gooseek/backend/internal/discover/personalization.go

package discover

import (
	"context"
	"encoding/json"
	"fmt"
	"math"
	"sort"
	"strings"
	"sync"
	"time"
)

type UserInterests struct {
	UserID           string                 `json:"userId"`
	Topics           map[string]float64     `json:"topics"`
	Sources          map[string]float64     `json:"sources"`
	Keywords         map[string]float64     `json:"keywords"`
	ViewHistory      []ViewEvent            `json:"viewHistory"`
	SavedArticles    []string               `json:"savedArticles"`
	BlockedSources   []string               `json:"blockedSources"`
	BlockedTopics    []string               `json:"blockedTopics"`
	PreferredLang    string                 `json:"preferredLang"`
	Region           string                 `json:"region"`
	ReadingLevel     string                 `json:"readingLevel"`
	Notifications    NotificationPrefs      `json:"notifications"`
	LastUpdated      time.Time              `json:"lastUpdated"`
	CustomCategories []CustomCategory       `json:"customCategories,omitempty"`
}

type ViewEvent struct {
	ArticleID    string    `json:"articleId"`
	URL          string    `json:"url"`
	Topic        string    `json:"topic"`
	Source       string    `json:"source"`
	Keywords     []string  `json:"keywords"`
	TimeSpent    int       `json:"timeSpentSeconds"`
	Completed    bool      `json:"completed"`
	Saved        bool      `json:"saved"`
	Shared       bool      `json:"shared"`
	Timestamp    time.Time `json:"timestamp"`
	Engagement   float64   `json:"engagement"`
}

type NotificationPrefs struct {
	Enabled        bool     `json:"enabled"`
	DailyDigest    bool     `json:"dailyDigest"`
	DigestTime     string   `json:"digestTime"`
	BreakingNews   bool     `json:"breakingNews"`
	TopicAlerts    []string `json:"topicAlerts"`
	Frequency      string   `json:"frequency"`
}

type CustomCategory struct {
	ID          string   `json:"id"`
	Name        string   `json:"name"`
	Keywords    []string `json:"keywords"`
	Sources     []string `json:"sources"`
	Weight      float64  `json:"weight"`
}

type PersonalizedFeed struct {
	UserID     string               `json:"userId"`
	Items      []FeedItem           `json:"items"`
	Categories []FeedCategory       `json:"categories"`
	TrendingIn []string             `json:"trendingIn"`
	UpdatedAt  time.Time            `json:"updatedAt"`
	NextUpdate time.Time            `json:"nextUpdate"`
}

type FeedItem struct {
	ID             string    `json:"id"`
	URL            string    `json:"url"`
	Title          string    `json:"title"`
	Summary        string    `json:"summary"`
	Thumbnail      string    `json:"thumbnail"`
	Source         string    `json:"source"`
	SourceLogo     string    `json:"sourceLogo"`
	Topic          string    `json:"topic"`
	Keywords       []string  `json:"keywords"`
	PublishedAt    time.Time `json:"publishedAt"`
	RelevanceScore float64   `json:"relevanceScore"`
	Reason         string    `json:"reason"`
	SourcesCount   int       `json:"sourcesCount"`
	ReadTime       int       `json:"readTimeMinutes"`
	HasDigest      bool      `json:"hasDigest"`
	IsBreaking     bool      `json:"isBreaking"`
	IsTrending     bool      `json:"isTrending"`
	IsSaved        bool      `json:"isSaved"`
	IsRead         bool      `json:"isRead"`
}

type FeedCategory struct {
	ID       string     `json:"id"`
	Name     string     `json:"name"`
	Icon     string     `json:"icon"`
	Color    string     `json:"color"`
	Items    []FeedItem `json:"items"`
	IsCustom bool       `json:"isCustom"`
}

type PersonalizationEngine struct {
	userStore   UserInterestStore
	contentRepo ContentRepository
	mu          sync.RWMutex
	config      PersonalizationConfig
}

