LokiStack/frontend/src/api/logs/index.ts

import { http, buildQuery } from '../http'
import type { PageData } from '@/types'

export interface LogSourceAccess { userId: number; labels: Record<string, string> }
export interface LogPolicyGroup { id: number; name: string; sourceId: number; labels: Record<string, string>; userIds: number[]; createdAt?: string; updatedAt?: string }
export interface LogSource {
  id: number; name: string; targetType: 'url' | 'cluster-svc'; clusterID?: number
  namespace: string; service: string; port: number; url: string
  authType: string; basicUser: string; hasBasicPass?: boolean
  headers?: Record<string, string>; headerNames?: string[]; skipTLSVerify?: boolean; hasBearerToken?: boolean
  isDefault: boolean; managed?: boolean; access?: LogSourceAccess[]
}
export interface LogEntry { timestamp: string; timestampNs: string; level: string; message: string; labels: Record<string, string> }
export interface HistogramBucket { timestamp: string; count: number }
export interface LogResult { entries: LogEntry[]; hasMore: boolean; nextCursor: string }
export interface LogQuery { sourceId: number; query: string; start: number; end: number; limit?: number; direction?: 'backward' | 'forward'; cursor?: string }
/** 指标序列采样点(日志分析页) */
export interface MetricPoint { timestamp: string; value: number }
/** 一条按标签分组的指标序列(如 sum by (app)(...) 中每个 app 一条) */
export interface MetricSeries { labels: Record<string, string>; points: MetricPoint[] }
export interface MetricQuery { sourceId: number; query: string; start: number; end: number }
/** 每日日志量（平台留存历史，不受 Loki retention 限制） */
export interface DailyVolumePoint {
  date: string
  bytes: number
  entries: number
  /** 当天未结束或抓取时数据不全 */
  partial: boolean
  /** 该天平台没有留存快照（超过留存期/当天平台没运行）——与"日志量真的是 0"不同 */
  missing?: boolean
}
/**
 * label 候选值的前端缓存（60 秒 + 同请求合并）。
 *
 * 为什么前端也要缓存：快捷 LogQL 的下拉在「切数据源 / 切命名空间 / 重开面板」
 * 时都会重新走一遍请求链，后端虽然只要 25~40ms，但串起来仍是三四次往返，
 * 用户看到的就是转圈。缓存后同一组合在 30 秒内直接命中内存，体感是"打开就有"。
 *
 * 时效性说明：新增 app / 新副本最长 30 秒后出现在下拉里。平台侧的变更
 * （采集器增删改、部署删除、数据源增删改）会由后端立即清缓存，
 * 所以"刚配好的 app 找不到"这类问题不会被缓存放大。
 *
 * 用 Promise 作为缓存值：并发的相同请求共享同一个飞行中的请求（去重），
 * 失败时立刻从缓存剔除，避免把错误结果缓存住。
 */
// 与后端 TTL 对齐（30 秒）：新上线的 app 要在半分钟内可被选中，
// 同时又能挡住"切命名空间/重开面板"造成的重复请求。
const LABEL_CACHE_TTL = 30_000
const labelValueCache = new Map<string, { expires: number; data: Promise<unknown> }>()

function cachedRequest<T>(key: string, run: () => Promise<T>): Promise<T> {
  const now = Date.now()
  const hit = labelValueCache.get(key)
  if (hit && hit.expires > now) return hit.data as Promise<T>
  const promise = run().catch((err: unknown) => {
    labelValueCache.delete(key)
    throw err
  })
  labelValueCache.set(key, { expires: now + LABEL_CACHE_TTL, data: promise })
  return promise
}

/** 手动失效（数据源被删除/重建后调用，避免拿到旧源的候选值） */
export function invalidateLabelCache(): void {
  labelValueCache.clear()
}

