gooseek/backend/internal/agent/travel_poi_collector.go

package agent

import (
	"context"
	"encoding/json"
	"fmt"
	"log"
	"regexp"
	"strings"
	"sync"
	"time"

	"github.com/gooseek/backend/internal/llm"
	"github.com/gooseek/backend/internal/search"
	"github.com/google/uuid"
)

// POI category queries for 2GIS Places API — concrete organization types
var poiCategoryQueries = map[string][]string{
	"attraction": {
		"достопримечательности",
		"памятники",
		"исторические здания",
		"смотровые площадки",
	},
	"museum": {
		"музеи",
		"галереи",
		"выставки",
	},
	"park": {
		"парки",
		"скверы",
		"сады",
		"набережные",
	},
	"restaurant": {
		"рестораны",
		"кафе",
	},
	"theater": {
		"театры",
		"кинотеатры",
		"филармония",
	},
	"entertainment": {
		"развлечения",
		"аквапарки",
		"зоопарки",
		"аттракционы",
		"боулинг",
	},
	"shopping": {
		"торговые центры",
		"рынки",
		"сувениры",
	},
	"religious": {
		"храмы",
		"церкви",
		"соборы",
		"мечети",
	},
}

// CollectPOIsEnriched collects POIs using 2GIS Places API as primary source,
// then enriches with descriptions from SearXNG + LLM.
func CollectPOIsEnriched(ctx context.Context, cfg TravelOrchestratorConfig, brief *TripBrief, destinations []destGeoEntry) ([]POICard, error) {
	if cfg.TravelData == nil {
		return nil, nil
	}

	var allPOIs []POICard

	// Phase 1: Collect concrete places from 2GIS for each destination
	for _, dest := range destinations {
		if dest.Lat == 0 && dest.Lng == 0 {
			continue
		}

		categories := selectCategoriesForBrief(brief)
		places := searchPlacesFrom2GIS(ctx, cfg, dest, categories)
		allPOIs = append(allPOIs, places...)
	}

	log.Printf("[travel-poi] 2GIS returned %d places total", len(allPOIs))

	// Phase 2: If 2GIS returned too few, supplement with SearXNG + LLM extraction
	if len(allPOIs) < 5 && cfg.SearchClient != nil {
		log.Printf("[travel-poi] 2GIS returned only %d places, supplementing with SearXNG+LLM", len(allPOIs))
		supplementPOIs := collectPOIsFromSearch(ctx, cfg, brief, destinations)
		allPOIs = append(allPOIs, supplementPOIs...)
	}

	// Phase 3: Enrich POIs with descriptions from SearXNG if available
	if cfg.SearchClient != nil && len(allPOIs) > 0 {
		allPOIs = enrichPOIDescriptions(ctx, cfg, brief, allPOIs)
	}

	// Phase 3b: Fetch photos via SearXNG images
	if cfg.SearchClient != nil && len(allPOIs) > 0 {
		allPOIs = enrichPOIPhotos(ctx, cfg, brief, allPOIs)
	}

	// Phase 4: Fallback geocoding for POIs without coordinates
	allPOIs = geocodePOIs(ctx, cfg, allPOIs)

	allPOIs = deduplicatePOIs(allPOIs)

	// Filter out POIs without coordinates — they can't be shown on map
	validPOIs := make([]POICard, 0, len(allPOIs))
	for _, p := range allPOIs {
		if p.Lat != 0 || p.Lng != 0 {
			validPOIs = append(validPOIs, p)
		}
	}

	if len(validPOIs) > 25 {
		validPOIs = validPOIs[:25]
	}

	log.Printf("[travel-poi] returning %d POIs with coordinates", len(validPOIs))
	return validPOIs, nil
}

// selectCategoriesForBrief picks relevant POI categories based on user interests.
func selectCategoriesForBrief(brief *TripBrief) []string {
	if len(brief.Interests) == 0 {
		return []string{"attraction", "museum", "park", "restaurant", "theater", "entertainment"}
	}

