jiang13-bbs/backend/service/visitstats_query.go

package service

import (
	"strconv"
	"time"

	"github.com/freefire/jiang13-bbs/model"
	"gorm.io/gorm"
)

// ============ 管理端聚合查询（visit_events SQL 聚合；长趋势用 site_daily_stats） ============

// VisitDayPoint 日趋势点
type VisitDayPoint struct {
	Date string `json:"date"`
	PV   int64  `json:"pv"`
	UV   int64  `json:"uv"`
}

// VisitNamedCount 命名计数（设备/来源/爬虫等）
type VisitNamedCount struct {
	Key   string `json:"key"`
	Label string `json:"label"`
	Count int64  `json:"count"`
}

// VisitPageRow 受访页面行
type VisitPageRow struct {
	Path  string  `json:"path"`
	PV    int64   `json:"pv"`
	UV    int64   `json:"uv"`
	Share float64 `json:"share"` // 占区间总 PV 比例（0–100）
}

// VisitOverview 概览 Tab
type VisitOverview struct {
	RangeDays   int               `json:"range_days"`
	Trend       []VisitDayPoint   `json:"trend"`
	TodayPV     int64             `json:"today_pv"`
	TodayUV     int64             `json:"today_uv"`
	YesterdayPV int64             `json:"yesterday_pv"`
	YesterdayUV int64             `json:"yesterday_uv"`
	LoggedInPV  int64             `json:"logged_in_pv"` // 今日登录 PV
	Devices     []VisitNamedCount `json:"devices"`
	Sources     []VisitNamedCount `json:"sources"`
	TopPages    []VisitPageRow    `json:"top_pages"`
}

// VisitMinutePoint 分钟级 PV 点
type VisitMinutePoint struct {
	Minute string `json:"minute"` // YYYY-MM-DD HH24:MI
	PV     int64  `json:"pv"`
}

// VisitEventItem 明细列表行
type VisitEventItem struct {
	ID        uint64    `json:"id"`
	CreatedAt time.Time `json:"created_at"`
	Kind      string    `json:"kind"`
	Path      string    `json:"path"`
	RefHost   string    `json:"ref_host"`
	RefKind   string    `json:"ref_kind"`
	UserID    uint      `json:"user_id"`
	Username  string    `json:"username"`
	IP        string    `json:"ip"`
	Device    string    `json:"device"`
	OS        string    `json:"os"`
	Browser   string    `json:"browser"`
	Bot       string    `json:"bot"`
	Status    int       `json:"status"`
}

// VisitRealtime 实时 Tab
type VisitRealtime struct {
	Minutes  []VisitMinutePoint `json:"minutes"`
	PV5min   int64              `json:"pv_5min"`
	PV30min  int64              `json:"pv_30min"`
	Online   int64              `json:"online"`
	TopPages []VisitNamedCount  `json:"top_pages"`
	Recent   []VisitEventItem   `json:"recent"`
}

// VisitSourcesKindDay 来源按日（kind 分列行）
type VisitSourcesKindDay struct {
	Date  string            `json:"date"`
	Items []VisitNamedCount `json:"items"`
}

// VisitSources 来源分析 Tab
type VisitSources struct {
	RangeDays  int                  `json:"range_days"`
	Kinds      []VisitNamedCount    `json:"kinds"`
	SearchHost []VisitNamedCount    `json:"search_hosts"`
	External   []VisitNamedCount    `json:"external"`
	Trend      []VisitSourcesKindDay `json:"trend"`
}

// VisitPages 受访页面 Tab
type VisitPages struct {
	RangeDays int            `json:"range_days"`
	Total     int64          `json:"total"` // 命中的 path 数
	Items     []VisitPageRow `json:"items"`
}

// VisitVisitors 访客画像 Tab
type VisitVisitors struct {
	RangeDays   int               `json:"range_days"`
	Devices     []VisitNamedCount `json:"devices"`
	OSes        []VisitNamedCount `json:"oses"`
	Browsers    []VisitNamedCount `json:"browsers"`
	LoggedInPV  int64             `json:"logged_in_pv"`
	GuestPV     int64             `json:"guest_pv"`
	NewVisitors int64             `json:"new_visitors"`
	Returning   int64             `json:"returning"`
	Hourly      []VisitNamedCount `json:"hourly"` // 0–23 时段 PV
}

