veza/veza-backend-api/cmd/tools/seed/seed_analytics.go

package main

import (
	"database/sql"
	"fmt"
	"time"
)

// SeedAnalytics creates track_plays, playback_history, daily_track_stats,
// geographic_play_stats, and analytics_events.
func SeedAnalytics(db *sql.DB, cfg Config, users []SeededUser, tracks []SeededTrack) error {
	fmt.Println("\n═══ ANALYTICS ═══")

	if len(tracks) == 0 || len(users) == 0 {
		return nil
	}

	// ── 1. Track plays (the big one) ─────────────────────────────────────────
	p := NewProgress("track_plays", cfg.PlayEvents)
	playRows := make([][]interface{}, 0, cfg.PlayEvents)

	monthsBack := cfg.AnalyticsMonths
	startDate := time.Now().AddDate(0, -monthsBack, 0)

	for i := 0; i < cfg.PlayEvents; i++ {
		user := users[rng.Intn(len(users))]
		// Power-law: popular tracks get way more plays
		track := tracks[PowerLaw(0, len(tracks)-1, 1.5)]

		playedAt := RandomTimeBetween(startDate, time.Now())
		playedAt = RealisticHour(playedAt)

		// Duration: 30-100% of track (most listen to 60%+)
		pctListened := randInt(30, 100)
		duration := track.Duration * pctListened / 100
		if duration < 1 {
			duration = 1
		}

		playRows = append(playRows, []interface{}{
			newUUID(), track.ID, user.ID, duration,
			playedAt, nil, nil, GenUserAgent(),
			GenSource(), GenCountry(),
			playedAt, playedAt, nil,
		})
	}

	_, err := BulkInsert(db, "track_plays",
		"id, track_id, user_id, duration, played_at, device, ip_address, user_agent, source, country_code, created_at, updated_at, deleted_at",
		playRows)
	if err != nil {
		return fmt.Errorf("insert track_plays: %w", err)
	}
	p.Update(cfg.PlayEvents)
	p.Done()

	// ── 2. Playback history ──────────────────────────────────────────────────
	historyCount := cfg.PlayEvents / 5 // 20% of plays recorded in history
	p = NewProgress("playback_history", historyCount)
	historyRows := make([][]interface{}, 0, historyCount)

	for i := 0; i < historyCount; i++ {
		user := users[rng.Intn(len(users))]
		track := tracks[rng.Intn(len(tracks))]
		playedAt := RandomTimeBetween(startDate, time.Now())
		duration := randInt(30, track.Duration)
		completion := 0
		if track.Duration > 0 {
			completion = duration * 100 / track.Duration
		}
		if completion > 100 {
			completion = 100
		}
		source := GenSource()
		device := pick([]string{"desktop", "mobile", "tablet"})

		historyRows = append(historyRows, []interface{}{
			newUUID(), user.ID, track.ID,
			duration, completion, source, nil, device, playedAt,
		})
	}

	_, err = BulkInsert(db, "playback_history",
		"id, user_id, track_id, played_duration, completion_percentage, source, source_id, device_type, played_at",
		historyRows)
	if err != nil {
		return fmt.Errorf("insert playback_history: %w", err)
	}
	p.Update(historyCount)
	p.Done()

	// ── 3. Daily track stats ─────────────────────────────────────────────────
	// Top 200 tracks, 180 days of stats
	topTrackCount := len(tracks) / 5
	if topTrackCount > 200 {
		topTrackCount = 200
	}
	days := monthsBack * 30
	statsCount := topTrackCount * days
	p = NewProgress("daily_track_stats", statsCount)
	statsRows := make([][]interface{}, 0, statsCount)

	for ti := 0; ti < topTrackCount; ti++ {
		track := tracks[ti]
		for d := 0; d < days; d++ {
			date := time.Now().AddDate(0, 0, -d).Format("2006-01-02")
			// Growth curve: more plays for newer days
			dayMultiplier := float64(days-d) / float64(days)
			basePlays := PowerLaw(0, 50, 1.0)
			totalPlays := int(float64(basePlays) * (0.5 + dayMultiplier))
			if totalPlays < 0 {
				totalPlays = 0
			}
			uniqueListeners := totalPlays * randInt(50, 90) / 100
			if uniqueListeners < 0 {
				uniqueListeners = 0
			}
			completeListens := uniqueListeners * randInt(40, 80) / 100
			totalPlayTime := totalPlays * track.Duration * randInt(60, 100) / 100
			avgCompletion := randFloat(0.4, 0.95)

			statsRows = append(statsRows, []interface{}{
				track.ID, date, totalPlays, uniqueListeners,
				completeListens, totalPlayTime, avgCompletion,
			})
		}
	}

