT/deploy/k8s" ]];   then echo "$REPO_ROOT"; return; fi
    if [[ -d "$HOME/claude-code-hub-k8s/k8s" ]]; then echo "$HOME/claude-code-hub-k8s"; return; fi
    echo ""
}
DEPLOY_DIR="$(resolve_deploy_dir)"

###############################################################################
# Runtime 探测 (决定是否使用 k3s ctr 等 k3s-only 命令)
###############################################################################
KUBECTL="kubectl"
RUNTIME=""
detect_runtime() {
    if [[ -n "$RUNTIME_OVERRIDE" ]]; then
        RUNTIME="$RUNTIME_OVERRIDE"
        # 即使 override,也要确保 KUBECTL 能真的运行:无 kubectl 时退到 sudo k3s kubectl
        if ! command -v kubectl &>/dev/null; then
            if [[ "$RUNTIME" == "k3s" ]] && command -v k3s &>/dev/null; then
                KUBECTL="sudo k3s kubectl"
            else
                err "RUNTIME_OVERRIDE=$RUNTIME 指定,但本机既无 kubectl 也无 k3s 可用"
                return 1
            fi
        fi
    elif command -v kubectl &>/dev/null && kubectl cluster-info &>/dev/null; then
        if kubectl get nodes -o jsonpath='{.items[*].status.nodeInfo.kubeletVersion}' 2>/dev/null | grep -q 'k3s'; then
            RUNTIME="k3s"
        else
            RUNTIME="kubectl"
        fi
    elif command -v k3s &>/dev/null; then
        KUBECTL="sudo k3s kubectl"
        RUNTIME="k3s"
    else
        err "未检测到 kubectl 或 k3s。请安装 kubectl 并确保 kubeconfig 可用"
        return 1
    fi
}

require_cluster() {
    detect_runtime
    if ! $KUBECTL get ns "$NAMESPACE" &>/dev/null; then
        err "命名空间 $NAMESPACE 不存在。请先运行 deploy-k8s.sh,或检查 CCH_NAMESPACE 配置"
        exit 1
    fi
}

###############################################################################
# Helpers
###############################################################################
# 解析应用访问地址,用于 health check
resolve_access_url() {
    if [[ -n "$INGRESS_HOST" ]]; then
        echo "http://$INGRESS_HOST"
        return
    fi
    # 尝试从 Service 拿 NodePort
    local np node_ip
    np=$($KUBECTL -n "$NAMESPACE" get svc claude-code-hub -o jsonpath='{.spec.ports[0].nodePort}' 2>/dev/null || echo "")
    node_ip=$($KUBECTL get nodes -o jsonpath='{.items[0].status.addresses[?(@.type=="ExternalIP")].address}' 2>/dev/null)
    if [[ -z "$node_ip" ]]; then
        node_ip=$($KUBECTL get nodes -o jsonpath='{.items[0].status.addresses[?(@.type=="InternalIP")].address}' 2>/dev/null)
    fi
    if [[ -n "$np" ]] && [[ -n "$node_ip" ]]; then
        echo "http://${node_ip}:${np}"
        return
    fi
    echo ""
}

# 在集群内通过 exec 做健康检查,避免依赖集群外网络
health_check_in_pod() {
    local pod
    pod=$($KUBECTL -n "$NAMESPACE" get pods -l app=claude-code-hub \
        -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || echo "")
    if [[ -z "$pod" ]]; then
        warn "没有找到 app Pod"
        return 1
    fi
    local status
    status=$($KUBECTL -n "$NAMESPACE" exec "$pod" -- \
        node -e "fetch('http://127.0.0.1:3000/api/health/ready').then(r=>r.json()).then(j=>{console.log(j.status);process.exit(j.components?.database?.status==='up'?0:1)}).catch(()=>process.exit(1))" 2>/dev/null) || true
    if [[ "$status" == "healthy" ]] || [[ "$status" == "degraded" ]]; then
        ok "Health check (in-pod): $status"
        return 0
    fi
    warn "Health check not healthy: ${status:-unknown}"
    return 1
}