// VisitSecurity 爬虫与安全 Tab
type VisitSecurity struct {
	RangeDays   int               `json:"range_days"`
	Bots        []VisitNamedCount `json:"bots"`
	BotTrend    []VisitDayPoint   `json:"bot_trend"`
	ProbeStatus []VisitNamedCount `json:"probe_status"`
	ProbeIPs    []VisitNamedCount `json:"probe_ips"`
	LoginFailed int64             `json:"login_failed"` // 近 N 日登录失败次数
	FailedUsers int64             `json:"failed_users"` // 涉及用户名数
}

// VisitEventFilter 明细筛选
type VisitEventFilter struct {
	Kind  string
	Bot   string
	Path  string
	IP    string
	Start time.Time
	End   time.Time
	Page  int
	Size  int
}

// VisitEventsResult 明细分页结果
type VisitEventsResult struct {
	Total int64            `json:"total"`
	Items []VisitEventItem `json:"items"`
}

// VisitSettingsPayload 设置 Tab（开关 + 保留天数 + 容量）
type VisitSettingsPayload struct {
	Enabled       bool       `json:"enabled"`
	RetentionDays int        `json:"retention_days"`
	Usage         VisitUsage `json:"usage"`
	Dropped       uint64     `json:"dropped"`
}

type pvUVRow struct {
	PV int64 `gorm:"column:pv"`
	UV int64 `gorm:"column:uv"`
}

func visitRangeStart(days int) time.Time {
	return todayDate().AddDate(0, 0, -(days - 1))
}

// groupCounts 通用 GROUP BY 计数（expr 为聚合表达式），按计数降序取前 limit 条。
// extraSQL 追加到 WHERE（如 " AND ref_kind = 'search'"）。
func (s *VisitStatsService) groupCounts(start time.Time, expr, kind, extraSQL string, limit int) ([]VisitNamedCount, error) {
	var rows []struct {
		Key   string
		Count int64
	}
	q := s.db.Model(&model.VisitEvent{}).
		Select(expr+" AS key, COUNT(*) AS count").
		Where("created_at >= ?", start)
	if kind != "" {
		q = q.Where("kind = ?", kind)
	}
	if extraSQL != "" {
		q = q.Where(extraSQL)
	}
	if err := q.Group("key").Order("count DESC").Limit(limit).Scan(&rows).Error; err != nil {
		return nil, err
	}
	out := make([]VisitNamedCount, 0, len(rows))
	for _, r := range rows {
		out = append(out, VisitNamedCount{Key: r.Key, Label: r.Key, Count: r.Count})
	}
	return out, nil
}

// Overview 概览：趋势（site_daily_stats）+ 今日/昨日 + 设备/来源/热门页面
func (s *VisitStatsService) Overview(rangeQ string) (*VisitOverview, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	out := &VisitOverview{RangeDays: days}

	// 日趋势：日聚合表（每天一行，不随明细清理丢失）
	if err := s.db.Model(&model.SiteDailyStats{}).
		Select("to_char(date, 'YYYY-MM-DD') AS date, pv, uv").
		Where("date >= ?", start).
		Order("date ASC").Scan(&out.Trend).Error; err != nil {
		return nil, err
	}
	// 空结果时 GORM 不赋值，nil 切片会序列化成 JSON null（前端期望 []）
	if out.Trend == nil {
		out.Trend = []VisitDayPoint{}
	}

	var today pvUVRow
	if err := s.db.Model(&model.SiteDailyStats{}).
		Select("COALESCE(SUM(pv),0) AS pv, COALESCE(SUM(uv),0) AS uv").
		Where("date = ?", todayDate()).Scan(&today).Error; err != nil {
		return nil, err
	}
	out.TodayPV, out.TodayUV = today.PV, today.UV

	var yesterday pvUVRow
	if err := s.db.Model(&model.SiteDailyStats{}).
		Select("COALESCE(SUM(pv),0) AS pv, COALESCE(SUM(uv),0) AS uv").
		Where("date = ?", todayDate().AddDate(0, 0, -1)).Scan(&yesterday).Error; err != nil {
		return nil, err
	}
	out.YesterdayPV, out.YesterdayUV = yesterday.PV, yesterday.UV

	if err := s.db.Model(&model.SiteDailyStats{}).
		Select("COALESCE(SUM(logged_in_pv),0) AS logged_in_pv").
		Where("date = ?", todayDate()).Scan(&out.LoggedInPV).Error; err != nil {
		return nil, err
	}

	var err error
	if out.Devices, err = s.groupCounts(start, "device", model.VisitKindPageview, "", 8); err != nil {
		return nil, err
	}
	if out.Sources, err = s.groupCounts(start, "ref_kind", model.VisitKindPageview, "", 8); err != nil {
		return nil, err
	}
	if out.TopPages, err = s.pageAgg(start, "", 1, 10); err != nil {
		return nil, err
	}
	return out, nil
}