	interestMapping := map[string][]string{
		"культура":        {"museum", "theater", "attraction", "religious"},
		"музеи":           {"museum"},
		"еда":             {"restaurant"},
		"рестораны":       {"restaurant"},
		"природа":         {"park"},
		"парки":           {"park"},
		"развлечения":     {"entertainment"},
		"шопинг":          {"shopping"},
		"история":         {"attraction", "museum", "religious"},
		"архитектура":     {"attraction", "religious"},
		"дети":            {"entertainment", "park"},
		"семья":           {"entertainment", "park", "museum"},
		"семейный":        {"entertainment", "park", "museum"},
		"активный отдых":  {"entertainment", "park"},
		"религия":         {"religious"},
		"театр":           {"theater"},
		"искусство":       {"museum", "theater"},
	}

	seen := make(map[string]bool)
	var categories []string

	for _, interest := range brief.Interests {
		lower := strings.ToLower(interest)
		for keyword, cats := range interestMapping {
			if strings.Contains(lower, keyword) {
				for _, c := range cats {
					if !seen[c] {
						seen[c] = true
						categories = append(categories, c)
					}
				}
			}
		}
	}

	if len(categories) == 0 {
		return []string{"attraction", "museum", "park", "restaurant", "theater", "entertainment"}
	}

	// Always include attractions as baseline
	if !seen["attraction"] {
		categories = append(categories, "attraction")
	}

	return categories
}

// searchPlacesFrom2GIS queries 2GIS Places API for concrete organizations.
func searchPlacesFrom2GIS(ctx context.Context, cfg TravelOrchestratorConfig, dest destGeoEntry, categories []string) []POICard {
	var (
		mu     sync.Mutex
		pois   []POICard
		wg     sync.WaitGroup
		seen   = make(map[string]bool)
	)

	for _, category := range categories {
		queries, ok := poiCategoryQueries[category]
		if !ok {
			continue
		}

		for _, q := range queries {
			wg.Add(1)
			go func(query, cat string) {
				defer wg.Done()

				searchCtx, cancel := context.WithTimeout(ctx, 10*time.Second)
				defer cancel()

				fullQuery := fmt.Sprintf("%s %s", query, dest.Name)
				places, err := cfg.TravelData.SearchPlaces(searchCtx, fullQuery, dest.Lat, dest.Lng, 10000)
				if err != nil {
					log.Printf("[travel-poi] 2GIS search error for '%s': %v", fullQuery, err)
					return
				}

				mu.Lock()
				defer mu.Unlock()

				for _, place := range places {
					if seen[place.ID] || seen[place.Name] {
						continue
					}
					seen[place.ID] = true
					seen[place.Name] = true

					mappedCategory := mapPurposeToCategory(place.Purpose, place.Type, cat)

				pois = append(pois, POICard{
					ID:          place.ID,
					Name:        place.Name,
					Category:    mappedCategory,
					Address:     fmt.Sprintf("%s, %s", dest.Name, place.Address),
					Lat:         place.Lat,
					Lng:         place.Lng,
					Rating:      place.Rating,
					ReviewCount: place.ReviewCount,
					Schedule:    place.Schedule,
				})
				}
			}(q, category)
		}
	}

	wg.Wait()
	return pois
}

// mapPurposeToCategory maps 2GIS purpose/type to our POI category.
func mapPurposeToCategory(purpose, itemType, fallbackCategory string) string {
	lower := strings.ToLower(purpose)

	switch {
	case strings.Contains(lower, "музей") || strings.Contains(lower, "галере") || strings.Contains(lower, "выставк"):
		return "museum"
	case strings.Contains(lower, "ресторан") || strings.Contains(lower, "кафе") || strings.Contains(lower, "бар"):
		return "restaurant"
	case strings.Contains(lower, "парк") || strings.Contains(lower, "сквер") || strings.Contains(lower, "сад"):
		return "park"
	case strings.Contains(lower, "театр") || strings.Contains(lower, "кинотеатр") || strings.Contains(lower, "филармон"):
		return "theater"
	case strings.Contains(lower, "храм") || strings.Contains(lower, "церков") || strings.Contains(lower, "собор") || strings.Contains(lower, "мечет"):
		return "religious"
	case strings.Contains(lower, "торгов") || strings.Contains(lower, "магазин") || strings.Contains(lower, "рынок"):
		return "shopping"
	case strings.Contains(lower, "развлеч") || strings.Contains(lower, "аквапарк") || strings.Contains(lower, "зоопарк") || strings.Contains(lower, "аттракц"):
		return "entertainment"
	case strings.Contains(lower, "памятник") || strings.Contains(lower, "достоприм"):
		return "attraction"
	}

