gooseek/backend/internal/agent/travel_hotels_collector.go

package agent

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

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

// CollectHotelsEnriched searches for hotels via SearXNG + Crawl4AI + LLM.
func CollectHotelsEnriched(ctx context.Context, cfg TravelOrchestratorConfig, brief *TripBrief, destinations []destGeoEntry) ([]HotelCard, error) {
	if cfg.SearchClient == nil {
		return nil, nil
	}

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

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

	hotels := extractHotelsWithLLM(ctx, cfg.LLM, brief, rawResults, crawledContent)

	hotels = geocodeHotels(ctx, cfg, hotels)

	hotels = deduplicateHotels(hotels)

	hotels = filterHotelsNearDestinations(hotels, destinations, 250)

	if len(hotels) > 10 {
		hotels = hotels[:10]
	}

	return hotels, nil
}

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

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

	for _, dest := range brief.Destinations {
		queries := generateHotelQueries(dest, brief.StartDate, brief.EndDate, brief.Travelers, brief.TravelStyle)

		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-hotels] 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, hotelSearchResult{
					Title:   r.Title,
					URL:     r.URL,
					Content: r.Content,
					Engine:  r.Engine,
				})
			}
		}
	}

	return results
}

func generateHotelQueries(destination, startDate, endDate string, travelers int, style string) []string {
	dateStr := ""
	if startDate != "" {
		dateStr = startDate
		if endDate != "" && endDate != startDate {
			dateStr += " " + endDate
		}
	}

	familyStr := ""
	if travelers >= 3 {
		familyStr = "семейный "
	}

	queries := []string{
		fmt.Sprintf("%sотели %s цены бронирование %s", familyStr, destination, dateStr),
		fmt.Sprintf("гостиницы %s рейтинг отзывы %d гостей", destination, travelers),
		fmt.Sprintf("лучшие отели %s для туристов", destination),
	}

	if travelers >= 3 {
		queries = append(queries, fmt.Sprintf("семейные отели %s с детьми", destination))
	}

	if style == "luxury" {
		queries = append(queries, fmt.Sprintf("5 звезд отели %s премиум", destination))
	} else if style == "budget" {
		queries = append(queries, fmt.Sprintf("хостелы %s дешево %d человек", destination, travelers))
	} else {
		queries = append(queries, fmt.Sprintf("где остановиться %s %d человек", destination, travelers))
	}

	return queries
}

func crawlHotelPages(ctx context.Context, crawl4aiURL string, results []hotelSearchResult) []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-hotels] crawl failed for %s: %v", r.URL, err)
			continue
		}

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

	return pages
}

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

	contextBuilder.WriteString("Результаты поиска отелей:\n\n")
	maxSearch := 10
	if len(searchResults) < maxSearch {
		maxSearch = len(searchResults)
	}
	for i := 0; i < maxSearch; i++ {
		r := searchResults[i]
		contextBuilder.WriteString(fmt.Sprintf("### %s\nURL: %s\n%s\n\n", r.Title, r.URL, truncateStr(r.Content, 300)))
	}

	if len(crawled) > 0 {
		contextBuilder.WriteString("\nПодробное содержание:\n\n")
		maxCrawled := 3
		if len(crawled) < maxCrawled {
			maxCrawled = len(crawled)
		}
		for i := 0; i < maxCrawled; i++ {
			p := crawled[i]
			contextBuilder.WriteString(fmt.Sprintf("### %s (%s)\n%s\n\n", p.Title, p.URL, truncateStr(p.Content, 2000)))
		}
	}

	nightsStr := "1 ночь"
	nightsCount := calculateNights(brief.StartDate, brief.EndDate)
	if brief.StartDate != "" && brief.EndDate != "" && brief.EndDate != brief.StartDate {
		nightsStr = fmt.Sprintf("с %s по %s (%d ночей)", brief.StartDate, brief.EndDate, nightsCount)
	}

	travelers := brief.Travelers
	if travelers < 1 {
		travelers = 1
	}
	rooms := calculateRooms(travelers)

	prompt := fmt.Sprintf(`Извлеки до 6 отелей в %s на %s для %d чел (%d номеров).