// pageAgg path 聚合 PV/UV/占比（q 为 path 模糊匹配；page 从 1 起）
func (s *VisitStatsService) pageAgg(start time.Time, q string, page, size int) ([]VisitPageRow, error) {
	var totalPV int64
	if err := s.db.Model(&model.VisitEvent{}).
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Scopes(pathFilter(q)).
		Count(&totalPV).Error; err != nil {
		return nil, err
	}
	if totalPV == 0 {
		return []VisitPageRow{}, nil
	}

	var rows []VisitPageRow
	if err := s.db.Model(&model.VisitEvent{}).
		Select("path, COUNT(*) AS pv, COUNT(DISTINCT vid_hash) AS uv").
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Scopes(pathFilter(q)).
		Group("path").Order("pv DESC").
		Offset((page - 1) * size).Limit(size).Scan(&rows).Error; err != nil {
		return nil, err
	}
	for i := range rows {
		rows[i].Share = float64(rows[i].PV) * 100 / float64(totalPV)
	}
	return rows, nil
}

func pathFilter(q string) func(*gorm.DB) *gorm.DB {
	if q == "" {
		return func(db *gorm.DB) *gorm.DB { return db }
	}
	return func(db *gorm.DB) *gorm.DB { return db.Where("path ILIKE ?", "%"+q+"%") }
}

// Realtime 实时：近 60 分钟分钟级 PV + 在线 + 近 5 分钟热门路径 + 最新 20 条
func (s *VisitStatsService) Realtime() (*VisitRealtime, error) {
	out := &VisitRealtime{}
	now := time.Now()

	if err := s.db.Model(&model.VisitEvent{}).
		Select("to_char(created_at, 'YYYY-MM-DD HH24:MI') AS minute, COUNT(*) AS pv").
		Where("created_at >= ? AND kind = ?", now.Add(-time.Hour), model.VisitKindPageview).
		Group("minute").Order("minute ASC").Scan(&out.Minutes).Error; err != nil {
		return nil, err
	}
	if out.Minutes == nil {
		out.Minutes = []VisitMinutePoint{}
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Where("created_at >= ? AND kind = ?", now.Add(-5*time.Minute), model.VisitKindPageview).
		Count(&out.PV5min).Error; err != nil {
		return nil, err
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Where("created_at >= ? AND kind = ?", now.Add(-30*time.Minute), model.VisitKindPageview).
		Count(&out.PV30min).Error; err != nil {
		return nil, err
	}
	// 在线口径与首页概览一致：users.last_seen_at 近 5 分钟且未封禁
	if err := s.db.Model(&model.User{}).
		Where("last_seen_at IS NOT NULL AND last_seen_at >= ? AND banned = ?", now.Add(-5*time.Minute), false).
		Count(&out.Online).Error; err != nil {
		return nil, err
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Select("path AS key, COUNT(*) AS count").
		Where("created_at >= ? AND kind = ?", now.Add(-5*time.Minute), model.VisitKindPageview).
		Group("path").Order("count DESC").Limit(10).
		Scan(&out.TopPages).Error; err != nil {
		return nil, err
	}
	for i := range out.TopPages {
		out.TopPages[i].Label = out.TopPages[i].Key
	}
	if out.TopPages == nil {
		out.TopPages = []VisitNamedCount{}
	}

	res, err := s.Events(VisitEventFilter{Kind: model.VisitKindPageview, Page: 1, Size: 20})
	if err != nil {
		return nil, err
	}
	out.Recent = res.Items
	return out, nil
}

// Sources 来源分析
func (s *VisitStatsService) Sources(rangeQ string) (*VisitSources, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	out := &VisitSources{RangeDays: days}
	var err error
	if out.Kinds, err = s.groupCounts(start, "ref_kind", model.VisitKindPageview, "", 8); err != nil {
		return nil, err
	}
	if out.SearchHost, err = s.groupCounts(start, "ref_host", model.VisitKindPageview,
		"ref_kind = '"+RefKindSearch+"'", 20); err != nil {
		return nil, err
	}
	if out.External, err = s.groupCounts(start, "ref_host", model.VisitKindPageview,
		"ref_kind = '"+RefKindExternal+"'", 20); err != nil {
		return nil, err
	}
	if out.Trend, err = s.SourcesTrend(rangeQ); err != nil {
		return nil, err
	}
	return out, nil
}

