steamcache2/vfs/predictive/predictive.go

package predictive

import (
	"context"
	"sync"
	"sync/atomic"
	"time"
)

// PredictiveCacheManager implements predictive caching strategies
type PredictiveCacheManager struct {
	accessPredictor *AccessPredictor
	cacheWarmer     *CacheWarmer
	prefetchQueue   chan PrefetchRequest
	ctx             context.Context
	cancel          context.CancelFunc
	wg              sync.WaitGroup
	stats           *PredictiveStats
}

// PrefetchRequest represents a request to prefetch content
type PrefetchRequest struct {
	Key         string
	Priority    int
	Reason      string
	RequestedAt time.Time
}

// PredictiveStats tracks predictive caching statistics
type PredictiveStats struct {
	PrefetchHits     int64
	PrefetchMisses   int64
	PrefetchRequests int64
	CacheWarmHits    int64
	CacheWarmMisses  int64
	mu               sync.RWMutex
}

// AccessPredictor predicts which files are likely to be accessed next
type AccessPredictor struct {
	accessHistory map[string]*AccessSequence
	patterns      map[string][]string // Key -> likely next keys
	mu            sync.RWMutex
}

// AccessSequence tracks access sequences for prediction
type AccessSequence struct {
	Key       string
	NextKeys  []string
	Frequency map[string]int64
	LastSeen  time.Time
	mu        sync.RWMutex
}

// CacheWarmer preloads popular content into cache
type CacheWarmer struct {
	popularContent map[string]*PopularContent
	warmerQueue    chan WarmRequest
	mu             sync.RWMutex
}

// PopularContent tracks popular content for warming
type PopularContent struct {
	Key         string
	AccessCount int64
	LastAccess  time.Time
	Size        int64
	Priority    int
}

// WarmRequest represents a cache warming request
type WarmRequest struct {
	Key         string
	Priority    int
	Reason      string
	Size        int64
	RequestedAt time.Time
	Source      string // Where the warming request came from
}

// ActiveWarmer tracks an active warming operation
type ActiveWarmer struct {
	Key       string
	StartTime time.Time
	Priority  int
	Reason    string
	mu        sync.RWMutex
}

// WarmingStats tracks cache warming statistics
type WarmingStats struct {
	WarmRequests      int64
	WarmSuccesses     int64
	WarmFailures      int64
	WarmBytes         int64
	WarmDuration      time.Duration
	PrefetchRequests  int64
	PrefetchSuccesses int64
	PrefetchFailures  int64
	PrefetchBytes     int64
	PrefetchDuration  time.Duration
}

// NewPredictiveCacheManager creates a new predictive cache manager
func NewPredictiveCacheManager() *PredictiveCacheManager {
	ctx, cancel := context.WithCancel(context.Background())

	pcm := &PredictiveCacheManager{
		accessPredictor: NewAccessPredictor(),
		cacheWarmer:     NewCacheWarmer(),
		prefetchQueue:   make(chan PrefetchRequest, 1000),
		ctx:             ctx,
		cancel:          cancel,
		stats:           &PredictiveStats{},
	}

	// Start background workers
	pcm.wg.Add(1)
	go pcm.prefetchWorker()

	pcm.wg.Add(1)
	go pcm.analysisWorker()

	return pcm
}

// NewAccessPredictor creates a new access predictor
func NewAccessPredictor() *AccessPredictor {
	return &AccessPredictor{
		accessHistory: make(map[string]*AccessSequence),
		patterns:      make(map[string][]string),
	}
}

// NewCacheWarmer creates a new cache warmer
func NewCacheWarmer() *CacheWarmer {
	return &CacheWarmer{
		popularContent: make(map[string]*PopularContent),
		warmerQueue:    make(chan WarmRequest, 100),
	}
}

// NewWarmingStats creates a new warming stats tracker
func NewWarmingStats() *WarmingStats {
	return &WarmingStats{}
}

// NewActiveWarmer creates a new active warmer tracker
func NewActiveWarmer(key string, priority int, reason string) *ActiveWarmer {
	return &ActiveWarmer{
		Key:       key,
		StartTime: time.Now(),
		Priority:  priority,
		Reason:    reason,
	}
}