%s

JSON массив (ТОЛЬКО JSON, без текста):
[{"id":"hotel-1","name":"Название","stars":3,"rating":8.5,"reviewCount":120,"pricePerNight":3500,"totalPrice":0,"currency":"RUB","address":"Город, ул. Улица, д. 1","bookingUrl":"https://...","amenities":["Wi-Fi","Завтрак"],"pros":["Центр города"],"checkIn":"%s","checkOut":"%s"}]

Правила:
- ТОЛЬКО реальные отели из текста
- pricePerNight — за 1 номер за 1 ночь в рублях. Если не указана — оцени по звёздам: 1★=1500, 2★=2500, 3★=3500, 4★=5000, 5★=8000
- totalPrice=0 (рассчитается автоматически)
- Адрес с городом для геокодинга
- Максимум 6 отелей, компактный JSON`,
		strings.Join(brief.Destinations, ", "),
		nightsStr,
		travelers,
		rooms,
		contextBuilder.String(),
		brief.StartDate,
		brief.EndDate,
	)

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

	response, err := llmClient.GenerateText(llmCtx, llm.StreamRequest{
		Messages: []llm.Message{{Role: llm.RoleUser, Content: prompt}},
		Options:  llm.StreamOptions{MaxTokens: 3000, Temperature: 0.1},
	})
	if err != nil {
		log.Printf("[travel-hotels] LLM extraction failed: %v", err)
		return fallbackHotelsFromSearch(searchResults, brief)
	}

	jsonMatch := regexp.MustCompile(`\[[\s\S]*\]`).FindString(response)
	if jsonMatch == "" {
		log.Printf("[travel-hotels] no JSON array found in LLM response (len=%d)", len(response))
		return fallbackHotelsFromSearch(searchResults, brief)
	}

	var hotels []HotelCard
	if err := json.Unmarshal([]byte(jsonMatch), &hotels); err != nil {
		log.Printf("[travel-hotels] JSON parse error: %v, response len=%d", err, len(jsonMatch))
		hotels = tryPartialHotelParse(jsonMatch)
		if len(hotels) == 0 {
			return fallbackHotelsFromSearch(searchResults, brief)
		}
	}

	nights := nightsCount
	guestRooms := rooms
	guests := travelers

	for i := range hotels {
		if hotels[i].ID == "" {
			hotels[i].ID = uuid.New().String()
		}
		if hotels[i].CheckIn == "" {
			hotels[i].CheckIn = brief.StartDate
		}
		if hotels[i].CheckOut == "" {
			hotels[i].CheckOut = brief.EndDate
		}
		hotels[i].Nights = nights
		hotels[i].Rooms = guestRooms
		hotels[i].Guests = guests

		if hotels[i].PricePerNight > 0 && hotels[i].TotalPrice == 0 {
			hotels[i].TotalPrice = hotels[i].PricePerNight * float64(nights) * float64(guestRooms)
		}
		if hotels[i].TotalPrice > 0 && hotels[i].PricePerNight == 0 && nights > 0 && guestRooms > 0 {
			hotels[i].PricePerNight = hotels[i].TotalPrice / float64(nights) / float64(guestRooms)
		}
		if hotels[i].PricePerNight == 0 {
			hotels[i].PricePerNight = estimatePriceByStars(hotels[i].Stars)
			hotels[i].TotalPrice = hotels[i].PricePerNight * float64(nights) * float64(guestRooms)
		}
	}

	log.Printf("[travel-hotels] extracted %d hotels from LLM", len(hotels))
	return hotels
}

func tryPartialHotelParse(jsonStr string) []HotelCard {
	var hotels []HotelCard
	objRegex := regexp.MustCompile(`\{[^{}]*"name"\s*:\s*"[^"]+[^{}]*\}`)
	matches := objRegex.FindAllString(jsonStr, -1)
	for _, m := range matches {
		var h HotelCard
		if err := json.Unmarshal([]byte(m), &h); err == nil && h.Name != "" {
			hotels = append(hotels, h)
		}
	}
	if len(hotels) > 0 {
		log.Printf("[travel-hotels] partial parse recovered %d hotels", len(hotels))
	}
	return hotels
}