// SourcesTrend 来源按日明细（kind 分列行）
func (s *VisitStatsService) SourcesTrend(rangeQ string) ([]VisitSourcesKindDay, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	var rows []struct {
		Date    string
		RefKind string
		Count   int64
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Select("to_char(created_at, 'YYYY-MM-DD') AS date, ref_kind, COUNT(*) AS count").
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Group("date, ref_kind").Order("date ASC").Scan(&rows).Error; err != nil {
		return nil, err
	}
	byDay := make(map[string][]VisitNamedCount)
	var dates []string
	seen := make(map[string]struct{})
	for _, r := range rows {
		if _, ok := seen[r.Date]; !ok {
			seen[r.Date] = struct{}{}
			dates = append(dates, r.Date)
		}
		byDay[r.Date] = append(byDay[r.Date], VisitNamedCount{Key: r.RefKind, Label: r.RefKind, Count: r.Count})
	}
	out := make([]VisitSourcesKindDay, 0, len(dates))
	for _, d := range dates {
		out = append(out, VisitSourcesKindDay{Date: d, Items: byDay[d]})
	}
	return out, nil
}

// Pages 受访页面（分页 + 模糊）
func (s *VisitStatsService) Pages(rangeQ, q string, page, size int) (*VisitPages, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	if page < 1 {
		page = 1
	}
	if size < 1 || size > 100 {
		size = 20
	}
	out := &VisitPages{RangeDays: days}
	if err := s.db.Model(&model.VisitEvent{}).
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Scopes(pathFilter(q)).
		Distinct("path").Count(&out.Total).Error; err != nil {
		return nil, err
	}
	items, err := s.pageAgg(start, q, page, size)
	if err != nil {
		return nil, err
	}
	if items == nil {
		items = []VisitPageRow{}
	}
	out.Items = items
	return out, nil
}

// Visitors 访客画像
func (s *VisitStatsService) Visitors(rangeQ string) (*VisitVisitors, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	out := &VisitVisitors{RangeDays: days}
	var err error
	if out.Devices, err = s.groupCounts(start, "device", model.VisitKindPageview, "", 8); err != nil {
		return nil, err
	}
	if out.OSes, err = s.groupCounts(start, "os", model.VisitKindPageview, "", 10); err != nil {
		return nil, err
	}
	if out.Browsers, err = s.groupCounts(start, "browser", model.VisitKindPageview, "", 10); err != nil {
		return nil, err
	}

	var lg pvUVRow
	if err := s.db.Model(&model.VisitEvent{}).
		Select("COALESCE(SUM(CASE WHEN user_id > 0 THEN 1 ELSE 0 END),0) AS pv, COALESCE(SUM(CASE WHEN user_id = 0 THEN 1 ELSE 0 END),0) AS uv").
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Scan(&lg).Error; err != nil {
		return nil, err
	}
	out.LoggedInPV, out.GuestPV = lg.PV, lg.UV

	// 新老访客：区间内 vid 首次出现时间晚于区间起点视为新访客
	var uvRows []struct {
		VidHash string
		FirstAt time.Time `gorm:"column:first_at"`
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Select("vid_hash, MIN(created_at) AS first_at").
		Where("created_at >= ? AND kind = ? AND vid_hash <> ''", start, model.VisitKindPageview).
		Group("vid_hash").Scan(&uvRows).Error; err != nil {
		return nil, err
	}
	for _, r := range uvRows {
		if r.FirstAt.After(start) {
			out.NewVisitors++
		} else {
			out.Returning++
		}
	}

	// 24 小时时段 PV（区间合计）
	var hours []struct {
		Hour  int
		Count int64
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Select("EXTRACT(HOUR FROM created_at)::int AS hour, COUNT(*) AS count").
		Where("created_at >= ? AND kind = ?", start, model.VisitKindPageview).
		Group("hour").Order("hour ASC").Scan(&hours).Error; err != nil {
		return nil, err
	}
	out.Hourly = make([]VisitNamedCount, 24)
	for i := range out.Hourly {
		label := strconv.Itoa(i)
		if i < 10 {
			label = "0" + label
		}
		out.Hourly[i].Key = label
		out.Hourly[i].Label = label + " 时"
	}
	for _, h := range hours {
		if h.Hour >= 0 && h.Hour < 24 {
			out.Hourly[h.Hour].Count = h.Count
		}
	}
	return out, nil
}