	if itemType == "attraction" {
		return "attraction"
	}

	return fallbackCategory
}

// collectPOIsFromSearch is the SearXNG + LLM fallback when 2GIS returns too few results.
func collectPOIsFromSearch(ctx context.Context, cfg TravelOrchestratorConfig, brief *TripBrief, destinations []destGeoEntry) []POICard {
	if cfg.SearchClient == nil {
		return nil
	}

	rawResults := searchForPOIs(ctx, cfg.SearchClient, brief)
	if len(rawResults) == 0 {
		return nil
	}

	var crawledContent []crawledPage
	if cfg.Crawl4AIURL != "" {
		crawledContent = crawlPOIPages(ctx, cfg.Crawl4AIURL, rawResults)
	}

	pois := extractPOIsWithLLM(ctx, cfg.LLM, brief, rawResults, crawledContent)

	return pois
}

// enrichPOIDescriptions adds descriptions to 2GIS POIs using SearXNG + LLM.
func enrichPOIDescriptions(ctx context.Context, cfg TravelOrchestratorConfig, brief *TripBrief, pois []POICard) []POICard {
	needsDescription := make([]int, 0)
	for i, p := range pois {
		if p.Description == "" {
			needsDescription = append(needsDescription, i)
		}
	}

	if len(needsDescription) == 0 {
		return pois
	}

	// Build a list of POI names for bulk enrichment via LLM
	var names []string
	for _, idx := range needsDescription {
		names = append(names, pois[idx].Name)
	}

	if len(names) > 15 {
		names = names[:15]
	}

	dest := strings.Join(brief.Destinations, ", ")
	prompt := fmt.Sprintf(`Ты — эксперт по туризму в %s. Для каждого места из списка напиши краткое описание (1-2 предложения), примерное время посещения в минутах и примерную стоимость входа в рублях (0 если бесплатно).

Места:
%s

Верни ТОЛЬКО JSON массив:
[
  {
    "name": "Точное название из списка",
    "description": "Краткое описание",
    "duration": число_минут,
    "price": цена_в_рублях,
    "rating": рейтинг_от_0_до_5
  }
]

Правила:
- Описание должно быть информативным и привлекательным для туриста
- duration: музей 60-120 мин, парк 30-90 мин, ресторан 60 мин, памятник 15-30 мин
- price: 0 для открытых мест, реальные цены для музеев/театров
- rating: если знаешь реальный рейтинг — используй, иначе оцени по популярности (3.5-5.0)
- Верни ТОЛЬКО JSON, без пояснений`, dest, strings.Join(names, "\n"))

	enrichCtx, cancel := context.WithTimeout(ctx, 30*time.Second)
	defer cancel()

	response, err := cfg.LLM.GenerateText(enrichCtx, llm.StreamRequest{
		Messages: []llm.Message{{Role: llm.RoleUser, Content: prompt}},
		Options:  llm.StreamOptions{MaxTokens: 4096, Temperature: 0.2},
	})
	if err != nil {
		log.Printf("[travel-poi] LLM enrichment failed: %v", err)
		return pois
	}

	jsonMatch := regexp.MustCompile(`\[[\s\S]*\]`).FindString(response)
	if jsonMatch == "" {
		return pois
	}

	var enrichments []struct {
		Name        string  `json:"name"`
		Description string  `json:"description"`
		Duration    int     `json:"duration"`
		Price       float64 `json:"price"`
		Rating      float64 `json:"rating"`
	}
	if err := json.Unmarshal([]byte(jsonMatch), &enrichments); err != nil {
		log.Printf("[travel-poi] enrichment JSON parse error: %v", err)
		return pois
	}

	enrichMap := make(map[string]int)
	for i, e := range enrichments {
		enrichMap[strings.ToLower(e.Name)] = i
	}

	for _, idx := range needsDescription {
		key := strings.ToLower(pois[idx].Name)
		if eIdx, ok := enrichMap[key]; ok {
			e := enrichments[eIdx]
			if e.Description != "" {
				pois[idx].Description = e.Description
			}
			if e.Duration > 0 {
				pois[idx].Duration = e.Duration
			}
			if e.Price > 0 {
				pois[idx].Price = e.Price
			}
			if e.Rating > 0 {
				pois[idx].Rating = e.Rating
			}
		}
	}