wait_for_deployment_rollout() {
    local timeout="$1"
    local stage="$2"
    if ! $KUBECTL -n "$NAMESPACE" rollout status deployment/claude-code-hub --timeout="$timeout"; then
        err "$stage 未在 $timeout 内完成"
        err "排查命令: kubectl -n $NAMESPACE describe deployment/claude-code-hub"
        err "排查命令: kubectl -n $NAMESPACE logs deploy/claude-code-hub --tail=100"
        return 1
    fi
}

restore_update_scaling() {
    local target_replicas="$1"
    local min_replicas="$2"
    if $KUBECTL -n "$NAMESPACE" get hpa claude-code-hub &>/dev/null; then
        $KUBECTL -n "$NAMESPACE" patch hpa claude-code-hub --type merge \
            -p "{\"spec\":{\"minReplicas\":$min_replicas}}" >/dev/null || true
    fi
    $KUBECTL -n "$NAMESPACE" scale deployment/claude-code-hub --replicas="$target_replicas" >/dev/null || true
}

###############################################################################
# Commands
###############################################################################
cmd_update() {
    require_cluster
    local TIMESTAMP
    TIMESTAMP=$(date +%Y%m%d%H%M%S)
    echo -e "${CYAN}=========================================${NC}"
    echo -e "${CYAN}  Claude Code Hub Upgrade - $TIMESTAMP${NC}"
    echo -e "${CYAN}=========================================${NC}"
    echo ""

    # Step 1: Backup
    info "Step 1/6: Backing up database..."
    if cmd_backup; then
        ok "Database backup complete"
    else
        # 非交互场景下 (无 TTY / CCH_NONINTERACTIVE=1) 直接放弃,避免自动化任务卡死
        if [[ "${CCH_NONINTERACTIVE:-0}" == "1" ]] || [[ ! -t 0 ]]; then
            err "Backup 失败,且当前非交互式环境 — 中止升级以保护数据"
            err "请先人工处理 (磁盘、权限、连接数),或设置 CCH_NONINTERACTIVE=0 并在 TTY 下重试"
            exit 1
        fi
        warn "Backup failed, continue without backup?"
        # read -t 60:60 秒内无输入则退出,避免 CI 卡死
        if ! read -t 60 -p "输入 yes 继续 (默认 60s 超时后中止): " answer; then
            echo ""
            err "超时,中止升级"; exit 1
        fi
        if [[ "$answer" != "yes" ]]; then
            err "已中止"; exit 1
        fi
    fi
    echo ""

    # Step 2: (k3s only) pre-pull image
    info "Step 2/6: Preparing image..."
    if [[ "$RUNTIME" == "k3s" ]]; then
        if sudo k3s ctr images pull "$IMAGE" >/dev/null 2>&1; then
            ok "Image pre-pulled via k3s ctr"
        else
            warn "k3s ctr pull 失败,依赖 Always imagePullPolicy"
        fi
    else
        info "Standard k8s: 依赖 imagePullPolicy=Always 在 rollout 时拉取"
    fi
    echo ""

    # Step 3: Scale down to 1 for migration
    local CURRENT_REPLICAS MIN_REPLICAS
    CURRENT_REPLICAS=$($KUBECTL -n "$NAMESPACE" get hpa claude-code-hub -o jsonpath='{.status.currentReplicas}' 2>/dev/null || echo "2")
    MIN_REPLICAS=$($KUBECTL -n "$NAMESPACE" get hpa claude-code-hub -o jsonpath='{.spec.minReplicas}' 2>/dev/null || echo "2")
    [[ -z "$CURRENT_REPLICAS" || "$CURRENT_REPLICAS" == "null" ]] && CURRENT_REPLICAS=2
    [[ -z "$MIN_REPLICAS"     || "$MIN_REPLICAS" == "null"     ]] && MIN_REPLICAS=2

    info "Step 3/6: Scaling down to 1 replica for migration (was $CURRENT_REPLICAS)..."
    if $KUBECTL -n "$NAMESPACE" get hpa claude-code-hub &>/dev/null; then
        $KUBECTL -n "$NAMESPACE" patch hpa claude-code-hub --type merge -p '{"spec":{"minReplicas":1}}' >/dev/null
    fi
    $KUBECTL -n "$NAMESPACE" scale deployment/claude-code-hub --replicas=1 >/dev/null
    if ! wait_for_deployment_rollout 180s "缩容到 1 副本"; then
        restore_update_scaling "$CURRENT_REPLICAS" "$MIN_REPLICAS"
        exit 1
    fi
    ok "Scaled to 1 replica"
    echo ""

