gooseek/backend/internal/search/media.go

package search

import (
	"context"
	"regexp"
	"strconv"
	"strings"

	"github.com/gooseek/backend/internal/types"
)

type MediaSearchOptions struct {
	MaxImages int
	MaxVideos int
}

type MediaSearchResult struct {
	Images []types.ImageData `json:"images"`
	Videos []types.VideoData `json:"videos"`
}

func (c *SearXNGClient) SearchMedia(ctx context.Context, query string, opts *MediaSearchOptions) (*MediaSearchResult, error) {
	if opts == nil {
		opts = &MediaSearchOptions{MaxImages: 8, MaxVideos: 6}
	}

	result := &MediaSearchResult{
		Images: make([]types.ImageData, 0),
		Videos: make([]types.VideoData, 0),
	}

	imageCh := make(chan []types.ImageData, 1)
	videoCh := make(chan []types.VideoData, 1)
	errCh := make(chan error, 2)

	go func() {
		images, err := c.searchImages(ctx, query, opts.MaxImages)
		if err != nil {
			errCh <- err
			imageCh <- nil
			return
		}
		errCh <- nil
		imageCh <- images
	}()

	go func() {
		videos, err := c.searchVideos(ctx, query, opts.MaxVideos)
		if err != nil {
			errCh <- err
			videoCh <- nil
			return
		}
		errCh <- nil
		videoCh <- videos
	}()

	<-errCh
	<-errCh
	result.Images = <-imageCh
	result.Videos = <-videoCh

	if result.Images == nil {
		result.Images = make([]types.ImageData, 0)
	}
	if result.Videos == nil {
		result.Videos = make([]types.VideoData, 0)
	}

	return result, nil
}

func (c *SearXNGClient) searchImages(ctx context.Context, query string, max int) ([]types.ImageData, error) {
	resp, err := c.Search(ctx, query, &SearchOptions{
		Categories: []string{"images"},
		PageNo:     1,
	})
	if err != nil {
		return nil, err
	}

	images := make([]types.ImageData, 0, max)
	seen := make(map[string]bool)

	for _, r := range resp.Results {
		if len(images) >= max {
			break
		}

		imgURL := r.ImgSrc
		if imgURL == "" {
			imgURL = r.ThumbnailSrc
		}
		if imgURL == "" {
			imgURL = r.Thumbnail
		}
		if imgURL == "" {
			continue
		}

		if seen[imgURL] {
			continue
		}
		seen[imgURL] = true

		images = append(images, types.ImageData{
			URL:       imgURL,
			Title:     r.Title,
			Source:    extractDomain(r.URL),
			SourceURL: r.URL,
		})
	}

	return images, nil
}

func (c *SearXNGClient) searchVideos(ctx context.Context, query string, max int) ([]types.VideoData, error) {
	resp, err := c.Search(ctx, query, &SearchOptions{
		Categories: []string{"videos"},
		PageNo:     1,
	})
	if err != nil {
		return nil, err
	}

	videos := make([]types.VideoData, 0, max)
	seen := make(map[string]bool)

	for _, r := range resp.Results {
		if len(videos) >= max {
			break
		}

		if seen[r.URL] {
			continue
		}
		seen[r.URL] = true

		platform := detectVideoPlatform(r.URL)

		video := types.VideoData{
			Title:     r.Title,
			URL:       r.URL,
			Thumbnail: r.Thumbnail,
			Duration:  toInt(r.Duration),
			Views:     toInt(r.Views),
			Author:    r.Author,
			Platform:  platform,
			EmbedURL:  r.IframeSrc,
		}

		videos = append(videos, video)
	}

	return videos, nil
}

var (
	youtubePattern = regexp.MustCompile(`youtube\.com|youtu\.be`)
	rutubePattern  = regexp.MustCompile(`rutube\.ru`)
	vkPattern      = regexp.MustCompile(`vk\.com`)
	dzenPattern    = regexp.MustCompile(`dzen\.ru`)
)

func detectVideoPlatform(url string) string {
	urlLower := strings.ToLower(url)

	if youtubePattern.MatchString(urlLower) {
		return "youtube"
	}
	if rutubePattern.MatchString(urlLower) {
		return "rutube"
	}
	if vkPattern.MatchString(urlLower) {
		return "vk"
	}
	if dzenPattern.MatchString(urlLower) {
		return "dzen"
	}

	return "other"
}

func extractDomain(rawURL string) string {
	rawURL = strings.TrimPrefix(rawURL, "https://")
	rawURL = strings.TrimPrefix(rawURL, "http://")
	rawURL = strings.TrimPrefix(rawURL, "www.")

	if idx := strings.Index(rawURL, "/"); idx > 0 {
		rawURL = rawURL[:idx]
	}

	return rawURL
}

func toInt(v interface{}) int {
	if v == nil {
		return 0
	}
	switch val := v.(type) {
	case int:
		return val
	case int64:
		return int(val)
	case float64:
		return int(val)
	case string:
		if i, err := strconv.Atoi(val); err == nil {
			return i
		}
		return 0
	default:
		return 0
	}
}