func estimatePriceByStars(stars int) float64 {
	switch {
	case stars >= 5:
		return 8000
	case stars == 4:
		return 5000
	case stars == 3:
		return 3500
	case stars == 2:
		return 2500
	default:
		return 2000
	}
}

func calculateNights(startDate, endDate string) int {
	if startDate == "" || endDate == "" {
		return 1
	}
	start, err1 := time.Parse("2006-01-02", startDate)
	end, err2 := time.Parse("2006-01-02", endDate)
	if err1 != nil || err2 != nil {
		return 1
	}
	nights := int(end.Sub(start).Hours() / 24)
	if nights < 1 {
		return 1
	}
	return nights
}

func calculateRooms(travelers int) int {
	if travelers <= 2 {
		return 1
	}
	return (travelers + 1) / 2
}

func fallbackHotelsFromSearch(results []hotelSearchResult, brief *TripBrief) []HotelCard {
	hotels := make([]HotelCard, 0, len(results))
	nights := calculateNights(brief.StartDate, brief.EndDate)
	travelers := brief.Travelers
	if travelers < 1 {
		travelers = 1
	}
	rooms := calculateRooms(travelers)

	for _, r := range results {
		if len(hotels) >= 5 {
			break
		}
		name := r.Title
		if len(name) > 80 {
			name = name[:80]
		}

		price := extractPriceFromSnippet(r.Content)
		if price == 0 {
			price = 3000
		}

		hotels = append(hotels, HotelCard{
			ID:            uuid.New().String(),
			Name:          name,
			Stars:         3,
			PricePerNight: price,
			TotalPrice:    price * float64(nights) * float64(rooms),
			Rooms:         rooms,
			Nights:        nights,
			Guests:        travelers,
			Currency:      "RUB",
			CheckIn:       brief.StartDate,
			CheckOut:      brief.EndDate,
			BookingURL:    r.URL,
		})
	}
	log.Printf("[travel-hotels] fallback: %d hotels from search results", len(hotels))
	return hotels
}

func extractPriceFromSnippet(text string) float64 {
	priceRegex := regexp.MustCompile(`(\d[\d\s]*\d)\s*(?:₽|руб|RUB|р\.)`)
	match := priceRegex.FindStringSubmatch(text)
	if len(match) >= 2 {
		numStr := strings.ReplaceAll(match[1], " ", "")
		var price float64
		if _, err := fmt.Sscanf(numStr, "%f", &price); err == nil && price > 100 && price < 500000 {
			return price
		}
	}
	return 0
}

func geocodeHotels(ctx context.Context, cfg TravelOrchestratorConfig, hotels []HotelCard) []HotelCard {
	for i := range hotels {
		if hotels[i].Address == "" || (hotels[i].Lat != 0 && hotels[i].Lng != 0) {
			continue
		}

		geoCtx, cancel := context.WithTimeout(ctx, 5*time.Second)
		geo, err := cfg.TravelData.Geocode(geoCtx, hotels[i].Address)
		cancel()

		if err != nil {
			log.Printf("[travel-hotels] geocode failed for '%s': %v", hotels[i].Address, err)
			continue
		}

		hotels[i].Lat = geo.Lat
		hotels[i].Lng = geo.Lng
	}

	return hotels
}

func filterHotelsNearDestinations(hotels []HotelCard, destinations []destGeoEntry, maxKm float64) []HotelCard {
	if len(destinations) == 0 {
		return hotels
	}
	filtered := make([]HotelCard, 0, len(hotels))
	for _, h := range hotels {
		if h.Lat == 0 && h.Lng == 0 {
			continue
		}
		minD := 1e18
		for _, d := range destinations {
			if d.Lat == 0 && d.Lng == 0 {
				continue
			}
			dd := distanceKm(h.Lat, h.Lng, d.Lat, d.Lng)
			if dd < minD {
				minD = dd
			}
		}
		if minD <= maxKm {
			filtered = append(filtered, h)
		} else {
			log.Printf("[travel-hotels] dropped far hotel '%s' (%.0fkm from destinations)", h.Name, minD)
		}
	}
	return filtered
}

func deduplicateHotels(hotels []HotelCard) []HotelCard {
	seen := make(map[string]bool)
	var unique []HotelCard

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

	return unique
}