// RecordAccess records a file access for prediction analysis (lightweight version)
func (pcm *PredictiveCacheManager) RecordAccess(key string, previousKey string, size int64) {
	// Only record if we have a previous key to avoid overhead
	if previousKey != "" {
		pcm.accessPredictor.RecordSequence(previousKey, key)
	}

	// Lightweight popular content tracking - only for large files
	if size > 1024*1024 { // Only track files > 1MB
		pcm.cacheWarmer.RecordAccess(key, size)
	}

	// Skip expensive prediction checks on every access
	// Only check occasionally to reduce overhead
}

// PredictNextAccess predicts the next likely file to be accessed
func (pcm *PredictiveCacheManager) PredictNextAccess(currentKey string) []string {
	return pcm.accessPredictor.PredictNext(currentKey)
}

// RequestPrefetch requests prefetching of predicted content
func (pcm *PredictiveCacheManager) RequestPrefetch(key string, priority int, reason string) {
	select {
	case pcm.prefetchQueue <- PrefetchRequest{
		Key:         key,
		Priority:    priority,
		Reason:      reason,
		RequestedAt: time.Now(),
	}:
		atomic.AddInt64(&pcm.stats.PrefetchRequests, 1)
	default:
		// Queue full, skip prefetch
	}
}

// RecordSequence records an access sequence for prediction
func (ap *AccessPredictor) RecordSequence(previousKey, currentKey string) {
	if previousKey == "" || currentKey == "" {
		return
	}

	ap.mu.Lock()
	defer ap.mu.Unlock()

	seq, exists := ap.accessHistory[previousKey]
	if !exists {
		seq = &AccessSequence{
			Key:       previousKey,
			NextKeys:  []string{},
			Frequency: make(map[string]int64),
			LastSeen:  time.Now(),
		}
		ap.accessHistory[previousKey] = seq
	}

	seq.mu.Lock()
	seq.Frequency[currentKey]++
	seq.LastSeen = time.Now()

	// Update next keys list (keep top 5)
	nextKeys := make([]string, 0, 5)
	for key, _ := range seq.Frequency {
		nextKeys = append(nextKeys, key)
		if len(nextKeys) >= 5 {
			break
		}
	}
	seq.NextKeys = nextKeys
	seq.mu.Unlock()
}

// PredictNext predicts the next likely files to be accessed
func (ap *AccessPredictor) PredictNext(currentKey string) []string {
	ap.mu.RLock()
	defer ap.mu.RUnlock()

	seq, exists := ap.accessHistory[currentKey]
	if !exists {
		return []string{}
	}

	seq.mu.RLock()
	defer seq.mu.RUnlock()

	// Return top predicted keys
	predictions := make([]string, len(seq.NextKeys))
	copy(predictions, seq.NextKeys)
	return predictions
}

// IsPredictedAccess checks if an access was predicted
func (ap *AccessPredictor) IsPredictedAccess(key string) bool {
	ap.mu.RLock()
	defer ap.mu.RUnlock()

	// Check if this key appears in any prediction lists
	for _, seq := range ap.accessHistory {
		seq.mu.RLock()
		for _, predictedKey := range seq.NextKeys {
			if predictedKey == key {
				seq.mu.RUnlock()
				return true
			}
		}
		seq.mu.RUnlock()
	}
	return false
}

// RecordAccess records a file access for cache warming (lightweight version)
func (cw *CacheWarmer) RecordAccess(key string, size int64) {
	// Use read lock first for better performance
	cw.mu.RLock()
	content, exists := cw.popularContent[key]
	cw.mu.RUnlock()

	if !exists {
		// Only acquire write lock when creating new entry
		cw.mu.Lock()
		// Double-check after acquiring write lock
		if content, exists = cw.popularContent[key]; !exists {
			content = &PopularContent{
				Key:         key,
				AccessCount: 1,
				LastAccess:  time.Now(),
				Size:        size,
				Priority:    1,
			}
			cw.popularContent[key] = content
		}
		cw.mu.Unlock()
	} else {
		// Lightweight update - just increment counter
		content.AccessCount++
		content.LastAccess = time.Now()