type PersonalizationConfig struct {
	MaxFeedItems       int
	DecayFactor        float64
	RecencyWeight      float64
	EngagementWeight   float64
	TopicMatchWeight   float64
	SourceTrustWeight  float64
	DiversityFactor    float64
	TrendingBoost      float64
	BreakingBoost      float64
}

type UserInterestStore interface {
	Get(ctx context.Context, userID string) (*UserInterests, error)
	Save(ctx context.Context, interests *UserInterests) error
	Delete(ctx context.Context, userID string) error
}

type ContentRepository interface {
	GetLatestContent(ctx context.Context, topics []string, limit int) ([]FeedItem, error)
	GetTrending(ctx context.Context, region string, limit int) ([]FeedItem, error)
	GetByKeywords(ctx context.Context, keywords []string, limit int) ([]FeedItem, error)
}

func DefaultConfig() PersonalizationConfig {
	return PersonalizationConfig{
		MaxFeedItems:       50,
		DecayFactor:        0.95,
		RecencyWeight:      0.25,
		EngagementWeight:   0.20,
		TopicMatchWeight:   0.30,
		SourceTrustWeight:  0.15,
		DiversityFactor:    0.10,
		TrendingBoost:      1.5,
		BreakingBoost:      2.0,
	}
}

func NewPersonalizationEngine(userStore UserInterestStore, contentRepo ContentRepository, cfg PersonalizationConfig) *PersonalizationEngine {
	return &PersonalizationEngine{
		userStore:   userStore,
		contentRepo: contentRepo,
		config:      cfg,
	}
}

func (e *PersonalizationEngine) GenerateForYouFeed(ctx context.Context, userID string) (*PersonalizedFeed, error) {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		interests = &UserInterests{
			UserID:        userID,
			Topics:        make(map[string]float64),
			Sources:       make(map[string]float64),
			Keywords:      make(map[string]float64),
			PreferredLang: "ru",
			Region:        "russia",
		}
	}

	var allItems []FeedItem
	var mu sync.Mutex
	var wg sync.WaitGroup

	topTopics := e.getTopInterests(interests.Topics, 5)
	wg.Add(1)
	go func() {
		defer wg.Done()
		items, _ := e.contentRepo.GetLatestContent(ctx, topTopics, 30)
		mu.Lock()
		allItems = append(allItems, items...)
		mu.Unlock()
	}()

	wg.Add(1)
	go func() {
		defer wg.Done()
		items, _ := e.contentRepo.GetTrending(ctx, interests.Region, 20)
		for i := range items {
			items[i].IsTrending = true
		}
		mu.Lock()
		allItems = append(allItems, items...)
		mu.Unlock()
	}()

	topKeywords := e.getTopKeywords(interests.Keywords, 10)
	if len(topKeywords) > 0 {
		wg.Add(1)
		go func() {
			defer wg.Done()
			items, _ := e.contentRepo.GetByKeywords(ctx, topKeywords, 15)
			mu.Lock()
			allItems = append(allItems, items...)
			mu.Unlock()
		}()
	}

	wg.Wait()

	allItems = e.deduplicateItems(allItems)
	allItems = e.filterBlockedContent(allItems, interests)

	for i := range allItems {
		allItems[i].RelevanceScore = e.calculateRelevance(allItems[i], interests)
		allItems[i].Reason = e.explainRecommendation(allItems[i], interests)
		allItems[i].IsRead = e.isArticleRead(allItems[i].URL, interests)
		allItems[i].IsSaved = e.isArticleSaved(allItems[i].URL, interests)
	}

	sort.Slice(allItems, func(i, j int) bool {
		return allItems[i].RelevanceScore > allItems[j].RelevanceScore
	})

	allItems = e.applyDiversity(allItems)

	if len(allItems) > e.config.MaxFeedItems {
		allItems = allItems[:e.config.MaxFeedItems]
	}

	categories := e.groupByCategory(allItems, interests)

	return &PersonalizedFeed{
		UserID:     userID,
		Items:      allItems,
		Categories: categories,
		TrendingIn: topTopics,
		UpdatedAt:  time.Now(),
		NextUpdate: time.Now().Add(15 * time.Minute),
	}, nil
}