export const logsApi = {
  sources: () => http.get<PageData<LogSource>>('/logs/sources'),
  createSource: (source: Partial<LogSource> & { basicPass?: string; bearerToken?: string }) => http.post<LogSource>('/logs/sources', source),
  updateSource: (id: number, data: Partial<LogSource>) => http.put(`/logs/sources/${id}`, data),
  deleteSource: (id: number) => http.delete(`/logs/sources/${id}`),
  testSource: (data: Partial<LogSource>) => http.post<{ success: boolean; message: string }>('/logs/sources/test', data),
  // 查询/统计都支持外部 signal 取消：用户改条件时必须断开旧请求，
  // 否则旧请求会在后端继续占用 Loki CPU（历史上把实例拖到 OOM/不 Ready）。
  query: (data: LogQuery, opts?: { signal?: AbortSignal }) =>
    http.post<LogResult>('/logs/query', data, { timeout: 60000, ...opts }),
  /** 日志上下文：before/after 为前后各取几条（后端默认 2、上限 50，原来写死 5） */
  context: (data: { sourceId: number; timestampNs: string; labels: Record<string, string>; before?: number; after?: number }) =>
    http.post<{ before: LogEntry[]; after: LogEntry[] }>('/logs/context', data),
  /** 总量统计（Loki index/stats，只读索引）：用于计算总页数 */
  stats: (data: LogQuery, opts?: { signal?: AbortSignal }) =>
    http.post<{ streams: number; chunks: number; bytes: number; entries: number }>(
      '/logs/stats', data, { timeout: 25000, ...opts }),
  /** 日志量统计：后端有 7 天范围上限与 30s 超时，前端超时略大于它 */
  /** 应用排行（长范围读平台预聚合快照，毫秒级，不扫日志内容） */
  appRanking: (sourceId: number, from: number, to: number, limit = 10, opts?: { signal?: AbortSignal; sort?: 'bytes' | 'entries' }) => {
    const { signal, sort = 'bytes' } = opts ?? {}
    return http.get<{
      list: Array<{ app: string; namespace: string; entries: number; bytes: number; fromHistory: boolean }>
      /** 快照覆盖的小时数 */
      hours: number
      /** 其中带体积的小时数（体积列是本轮新增，早期快照只有条数） */
      bytesHours: number
    }>(
      `/logs/app-ranking?sourceId=${sourceId}&from=${from}&to=${to}&limit=${limit}&sort=${sort}`,
      signal ? { signal } : undefined,
    )
  },

  /** 每日日志量历史（读平台库，毫秒级） */
  dailyVolume: (sourceId: number, days: number, opts?: { signal?: AbortSignal }) =>
    http.get<DailyVolumePoint[]>(`/logs/daily-volume?sourceId=${sourceId}&days=${days}`, opts),
  histogram: (data: LogQuery, opts?: { signal?: AbortSignal }) =>
    http.post<Array<Record<string, unknown>>>('/logs/histogram', data, { timeout: 40000, ...opts }),
  /** LogQL 指标查询(admin):query 中的 [$step] 占位符由后端按时间跨度替换 */
  metrics: (data: MetricQuery, opts?: { signal?: AbortSignal }) => http.post<MetricSeries[]>('/logs/metrics', data, { timeout: 125000, ...opts }),
  labels: (sourceId: number, start?: number, end?: number) => http.get<string[]>(`/logs/labels${buildQuery({ sourceId, start, end } as { page?: number; pageSize?: number })}`),
  labelValues: (sourceId: number, name: string, query = '', start?: number, end?: number) =>
    cachedRequest<string[]>(
      `lv|${sourceId}|${name}|${query}|${start ?? 0}|${end ?? 0}`,
      () => http.get<string[]>(`/logs/label/${encodeURIComponent(name)}/values${buildQuery({ sourceId, query, start, end } as { page?: number; pageSize?: number })}`),
    ),
  /**
   * 批量取多个 label 的候选值：`/logs/label-values?names=app,service_name,container&query={...}`
   *
   * 快捷查询的 app 下拉要做 app → service_name → container 三级回退，
   * 逐个请求要 3 个往返（每个都经 K8s API 代理到 Loki 索引）；
   * 批量接口后端内部并发，一次往返拿全，前端只看"有没有值"。
   * 返回里缺某个 label 或取失败 = 空数组（不会抛错）。
   */
  labelValuesBatch: (sourceId: number, names: string[], query = '', start?: number, end?: number, opts?: { signal?: AbortSignal }) =>
    // 带 signal 的调用（用户主动取消）不参与缓存，避免把"被取消"的请求共享给别人
    opts?.signal
      ? http.get<Record<string, string[]>>(
          `/logs/label-values${buildQuery({ sourceId, names: names.join(','), query, start, end } as { page?: number; pageSize?: number })}`,
          opts,
        )
      : cachedRequest<Record<string, string[]>>(
          `lvb|${sourceId}|${names.join(',')}|${query}|${start ?? 0}|${end ?? 0}`,
          () => http.get<Record<string, string[]>>(
            `/logs/label-values${buildQuery({ sourceId, names: names.join(','), query, start, end } as { page?: number; pageSize?: number })}`,
          ),
        ),
  /** 当前用户在某数据源上的授权范围（策略 labels；管理员 unrestricted=true） */
  myScope: (sourceId: number) => http.get<{ labels: Record<string, string>; unrestricted: boolean }>(`/logs/my-scope?sourceId=${sourceId}`),
  policyGroups: () => http.get<LogPolicyGroup[]>('/logs/policy-groups'),
  createPolicyGroup: (data: Omit<LogPolicyGroup, 'id'>) => http.post<LogPolicyGroup>('/logs/policy-groups', data),
  updatePolicyGroup: (id: number, data: Omit<LogPolicyGroup, 'id'>) => http.put<LogPolicyGroup>(`/logs/policy-groups/${id}`, data),
  deletePolicyGroup: (id: number) => http.delete(`/logs/policy-groups/${id}`),
}