		// Only update priority occasionally to reduce overhead
		if content.AccessCount%5 == 0 {
			if content.AccessCount > 10 {
				content.Priority = 3
			} else if content.AccessCount > 5 {
				content.Priority = 2
			}
		}
	}
}

// GetPopularContent returns the most popular content for warming
func (cw *CacheWarmer) GetPopularContent(limit int) []*PopularContent {
	cw.mu.RLock()
	defer cw.mu.RUnlock()

	// Sort by access count and return top items
	popular := make([]*PopularContent, 0, len(cw.popularContent))
	for _, content := range cw.popularContent {
		popular = append(popular, content)
	}

	// Simple sort by access count (in production, use proper sorting)
	// For now, just return the first 'limit' items
	if len(popular) > limit {
		popular = popular[:limit]
	}

	return popular
}

// RequestWarming requests warming of a specific key
func (cw *CacheWarmer) RequestWarming(key string, priority int, reason string, size int64) {
	select {
	case cw.warmerQueue <- WarmRequest{
		Key:         key,
		Priority:    priority,
		Reason:      reason,
		Size:        size,
		RequestedAt: time.Now(),
		Source:      "predictive",
	}:
		// Successfully queued
	default:
		// Queue full, skip warming
	}
}

// prefetchWorker processes prefetch requests
func (pcm *PredictiveCacheManager) prefetchWorker() {
	defer pcm.wg.Done()

	for {
		select {
		case <-pcm.ctx.Done():
			return
		case req := <-pcm.prefetchQueue:
			// Process prefetch request
			pcm.processPrefetchRequest(req)
		}
	}
}

// analysisWorker performs periodic analysis and cache warming
func (pcm *PredictiveCacheManager) analysisWorker() {
	defer pcm.wg.Done()

	ticker := time.NewTicker(30 * time.Second) // Analyze every 30 seconds
	defer ticker.Stop()

	for {
		select {
		case <-pcm.ctx.Done():
			return
		case <-ticker.C:
			pcm.performAnalysis()
		}
	}
}

// processPrefetchRequest processes a prefetch request
func (pcm *PredictiveCacheManager) processPrefetchRequest(req PrefetchRequest) {
	// In a real implementation, this would:
	// 1. Check if content is already cached
	// 2. If not, fetch and cache it
	// 3. Update statistics

	// For now, just log the prefetch request
	// In production, integrate with the actual cache system
}

// performAnalysis performs periodic analysis and cache warming
func (pcm *PredictiveCacheManager) performAnalysis() {
	// Get popular content for warming
	popular := pcm.cacheWarmer.GetPopularContent(10)

	// Request warming for popular content
	for _, content := range popular {
		if content.AccessCount > 5 { // Only warm frequently accessed content
			select {
			case pcm.cacheWarmer.warmerQueue <- WarmRequest{
				Key:      content.Key,
				Priority: content.Priority,
				Reason:   "popular_content",
			}:
			default:
				// Queue full, skip
			}
		}
	}
}

// GetStats returns predictive caching statistics
func (pcm *PredictiveCacheManager) GetStats() *PredictiveStats {
	pcm.stats.mu.RLock()
	defer pcm.stats.mu.RUnlock()

	return &PredictiveStats{
		PrefetchHits:     atomic.LoadInt64(&pcm.stats.PrefetchHits),
		PrefetchMisses:   atomic.LoadInt64(&pcm.stats.PrefetchMisses),
		PrefetchRequests: atomic.LoadInt64(&pcm.stats.PrefetchRequests),
		CacheWarmHits:    atomic.LoadInt64(&pcm.stats.CacheWarmHits),
		CacheWarmMisses:  atomic.LoadInt64(&pcm.stats.CacheWarmMisses),
	}
}

// Stop stops the predictive cache manager
func (pcm *PredictiveCacheManager) Stop() {
	pcm.cancel()
	pcm.wg.Wait()
}