    # Step 4: Update image + migrate
    info "Step 4/6: Updating image on single instance (auto-migration)..."
    if [[ "$RUNTIME" == "k3s" ]]; then
        # k3s: 用 digest 固定,避免 tag 相同导致 no-op rollout
        local IMAGE_DIGEST IMAGE_BY_DIGEST
        IMAGE_DIGEST=$(sudo k3s ctr images ls 2>/dev/null | awk -v img="$IMAGE" '$1==img {print $3; exit}')
        if [[ -n "$IMAGE_DIGEST" ]] && [[ "${IMAGE_DIGEST#sha256:}" != "$IMAGE_DIGEST" ]]; then
            IMAGE_BY_DIGEST="${IMAGE%:*}@${IMAGE_DIGEST}"
            info "  digest: $IMAGE_DIGEST"
            $KUBECTL -n "$NAMESPACE" set image deployment/claude-code-hub app="$IMAGE_BY_DIGEST" >/dev/null
        else
            warn "无法解析 digest,回落到 rollout restart"
            $KUBECTL -n "$NAMESPACE" set image deployment/claude-code-hub app="$IMAGE" >/dev/null || true
            $KUBECTL -n "$NAMESPACE" rollout restart deployment/claude-code-hub >/dev/null
        fi
    else
        # 标准 k8s: set image 到目标 tag,触发 rollout;相同 tag 时强制 restart 拿最新 digest
        local CURRENT_IMAGE
        CURRENT_IMAGE=$($KUBECTL -n "$NAMESPACE" get deployment/claude-code-hub \
            -o jsonpath='{.spec.template.spec.containers[0].image}' 2>/dev/null || echo "")
        if [[ "$CURRENT_IMAGE" == "$IMAGE" ]]; then
            info "镜像 tag 未变 ($IMAGE),执行 rollout restart 重新拉取"
            $KUBECTL -n "$NAMESPACE" rollout restart deployment/claude-code-hub >/dev/null
        else
            $KUBECTL -n "$NAMESPACE" set image deployment/claude-code-hub app="$IMAGE" >/dev/null
        fi
    fi
    if ! wait_for_deployment_rollout 600s "镜像更新 rollout"; then
        err "新镜像 rollout 失败,正在回滚..."
        $KUBECTL -n "$NAMESPACE" rollout undo deployment/claude-code-hub >/dev/null || true
        restore_update_scaling "$CURRENT_REPLICAS" "$MIN_REPLICAS"
        wait_for_deployment_rollout 300s "回滚后的 deployment" || true
        exit 1
    fi
    sleep 3

    if health_check_in_pod; then
        ok "Migration + startup OK"
    else
        err "DB 未通过健康检查,正在回滚..."
        $KUBECTL -n "$NAMESPACE" rollout undo deployment/claude-code-hub >/dev/null
        restore_update_scaling "$CURRENT_REPLICAS" "$MIN_REPLICAS"
        exit 1
    fi
    echo ""

    # Step 5: Scale back
    local desired_replicas="$CURRENT_REPLICAS"
    if [[ "$desired_replicas" -lt "$MIN_REPLICAS" ]]; then
        desired_replicas="$MIN_REPLICAS"
    fi
    info "Step 5/6: Scaling back to $desired_replicas replicas..."
    restore_update_scaling "$desired_replicas" "$MIN_REPLICAS"
    if ! wait_for_deployment_rollout 300s "恢复副本"; then
        err "副本恢复失败,当前 deployment 可能仍停留在单副本"
        exit 1
    fi
    ok "Running with $desired_replicas replicas"
    echo ""