	_, err = BulkInsertRaw(db, "daily_track_stats",
		"track_id, date, total_plays, unique_listeners, complete_listens, total_play_time, avg_completion_rate",
		statsRows, "ON CONFLICT (track_id, date) DO NOTHING")
	if err != nil {
		return fmt.Errorf("insert daily_track_stats: %w", err)
	}
	p.Update(len(statsRows))
	p.Done()

	// ── 4. Geographic play stats ─────────────────────────────────────────────
	geoCount := topTrackCount * 10 // ~10 countries per top track
	p = NewProgress("geographic_play_stats", geoCount)
	geoRows := make([][]interface{}, 0, geoCount)

	for ti := 0; ti < topTrackCount; ti++ {
		track := tracks[ti]
		countries := pickN(countryCodes, randInt(3, 10))
		for _, cc := range countries {
			date := time.Now().AddDate(0, 0, -randInt(0, days)).Format("2006-01-02")
			geoRows = append(geoRows, []interface{}{
				newUUID(), track.ID, cc, "", date,
				int64(randInt(1, 5000)), int64(randInt(1, 3000)),
				time.Now(), time.Now(),
			})
		}
	}

	_, _ = BulkInsert(db, "geographic_play_stats",
		"id, track_id, country_code, region, date, play_count, unique_listeners, created_at, updated_at",
		geoRows)
	p.Update(len(geoRows))
	p.Done()

	// ── 5. Analytics events ──────────────────────────────────────────────────
	eventCount := cfg.PlayEvents / 4
	p = NewProgress("analytics_events", eventCount)
	eventRows := make([][]interface{}, 0, eventCount)
	eventTypes := []string{
		"page_view", "track_play", "search", "playlist_create",
		"follow", "signup", "login", "track_upload", "profile_view",
		"playlist_view", "marketplace_view", "product_view",
	}

	for i := 0; i < eventCount; i++ {
		user := users[rng.Intn(len(users))]
		evt := pick(eventTypes)
		createdAt := RandomTimeBetween(startDate, time.Now())
		createdAt = RealisticHour(createdAt)
		payload := fmt.Sprintf(`{"source":"%s","page":"/dashboard","session_id":"%s"}`, GenSource(), newUUID()[:8])

		eventRows = append(eventRows, []interface{}{
			newUUID(), evt, user.ID, payload, createdAt,
		})
	}

	_, err = BulkInsert(db, "analytics_events",
		"id, event_name, user_id, payload, created_at",
		eventRows)
	if err != nil {
		return fmt.Errorf("insert analytics_events: %w", err)
	}
	p.Update(eventCount)
	p.Done()

	// ── 6. Update track play/like counts ─────────────────────────────────────
	p = NewProgress("update track counts", 1)
	_, _ = db.Exec("UPDATE tracks SET play_count = (SELECT COUNT(*) FROM track_plays WHERE track_plays.track_id = tracks.id)")
	_, _ = db.Exec("UPDATE tracks SET like_count = (SELECT COUNT(*) FROM track_likes WHERE track_likes.track_id = tracks.id)")
	_, _ = db.Exec("UPDATE tracks SET comment_count = (SELECT COUNT(*) FROM track_comments WHERE track_comments.track_id = tracks.id)")
	p.Update(1)
	p.Done()

	// ── 7. Update user profile counts ────────────────────────────────────────
	p = NewProgress("update profile counts", 1)
	_, _ = db.Exec("UPDATE user_profiles SET follower_count = (SELECT COUNT(*) FROM follows WHERE follows.followed_id = user_profiles.user_id)")
	_, _ = db.Exec("UPDATE user_profiles SET following_count = (SELECT COUNT(*) FROM follows WHERE follows.follower_id = user_profiles.user_id)")
	_, _ = db.Exec("UPDATE user_profiles SET track_count = (SELECT COUNT(*) FROM tracks WHERE tracks.creator_id = user_profiles.user_id AND tracks.deleted_at IS NULL)")
	_, _ = db.Exec("UPDATE user_profiles SET playlist_count = (SELECT COUNT(*) FROM playlists WHERE playlists.user_id = user_profiles.user_id AND playlists.deleted_at IS NULL)")
	p.Update(1)
	p.Done()

	return nil
}