	return pois
}

func enrichPOIPhotos(ctx context.Context, cfg TravelOrchestratorConfig, brief *TripBrief, pois []POICard) []POICard {
	dest := ""
	if len(brief.Destinations) > 0 {
		dest = brief.Destinations[0]
	}

	maxEnrich := 15
	if len(pois) < maxEnrich {
		maxEnrich = len(pois)
	}

	var wg sync.WaitGroup
	var mu sync.Mutex

	for i := 0; i < maxEnrich; i++ {
		if len(pois[i].Photos) > 0 {
			continue
		}
		wg.Add(1)
		go func(idx int) {
			defer wg.Done()

			query := pois[idx].Name
			if dest != "" {
				query = pois[idx].Name + " " + dest
			}

			searchCtx, cancel := context.WithTimeout(ctx, 6*time.Second)
			defer cancel()

			resp, err := cfg.SearchClient.Search(searchCtx, query, &search.SearchOptions{
				Categories: []string{"images"},
				PageNo:     1,
			})
			if err != nil {
				return
			}

			var photos []string
			seen := make(map[string]bool)
			for _, r := range resp.Results {
				if len(photos) >= 3 {
					break
				}
				imgURL := r.ImgSrc
				if imgURL == "" {
					imgURL = r.ThumbnailSrc
				}
				if imgURL == "" {
					imgURL = r.Thumbnail
				}
				if imgURL == "" || seen[imgURL] {
					continue
				}
				seen[imgURL] = true
				photos = append(photos, imgURL)
			}

			if len(photos) > 0 {
				mu.Lock()
				pois[idx].Photos = photos
				mu.Unlock()
			}
		}(i)
	}

	wg.Wait()

	photosFound := 0
	for _, p := range pois {
		if len(p.Photos) > 0 {
			photosFound++
		}
	}
	log.Printf("[travel-poi] enriched %d/%d POIs with photos", photosFound, len(pois))

	return pois
}

type poiSearchResult struct {
	Title   string
	URL     string
	Content string
	Engine  string
}

func searchForPOIs(ctx context.Context, client *search.SearXNGClient, brief *TripBrief) []poiSearchResult {
	var results []poiSearchResult
	seen := make(map[string]bool)

	for _, dest := range brief.Destinations {
		queries := generatePOIQueries(dest, brief.Interests)

		for _, q := range queries {
			searchCtx, cancel := context.WithTimeout(ctx, 10*time.Second)
			resp, err := client.Search(searchCtx, q, &search.SearchOptions{
				Categories: []string{"general"},
				PageNo:     1,
			})
			cancel()

			if err != nil {
				log.Printf("[travel-poi] search error for '%s': %v", q, err)
				continue
			}

			for _, r := range resp.Results {
				if r.URL == "" || seen[r.URL] {
					continue
				}
				seen[r.URL] = true
				results = append(results, poiSearchResult{
					Title:   r.Title,
					URL:     r.URL,
					Content: r.Content,
					Engine:  r.Engine,
				})
			}
		}
	}

	return results
}

func generatePOIQueries(destination string, interests []string) []string {
	queries := []string{
		fmt.Sprintf("достопримечательности %s что посмотреть конкретные места", destination),
		fmt.Sprintf("лучшие рестораны %s рейтинг адреса", destination),
		fmt.Sprintf("музеи %s список адреса", destination),
	}

	for _, interest := range interests {
		queries = append(queries, fmt.Sprintf("%s %s адреса", interest, destination))
	}

	return queries
}

func crawlPOIPages(ctx context.Context, crawl4aiURL string, results []poiSearchResult) []crawledPage {
	maxCrawl := 5
	if len(results) < maxCrawl {
		maxCrawl = len(results)
	}

	var pages []crawledPage

	for _, r := range results[:maxCrawl] {
		crawlCtx, cancel := context.WithTimeout(ctx, 20*time.Second)
		page, err := crawlSinglePage(crawlCtx, crawl4aiURL, r.URL)
		cancel()

		if err != nil {
			log.Printf("[travel-poi] crawl failed for %s: %v", r.URL, err)
			continue
		}

		if page != nil && len(page.Content) > 100 {
			pages = append(pages, *page)
		}
	}

	return pages
}

func extractPOIsWithLLM(ctx context.Context, llmClient llm.Client, brief *TripBrief, searchResults []poiSearchResult, crawled []crawledPage) []POICard {
	var contextBuilder strings.Builder