    # Step 6: Final health check
    info "Step 6/6: Final health check..."
    sleep 3
    if health_check_in_pod; then
        ok "Upgrade complete"
    else
        warn "Upgrade done but health check failed. Check: cch logs"
    fi
    echo ""
    $KUBECTL -n "$NAMESPACE" get pods -o wide
}

cmd_status() {
    require_cluster
    echo -e "${CYAN}Pods:${NC}"
    $KUBECTL -n "$NAMESPACE" get pods -o wide
    echo ""
    echo -e "${CYAN}HPA:${NC}"
    $KUBECTL -n "$NAMESPACE" get hpa 2>/dev/null || echo "(no HPA)"
    echo ""
    echo -e "${CYAN}Resources (top):${NC}"
    $KUBECTL -n "$NAMESPACE" top pods 2>/dev/null || warn "metrics-server 未就绪,跳过 top"
}

cmd_logs() {
    require_cluster
    local TAIL="100"
    if [[ "${1:-}" =~ ^[0-9]+$ ]]; then
        TAIL="$1"
        shift
    fi
    $KUBECTL -n "$NAMESPACE" logs deploy/claude-code-hub --all-containers --tail="$TAIL" "$@"
}

cmd_follow() {
    require_cluster
    $KUBECTL -n "$NAMESPACE" logs -f deploy/claude-code-hub --all-containers --tail=50
}

cmd_restart() {
    require_cluster
    info "Rolling restart..."
    $KUBECTL -n "$NAMESPACE" rollout restart deployment/claude-code-hub
    $KUBECTL -n "$NAMESPACE" rollout status  deployment/claude-code-hub --timeout=300s
    ok "Restart complete"
}

cmd_rollback() {
    require_cluster
    warn "Rolling back to previous revision..."
    $KUBECTL -n "$NAMESPACE" rollout undo    deployment/claude-code-hub
    $KUBECTL -n "$NAMESPACE" rollout status  deployment/claude-code-hub --timeout=300s
    ok "Rollback complete"
}

cmd_scale() {
    require_cluster
    local N="${1:-}"
    if [[ -z "$N" ]]; then err "Usage: cch scale <replicas>"; exit 1; fi
    if ! [[ "$N" =~ ^[0-9]+$ ]] || [[ "$N" -lt 1 ]]; then
        err "replicas 必须是正整数: $N"; exit 1
    fi
    if $KUBECTL -n "$NAMESPACE" get hpa claude-code-hub &>/dev/null; then
        local hpa_min hpa_max
        hpa_min=$($KUBECTL -n "$NAMESPACE" get hpa claude-code-hub -o jsonpath='{.spec.minReplicas}' 2>/dev/null || echo "")
        hpa_max=$($KUBECTL -n "$NAMESPACE" get hpa claude-code-hub -o jsonpath='{.spec.maxReplicas}' 2>/dev/null || echo "")
        [[ -z "$hpa_min" || "$hpa_min" == "null" ]] && hpa_min=1
        [[ -z "$hpa_max" || "$hpa_max" == "null" ]] && hpa_max=0

        if [[ "$N" -lt "$hpa_min" ]]; then
            err "HPA minReplicas=$hpa_min 阻止缩到 $N。请先调整 HPA 或重新运行 deploy-k8s.sh 传入匹配的 --hpa-min"
            exit 1
        fi
        if [[ "$hpa_max" -gt 0 ]] && [[ "$N" -gt "$hpa_max" ]]; then
            err "HPA maxReplicas=$hpa_max 阻止扩到 $N。请先调整 HPA 或重新运行 deploy-k8s.sh 传入匹配的 --hpa-max"
            exit 1
        fi
    fi
    $KUBECTL -n "$NAMESPACE" scale deployment/claude-code-hub --replicas="$N"
    info "Scaled to $N replicas"
}