func (e *PersonalizationEngine) RecordView(ctx context.Context, userID string, event ViewEvent) error {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		interests = &UserInterests{
			UserID:   userID,
			Topics:   make(map[string]float64),
			Sources:  make(map[string]float64),
			Keywords: make(map[string]float64),
		}
	}

	event.Engagement = e.calculateEngagement(event)

	interests.ViewHistory = append([]ViewEvent{event}, interests.ViewHistory...)
	if len(interests.ViewHistory) > 500 {
		interests.ViewHistory = interests.ViewHistory[:500]
	}

	topicWeight := event.Engagement * 0.1
	interests.Topics[event.Topic] += topicWeight

	sourceWeight := event.Engagement * 0.05
	interests.Sources[event.Source] += sourceWeight

	keywordWeight := event.Engagement * 0.02
	for _, kw := range event.Keywords {
		interests.Keywords[kw] += keywordWeight
	}

	if event.Saved {
		interests.SavedArticles = append(interests.SavedArticles, event.URL)
	}

	interests.LastUpdated = time.Now()

	e.decayInterests(interests)

	return e.userStore.Save(ctx, interests)
}

func (e *PersonalizationEngine) UpdateTopicPreference(ctx context.Context, userID, topic string, weight float64) error {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		interests = &UserInterests{
			UserID:   userID,
			Topics:   make(map[string]float64),
			Sources:  make(map[string]float64),
			Keywords: make(map[string]float64),
		}
	}

	interests.Topics[topic] = weight
	interests.LastUpdated = time.Now()

	return e.userStore.Save(ctx, interests)
}

func (e *PersonalizationEngine) BlockSource(ctx context.Context, userID, source string) error {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		return err
	}

	for _, blocked := range interests.BlockedSources {
		if blocked == source {
			return nil
		}
	}

	interests.BlockedSources = append(interests.BlockedSources, source)
	interests.LastUpdated = time.Now()

	return e.userStore.Save(ctx, interests)
}

func (e *PersonalizationEngine) BlockTopic(ctx context.Context, userID, topic string) error {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		return err
	}

	for _, blocked := range interests.BlockedTopics {
		if blocked == topic {
			return nil
		}
	}

	interests.BlockedTopics = append(interests.BlockedTopics, topic)
	delete(interests.Topics, topic)
	interests.LastUpdated = time.Now()

	return e.userStore.Save(ctx, interests)
}

func (e *PersonalizationEngine) AddCustomCategory(ctx context.Context, userID string, category CustomCategory) error {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		return err
	}

	interests.CustomCategories = append(interests.CustomCategories, category)
	interests.LastUpdated = time.Now()

	return e.userStore.Save(ctx, interests)
}

func (e *PersonalizationEngine) GetUserTopics(ctx context.Context, userID string) (map[string]float64, error) {
	interests, err := e.userStore.Get(ctx, userID)
	if err != nil {
		return nil, err
	}
	return interests.Topics, nil
}

func (e *PersonalizationEngine) calculateRelevance(item FeedItem, interests *UserInterests) float64 {
	score := 0.0

	if topicScore, ok := interests.Topics[item.Topic]; ok {
		score += topicScore * e.config.TopicMatchWeight
	}

	if sourceScore, ok := interests.Sources[item.Source]; ok {
		score += sourceScore * e.config.SourceTrustWeight
	}

	keywordScore := 0.0
	for _, kw := range item.Keywords {
		if kwScore, ok := interests.Keywords[strings.ToLower(kw)]; ok {
			keywordScore += kwScore
		}
	}
	score += keywordScore * 0.1

	hoursSincePublish := time.Since(item.PublishedAt).Hours()
	recencyScore := math.Max(0, 1.0-hoursSincePublish/168.0)
	score += recencyScore * e.config.RecencyWeight

	if item.IsTrending {
		score *= e.config.TrendingBoost
	}

	if item.IsBreaking {
		score *= e.config.BreakingBoost
	}

	return score
}

func (e *PersonalizationEngine) calculateEngagement(event ViewEvent) float64 {
	engagement := 0.0