	contextBuilder.WriteString("Результаты поиска мест и организаций:\n\n")
	for i, r := range searchResults {
		if i >= 15 {
			break
		}
		contextBuilder.WriteString(fmt.Sprintf("### %s\nURL: %s\n%s\n\n", r.Title, r.URL, truncateStr(r.Content, 500)))
	}

	if len(crawled) > 0 {
		contextBuilder.WriteString("\nПодробное содержание страниц:\n\n")
		for _, p := range crawled {
			contextBuilder.WriteString(fmt.Sprintf("### %s (%s)\n%s\n\n", p.Title, p.URL, truncateStr(p.Content, 3000)))
		}
	}

	prompt := fmt.Sprintf(`Ты — эксперт по туризму. Из предоставленного контента извлеки КОНКРЕТНЫЕ достопримечательности, рестораны, музеи, парки и интересные места в %s.

%s

КРИТИЧЕСКИ ВАЖНО:
- Извлекай ТОЛЬКО конкретные места с названиями (не статьи, не списки, не обзоры)
- Каждое место должно быть реальной организацией или объектом, который можно найти на карте
- НЕ включай заголовки статей типа "ТОП-25 достопримечательностей" — это НЕ место
- Адрес ОБЯЗАТЕЛЕН и должен включать город для геокодинга

Верни JSON массив:
[
  {
    "id": "уникальный id",
    "name": "Конкретное название места (не статьи!)",
    "description": "Краткое описание (1-2 предложения)",
    "category": "attraction|restaurant|museum|park|theater|shopping|entertainment|religious|viewpoint",
    "rating": число_от_0_до_5_или_0,
    "address": "Город, улица, дом (точный адрес)",
    "duration": время_посещения_в_минутах,
    "price": цена_входа_в_рублях_или_0,
    "currency": "RUB",
    "url": "ссылка на источник"
  }
]

Верни ТОЛЬКО JSON массив, без пояснений.`,
		strings.Join(brief.Destinations, ", "),
		contextBuilder.String(),
	)

	response, err := llmClient.GenerateText(ctx, llm.StreamRequest{
		Messages: []llm.Message{{Role: llm.RoleUser, Content: prompt}},
		Options:  llm.StreamOptions{MaxTokens: 4096, Temperature: 0.2},
	})
	if err != nil {
		log.Printf("[travel-poi] LLM extraction failed: %v", err)
		return nil
	}

	jsonMatch := regexp.MustCompile(`\[[\s\S]*\]`).FindString(response)
	if jsonMatch == "" {
		return nil
	}

	var pois []POICard
	if err := json.Unmarshal([]byte(jsonMatch), &pois); err != nil {
		log.Printf("[travel-poi] JSON parse error: %v", err)
		return nil
	}

	for i := range pois {
		if pois[i].ID == "" {
			pois[i].ID = uuid.New().String()
		}
	}

	return pois
}

func geocodePOIs(ctx context.Context, cfg TravelOrchestratorConfig, pois []POICard) []POICard {
	for i := range pois {
		if pois[i].Lat != 0 && pois[i].Lng != 0 {
			continue
		}

		// Try geocoding by address first, then by name + city
		queries := []string{}
		if pois[i].Address != "" {
			queries = append(queries, pois[i].Address)
		}
		if pois[i].Name != "" {
			queries = append(queries, pois[i].Name)
		}

		for _, query := range queries {
			geoCtx, cancel := context.WithTimeout(ctx, 5*time.Second)
			geo, err := cfg.TravelData.Geocode(geoCtx, query)
			cancel()

			if err != nil {
				continue
			}

			pois[i].Lat = geo.Lat
			pois[i].Lng = geo.Lng
			log.Printf("[travel-poi] geocoded '%s' -> %.4f, %.4f", query, geo.Lat, geo.Lng)
			break
		}

		if pois[i].Lat == 0 && pois[i].Lng == 0 {
			log.Printf("[travel-poi] failed to geocode POI '%s' (address: '%s')", pois[i].Name, pois[i].Address)
		}
	}

	return pois
}

func deduplicatePOIs(pois []POICard) []POICard {
	seen := make(map[string]bool)
	var unique []POICard

	for _, p := range pois {
		key := strings.ToLower(p.Name)
		if len(key) > 50 {
			key = key[:50]
		}
		if seen[key] {
			continue
		}
		seen[key] = true
		unique = append(unique, p)
	}

	return unique
}