cmd_backup() {
    require_cluster
    local backup_dir="${CCH_BACKUP_DIR:-$HOME/backups/claude-code-hub}"
    mkdir -p "$backup_dir"
    local ts file
    ts=$(date +%Y%m%d_%H%M%S)
    file="$backup_dir/claude_code_hub_${ts}.sql.gz"
    info "Backing up PostgreSQL -> $file"
    if ! $KUBECTL -n "$NAMESPACE" exec sts/postgres -- \
           pg_dump -U claude_code_hub -d claude_code_hub --no-owner --no-privileges \
           | gzip > "$file"; then
        err "备份失败"; rm -f "$file"; return 1
    fi
    local size
    size=$(du -h "$file" 2>/dev/null | cut -f1)
    ok "Backup complete: $file ($size)"
    # 保留最近 30 份 (BSD xargs 无 -r,用 while read 代替)
    local keep="${CCH_BACKUP_KEEP:-30}"
    if ! [[ "$keep" =~ ^[0-9]+$ ]] || [[ "$keep" -lt 1 ]]; then
        warn "CCH_BACKUP_KEEP 必须是正整数,当前值: $keep; 使用默认值 30"
        keep=30
    fi
    # shellcheck disable=SC2012
    ls -t "$backup_dir"/claude_code_hub_*.sql.gz 2>/dev/null \
        | tail -n +"$((keep+1))" \
        | while IFS= read -r old; do rm -f "$old"; done
}

cmd_env() {
    require_cluster
    if command -v python3 &>/dev/null; then
        $KUBECTL -n "$NAMESPACE" get deployment claude-code-hub \
            -o jsonpath='{.spec.template.spec.containers[0].env}' | python3 -m json.tool
    else
        $KUBECTL -n "$NAMESPACE" get deployment claude-code-hub \
            -o jsonpath='{.spec.template.spec.containers[0].env}'
        echo ""
    fi
}

cmd_secret() {
    require_cluster
    local KEY="${1:-admin-token}"
    $KUBECTL -n "$NAMESPACE" get secret claude-code-hub-secrets \
        -o jsonpath="{.data.$KEY}" | b64d
    echo ""
}

cmd_shell() {
    require_cluster
    $KUBECTL -n "$NAMESPACE" exec -it deploy/claude-code-hub -- sh
}

cmd_dbshell() {
    require_cluster
    $KUBECTL -n "$NAMESPACE" exec -it sts/postgres -- \
        psql -U claude_code_hub -d claude_code_hub
}

cmd_info() {
    require_cluster
    local url token img digest
    url=$(resolve_access_url)
    token=$($KUBECTL -n "$NAMESPACE" get secret claude-code-hub-secrets \
        -o jsonpath='{.data.admin-token}' 2>/dev/null | b64d)
    img=$($KUBECTL -n "$NAMESPACE" get deployment claude-code-hub \
        -o jsonpath='{.spec.template.spec.containers[0].image}' 2>/dev/null)
    digest=$($KUBECTL -n "$NAMESPACE" get pods -l app=claude-code-hub \
        -o jsonpath='{.items[0].status.containerStatuses[0].imageID}' 2>/dev/null)

    echo -e "${CYAN}Namespace:${NC}      $NAMESPACE"
    echo -e "${CYAN}Runtime:${NC}        $RUNTIME"
    echo -e "${CYAN}Image (desired):${NC} $img"
    echo -e "${CYAN}Image (running):${NC} $digest"
    if [[ -n "$url" ]]; then
        echo -e "${CYAN}Access URL:${NC}     $url"
    else
        echo -e "${CYAN}Access URL:${NC}     (no ingress/nodeport detected; use 'kubectl port-forward')"
    fi
    echo -e "${CYAN}Admin token:${NC}    ${YELLOW}${token}${NC}"
    echo ""
    $KUBECTL -n "$NAMESPACE" get deployment,statefulset,hpa,svc,ingress 2>/dev/null || true
}

cmd_version() {
    echo "cch v${VERSION}"
    echo "  runtime     : ${RUNTIME:-(not detected)}"
    echo "  namespace   : $NAMESPACE"
    echo "  image       : $IMAGE"
    echo "  deploy-dir  : ${DEPLOY_DIR:-(not found)}"
    echo "  config-file : $CCH_CONFIG_FILE"
}

cmd_doctor() {
    echo -e "${CYAN}cch doctor${NC}"
    local pass=0 fail=0 warn_n=0

    check_pass() { ok "$1"; pass=$((pass+1)); }
    check_warn() { warn "$1"; warn_n=$((warn_n+1)); }
    check_fail() { err "$1"; fail=$((fail+1)); }

    if detect_runtime; then
        check_pass "Runtime detected (runtime=$RUNTIME, kubectl=$KUBECTL)"
    else
        check_warn "运行时探测失败,将继续尝试默认 kubectl 诊断"
    fi