	if event.TimeSpent > 0 {
		readTimeScore := math.Min(1.0, float64(event.TimeSpent)/300.0)
		engagement += readTimeScore * 0.4
	}

	if event.Completed {
		engagement += 0.3
	}

	if event.Saved {
		engagement += 0.2
	}

	if event.Shared {
		engagement += 0.1
	}

	return engagement
}

func (e *PersonalizationEngine) explainRecommendation(item FeedItem, interests *UserInterests) string {
	if item.IsBreaking {
		return "Срочная новость"
	}

	if item.IsTrending {
		return "Популярно сейчас"
	}

	if topicScore, ok := interests.Topics[item.Topic]; ok && topicScore > 0.5 {
		return fmt.Sprintf("Из вашей категории: %s", item.Topic)
	}

	if sourceScore, ok := interests.Sources[item.Source]; ok && sourceScore > 0.3 {
		return fmt.Sprintf("Из источника, который вы читаете: %s", item.Source)
	}

	for _, kw := range item.Keywords {
		if kwScore, ok := interests.Keywords[strings.ToLower(kw)]; ok && kwScore > 0.2 {
			return fmt.Sprintf("По вашему интересу: %s", kw)
		}
	}

	return "Рекомендуем для вас"
}

func (e *PersonalizationEngine) getTopInterests(interests map[string]float64, limit int) []string {
	type kv struct {
		Key   string
		Value float64
	}

	var sorted []kv
	for k, v := range interests {
		sorted = append(sorted, kv{k, v})
	}

	sort.Slice(sorted, func(i, j int) bool {
		return sorted[i].Value > sorted[j].Value
	})

	result := make([]string, 0, limit)
	for i, item := range sorted {
		if i >= limit {
			break
		}
		result = append(result, item.Key)
	}

	return result
}

func (e *PersonalizationEngine) getTopKeywords(keywords map[string]float64, limit int) []string {
	return e.getTopInterests(keywords, limit)
}

func (e *PersonalizationEngine) deduplicateItems(items []FeedItem) []FeedItem {
	seen := make(map[string]bool)
	result := make([]FeedItem, 0, len(items))

	for _, item := range items {
		if !seen[item.URL] {
			seen[item.URL] = true
			result = append(result, item)
		}
	}

	return result
}

func (e *PersonalizationEngine) filterBlockedContent(items []FeedItem, interests *UserInterests) []FeedItem {
	blockedSources := make(map[string]bool)
	for _, s := range interests.BlockedSources {
		blockedSources[strings.ToLower(s)] = true
	}

	blockedTopics := make(map[string]bool)
	for _, t := range interests.BlockedTopics {
		blockedTopics[strings.ToLower(t)] = true
	}

	result := make([]FeedItem, 0, len(items))
	for _, item := range items {
		if blockedSources[strings.ToLower(item.Source)] {
			continue
		}
		if blockedTopics[strings.ToLower(item.Topic)] {
			continue
		}
		result = append(result, item)
	}

	return result
}

func (e *PersonalizationEngine) applyDiversity(items []FeedItem) []FeedItem {
	if len(items) <= 10 {
		return items
	}

	topicCounts := make(map[string]int)
	sourceCounts := make(map[string]int)
	maxPerTopic := len(items) / 5
	maxPerSource := len(items) / 4

	if maxPerTopic < 3 {
		maxPerTopic = 3
	}
	if maxPerSource < 3 {
		maxPerSource = 3
	}

	result := make([]FeedItem, 0, len(items))
	deferred := make([]FeedItem, 0)

	for _, item := range items {
		if topicCounts[item.Topic] >= maxPerTopic || sourceCounts[item.Source] >= maxPerSource {
			deferred = append(deferred, item)
			continue
		}

		topicCounts[item.Topic]++
		sourceCounts[item.Source]++
		result = append(result, item)
	}

	for _, item := range deferred {
		if len(result) >= e.config.MaxFeedItems {
			break
		}
		result = append(result, item)
	}