// Security 爬虫与安全
func (s *VisitStatsService) Security(rangeQ string) (*VisitSecurity, error) {
	days := parseRangeDays(rangeQ)
	start := visitRangeStart(days)
	out := &VisitSecurity{RangeDays: days}
	var err error
	if out.Bots, err = s.groupCounts(start, "bot", model.VisitKindBot, "", 20); err != nil {
		return nil, err
	}
	if err := s.db.Model(&model.VisitEvent{}).
		Select("to_char(created_at, 'YYYY-MM-DD') AS date, COUNT(*) AS pv").
		Where("created_at >= ? AND kind = ?", start, model.VisitKindBot).
		Group("date").Order("date ASC").Scan(&out.BotTrend).Error; err != nil {
		return nil, err
	}
	if out.BotTrend == nil {
		out.BotTrend = []VisitDayPoint{}
	}
	if out.ProbeStatus, err = s.groupCounts(start, "status::text", model.VisitKindProbe, "", 15); err != nil {
		return nil, err
	}
	if out.ProbeIPs, err = s.groupCounts(start, "ip", model.VisitKindProbe, "", 20); err != nil {
		return nil, err
	}

	if err := s.db.Model(&model.LoginLog{}).
		Where("success = ? AND created_at >= ?", false, start).
		Count(&out.LoginFailed).Error; err != nil {
		return nil, err
	}
	if err := s.db.Raw(`SELECT COUNT(DISTINCT username) FROM login_logs WHERE success = ? AND created_at >= ?`,
		false, start).Scan(&out.FailedUsers).Error; err != nil {
		return nil, err
	}
	return out, nil
}

// Events 访问明细（筛选 + 分页 + 补用户昵称）
func (s *VisitStatsService) Events(f VisitEventFilter) (*VisitEventsResult, error) {
	if f.Page < 1 {
		f.Page = 1
	}
	if f.Size < 1 || f.Size > 200 {
		f.Size = 50
	}
	apply := func(db *gorm.DB) *gorm.DB {
		if f.Kind != "" {
			db = db.Where("kind = ?", f.Kind)
		}
		if f.Bot != "" {
			db = db.Where("bot = ?", f.Bot)
		}
		if f.Path != "" {
			db = db.Where("path ILIKE ?", "%"+f.Path+"%")
		}
		if f.IP != "" {
			db = db.Where("ip = ?", f.IP)
		}
		if !f.Start.IsZero() {
			db = db.Where("created_at >= ?", f.Start)
		}
		if !f.End.IsZero() {
			db = db.Where("created_at < ?", f.End)
		}
		return db
	}

	out := &VisitEventsResult{}
	if err := apply(s.db.Model(&model.VisitEvent{})).Count(&out.Total).Error; err != nil {
		return nil, err
	}
	var events []model.VisitEvent
	if err := apply(s.db.Model(&model.VisitEvent{})).
		Order("id DESC").
		Offset((f.Page - 1) * f.Size).Limit(f.Size).
		Find(&events).Error; err != nil {
		return nil, err
	}

	// 批量补昵称
	userIDs := make([]uint, 0, len(events))
	for _, e := range events {
		if e.UserID > 0 {
			userIDs = append(userIDs, e.UserID)
		}
	}
	nameMap := make(map[uint]string, len(userIDs))
	if len(userIDs) > 0 {
		var users []model.User
		if err := s.db.Select("id, username, nickname").
			Where("id IN ?", userIDs).Find(&users).Error; err != nil {
			return nil, err
		}
		for _, u := range users {
			nameMap[u.ID] = u.Nickname
		}
	}
	out.Items = make([]VisitEventItem, 0, len(events))
	for _, e := range events {
		out.Items = append(out.Items, VisitEventItem{
			ID: e.ID, CreatedAt: e.CreatedAt, Kind: e.Kind, Path: e.Path,
			RefHost: e.RefHost, RefKind: e.RefKind,
			UserID: e.UserID, Username: nameMap[e.UserID],
			IP: e.IP, Device: e.Device, OS: e.OS, Browser: e.Browser,
			Bot: e.Bot, Status: e.Status,
		})
	}
	return out, nil
}

// SettingsPayload 设置 Tab 数据
func (s *VisitStatsService) SettingsPayload() (*VisitSettingsPayload, error) {
	out := &VisitSettingsPayload{
		Enabled:       s.Enabled(),
		RetentionDays: s.retentionDays(),
		Dropped:       s.DroppedCount(),
	}
	u, err := s.Usage()
	if err != nil {
		return nil, err
	}
	out.Usage = u
	return out, nil
}

// SaveSettings 保存采集开关与保留天数
func (s *VisitStatsService) SaveSettings(enabled bool, retentionDays int) error {
	if err := s.setting.SetAnalyticsEnabled(enabled); err != nil {
		return err
	}
	return s.setting.SetAnalyticsRetentionDays(retentionDays)
}