    # kubectl
    if command -v kubectl &>/dev/null; then
        check_pass "kubectl installed: $(kubectl version --client -o yaml 2>/dev/null | awk '/gitVersion/{print $2; exit}')"
    elif [[ "$RUNTIME" == "k3s" ]] && command -v k3s &>/dev/null; then
        check_warn "kubectl 未安装,将使用 sudo k3s kubectl"
    else
        check_fail "kubectl 未安装"
    fi

    # Cluster reachable
    if $KUBECTL cluster-info &>/dev/null; then
        check_pass "Cluster reachable (runtime=$RUNTIME)"
    else
        check_fail "无法连接集群。请检查 kubeconfig / context"
        echo "Summary: $pass passed, $warn_n warnings, $fail failures"; return
    fi

    # Namespace
    if $KUBECTL get ns "$NAMESPACE" &>/dev/null; then
        check_pass "Namespace $NAMESPACE exists"
    else
        check_fail "Namespace $NAMESPACE 不存在 — 请先运行 deploy-k8s.sh"
    fi

    # Secret
    if $KUBECTL -n "$NAMESPACE" get secret claude-code-hub-secrets &>/dev/null; then
        check_pass "Secret claude-code-hub-secrets 存在"
    else
        check_fail "Secret 缺失"
    fi

    # Postgres / Redis / App
    for comp in postgres redis; do
        local rs
        rs=$($KUBECTL -n "$NAMESPACE" get sts "$comp" -o jsonpath='{.status.readyReplicas}' 2>/dev/null || echo "")
        if [[ "$rs" == "1" ]]; then check_pass "$comp StatefulSet ready"
        else check_fail "$comp 未就绪 (ready=$rs)"; fi
    done
    local app_ready
    app_ready=$($KUBECTL -n "$NAMESPACE" get deployment claude-code-hub \
        -o jsonpath='{.status.readyReplicas}' 2>/dev/null || echo "0")
    if [[ "$app_ready" -gt 0 ]]; then
        check_pass "App ready replicas: $app_ready"
    else
        check_fail "App 没有就绪 Pod"
    fi

    # HPA
    if $KUBECTL -n "$NAMESPACE" get hpa claude-code-hub &>/dev/null; then
        check_pass "HPA configured"
    else
        check_warn "HPA 不存在 (非必需)"
    fi

    # Ingress
    if $KUBECTL -n "$NAMESPACE" get ingress claude-code-hub &>/dev/null || \
       $KUBECTL -n "$NAMESPACE" get ingressroute claude-code-hub &>/dev/null 2>&1; then
        check_pass "Ingress resource present"
    else
        check_warn "未检测到 Ingress (如使用 NodePort 可忽略)"
    fi

    # StorageClass
    if [[ "$RUNTIME" == "k3s" ]] && $KUBECTL get sc local-path &>/dev/null; then
        check_pass "StorageClass local-path (k3s default)"
    fi

    # In-pod health
    if health_check_in_pod &>/dev/null; then
        check_pass "In-pod health check"
    else
        check_warn "In-pod health check failed (服务可能启动中)"
    fi

    echo ""
    echo "Summary: ${GREEN}$pass passed${NC}, ${YELLOW}$warn_n warnings${NC}, ${RED}$fail failures${NC}"
}

cmd_install() {
    # 快捷路径:仅当用户在仓库内运行时有用。复制 manifest 到 deploy-dir 并提示跑 deploy-k8s.sh
    if [[ -x "$SCRIPT_DIR/deploy-k8s.sh" ]]; then
        info "转交给 scripts/deploy-k8s.sh (推荐使用完整的安装流程)"
        exec bash "$SCRIPT_DIR/deploy-k8s.sh" "$@"
    fi
    err "scripts/deploy-k8s.sh 未找到。请在仓库内运行或手动调用"
    exit 1
}