	return result
}

func (e *PersonalizationEngine) groupByCategory(items []FeedItem, interests *UserInterests) []FeedCategory {
	categoryMap := make(map[string][]FeedItem)

	for _, item := range items {
		categoryMap[item.Topic] = append(categoryMap[item.Topic], item)
	}

	categories := make([]FeedCategory, 0, len(categoryMap))
	categoryMeta := map[string]struct {
		Icon  string
		Color string
	}{
		"tech":        {"💻", "#3B82F6"},
		"finance":     {"💰", "#10B981"},
		"sports":      {"⚽", "#F59E0B"},
		"politics":    {"🏛️", "#6366F1"},
		"science":     {"🔬", "#8B5CF6"},
		"health":      {"🏥", "#EC4899"},
		"entertainment": {"🎬", "#F97316"},
		"world":       {"🌍", "#14B8A6"},
		"business":    {"📊", "#6B7280"},
		"culture":     {"🎭", "#A855F7"},
	}

	for topic, topicItems := range categoryMap {
		if len(topicItems) < 2 {
			continue
		}

		meta, ok := categoryMeta[strings.ToLower(topic)]
		if !ok {
			meta = struct {
				Icon  string
				Color string
			}{"📰", "#6B7280"}
		}

		categories = append(categories, FeedCategory{
			ID:    topic,
			Name:  topic,
			Icon:  meta.Icon,
			Color: meta.Color,
			Items: topicItems,
		})
	}

	for _, custom := range interests.CustomCategories {
		customItems := make([]FeedItem, 0)
		for _, item := range items {
			for _, kw := range custom.Keywords {
				if containsKeyword(item, kw) {
					customItems = append(customItems, item)
					break
				}
			}
		}
		if len(customItems) > 0 {
			categories = append(categories, FeedCategory{
				ID:       custom.ID,
				Name:     custom.Name,
				Icon:     "⭐",
				Color:    "#FBBF24",
				Items:    customItems,
				IsCustom: true,
			})
		}
	}

	sort.Slice(categories, func(i, j int) bool {
		iScore := interests.Topics[categories[i].ID]
		jScore := interests.Topics[categories[j].ID]
		return iScore > jScore
	})

	return categories
}

func (e *PersonalizationEngine) decayInterests(interests *UserInterests) {
	for k := range interests.Topics {
		interests.Topics[k] *= e.config.DecayFactor
		if interests.Topics[k] < 0.01 {
			delete(interests.Topics, k)
		}
	}

	for k := range interests.Sources {
		interests.Sources[k] *= e.config.DecayFactor
		if interests.Sources[k] < 0.01 {
			delete(interests.Sources, k)
		}
	}

	for k := range interests.Keywords {
		interests.Keywords[k] *= e.config.DecayFactor
		if interests.Keywords[k] < 0.01 {
			delete(interests.Keywords, k)
		}
	}
}

func (e *PersonalizationEngine) isArticleRead(url string, interests *UserInterests) bool {
	for _, event := range interests.ViewHistory {
		if event.URL == url {
			return true
		}
	}
	return false
}

func (e *PersonalizationEngine) isArticleSaved(url string, interests *UserInterests) bool {
	for _, saved := range interests.SavedArticles {
		if saved == url {
			return true
		}
	}
	return false
}

func containsKeyword(item FeedItem, keyword string) bool {
	kw := strings.ToLower(keyword)
	if strings.Contains(strings.ToLower(item.Title), kw) {
		return true
	}
	if strings.Contains(strings.ToLower(item.Summary), kw) {
		return true
	}
	for _, itemKw := range item.Keywords {
		if strings.ToLower(itemKw) == kw {
			return true
		}
	}
	return false
}

func (u *UserInterests) ToJSON() ([]byte, error) {
	return json.Marshal(u)
}

func ParseUserInterests(data []byte) (*UserInterests, error) {
	var interests UserInterests
	if err := json.Unmarshal(data, &interests); err != nil {
		return nil, err
	}
	return &interests, nil
}