cmd_uninstall() {
    detect_runtime
    # 非交互场景必须显式通过 CCH_CONFIRM_UNINSTALL=<namespace> 授权,避免误删
    if [[ ! -t 0 ]]; then
        if [[ "${CCH_CONFIRM_UNINSTALL:-}" != "$NAMESPACE" ]]; then
            err "非交互环境检测到。要 uninstall 必须显式设置:"
            err "  CCH_CONFIRM_UNINSTALL=$NAMESPACE cch uninstall"
            exit 1
        fi
        info "已通过 CCH_CONFIRM_UNINSTALL 授权,继续卸载"
    else
        echo -e "${RED}!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!${NC}"
        echo -e "${RED}  即将删除 namespace: $NAMESPACE${NC}"
        echo -e "${RED}  这会永久删除所有 Pod、PVC(数据库数据)、Secret${NC}"
        echo -e "${RED}!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!${NC}"
        if ! read -t 120 -p "确认?输入 namespace 名称 '$NAMESPACE' 继续: " input; then
            echo ""; info "超时,已取消"; exit 0
        fi
        if [[ "$input" != "$NAMESPACE" ]]; then
            info "已取消"; exit 0
        fi
    fi
    $KUBECTL delete namespace "$NAMESPACE" --timeout=180s
    ok "Namespace $NAMESPACE 已删除"
    info "manifest 目录 ($DEPLOY_DIR) 未删除。如需清理: rm -rf $DEPLOY_DIR"
}

###############################################################################
# Help
###############################################################################
cmd_help() {
    cat <<EOF
cch - Claude Code Hub Kubernetes Management CLI v${VERSION}

Usage: cch <command> [args]

Lifecycle:
  update          Pull latest image, auto-migrate DB, rolling deploy (带回滚)
  restart         Rolling restart (no image change)
  rollback        Rollback to previous deployment revision
  scale <n>       Scale app to n replicas

Install / Teardown:
  install [opts]  调用 scripts/deploy-k8s.sh (透传参数)
  uninstall       Delete namespace + PVCs (破坏性,带二次确认)

Observe:
  status          Show pods, HPA, resource usage
  logs [n] [args] Show last n log lines (default 100), or pass through kubectl log flags
  follow          Tail logs in real-time
  env             Show app environment variables (JSON)
  info            展示访问 URL、Admin Token、镜像 digest
  doctor          诊断 (kubectl / 集群 / 资源 / 健康)
  version         显示版本与当前配置

Data:
  backup          Backup PostgreSQL (gzip, 保留最近 30 份)
  secret [key]    Show secret value (default: admin-token)
  dbshell         Open psql shell
  shell           Open sh in app pod

Config:
  环境变量(或 ~/.config/cch/config):
    CCH_NAMESPACE   K8s namespace (default: claude-code-hub)
    CCH_IMAGE       应用镜像 (default: ghcr.io/ding113/claude-code-hub:latest)
    CCH_DEPLOY_DIR  manifest 目录 (default: 自动查找)
    CCH_RUNTIME     覆盖 runtime: k3s | kubectl
    CCH_INGRESS_HOST Ingress 域名,用于访问 URL 解析
    CCH_BACKUP_DIR  备份目录 (default: ~/backups/claude-code-hub)
    CCH_BACKUP_KEEP 保留数量 (default: 30)
    NO_COLOR        禁用彩色输出

Examples:
  cch status
  cch logs 500
  cch update
  CCH_NAMESPACE=staging cch status
  cch install -y                      # 一键部署
  cch backup
  cch info
  cch doctor
EOF
}

###############################################################################
# Dispatch
###############################################################################
case "${1:-help}" in
    update)    shift; cmd_update "$@" ;;
    status)    shift; cmd_status "$@" ;;
    logs)      shift; cmd_logs "$@" ;;
    follow)    cmd_follow ;;
    restart)   cmd_restart ;;
    rollback)  cmd_rollback ;;
    backup)    cmd_backup ;;
    scale)     shift; cmd_scale "$@" ;;
    env)       cmd_env ;;
    secret)    shift; cmd_secret "$@" ;;
    shell)     cmd_shell ;;
    dbshell)   cmd_dbshell ;;
    info)      cmd_info ;;
    version|--version|-v) cmd_version ;;
    doctor)    cmd_doctor ;;
    install)   shift; cmd_install "$@" ;;
    uninstall) cmd_uninstall ;;
    help|--help|-h|*) cmd_help ;;
esac