Fix Artgrid collector matching and split ranker

backend/services/cse.go

@@ -90,6 +90,7 @@ func (s *SearchService) SearchMedia(queries []string, enabledPlatforms map[strin
 						continue
 					}
 					for _, item := range items {
+						item = normalizeResultForCollector(collector.Name(), item)
 						if item.Link == "" || seen[item.Link] || !collector.Accept(item) {
 							continue
 						}
@@ -379,10 +380,16 @@ func isRenderableArtgridResult(result SearchResult) bool {
 	if err != nil {
 		return false
 	}
-	if !strings.Contains(strings.ToLower(parsed.Host), "artgrid.io") {
+	host := strings.ToLower(parsed.Host)
+	switch {
+	case strings.Contains(host, "artgrid.io"):
+		return regexp.MustCompile(`^/clip/[0-9]+/`).MatchString(parsed.Path)
+	case strings.Contains(host, "artlist.io"):
+		trimmedPath := strings.TrimSuffix(parsed.Path, "/")
+		return regexp.MustCompile(`^/stock-footage/clip/.+/[0-9]+$`).MatchString(trimmedPath)
+	default:
 		return false
 	}
-	return regexp.MustCompile(`^/clip/[0-9]+/`).MatchString(parsed.Path)
 }
 
 func normalizeSource(source, link, engine string) string {
@@ -391,7 +398,7 @@ func normalizeSource(source, link, engine string) string {
 		return source
 	case strings.Contains(strings.ToLower(link), "envato") || strings.Contains(strings.ToLower(link), "videohive"):
 		return "Envato"
-	case strings.Contains(strings.ToLower(link), "artgrid"):
+	case strings.Contains(strings.ToLower(link), "artgrid"), strings.Contains(strings.ToLower(link), "artlist.io/stock-footage/clip/"):
 		return "Artgrid"
 	case strings.Contains(strings.ToLower(engine), "google"):
 		return "Google Video"
@@ -473,13 +480,60 @@ func extractArtgridBackgroundThumbnail(html, clipID string) string {
 }
 
 func extractArtgridClipID(link string) string {
-	matches := regexp.MustCompile(`/clip/([0-9]+)/`).FindStringSubmatch(link)
-	if len(matches) == 2 {
-		return matches[1]
+	patterns := []*regexp.Regexp{
+		regexp.MustCompile(`/clip/([0-9]+)/`),
+		regexp.MustCompile(`/stock-footage/clip/[^/]+/([0-9]+)$`),
+		regexp.MustCompile(`/stock-footage/clip/.+/([0-9]+)$`),
+	}
+	for _, pattern := range patterns {
+		matches := pattern.FindStringSubmatch(strings.TrimSuffix(link, "/"))
+		if len(matches) == 2 {
+			return matches[1]
+		}
 	}
 	return ""
 }
 
+func canonicalizeArtgridLink(link string) string {
+	trimmed := strings.TrimSpace(link)
+	if trimmed == "" {
+		return ""
+	}
+	clipID := extractArtgridClipID(trimmed)
+	if clipID == "" {
+		return trimmed
+	}
+	if strings.Contains(strings.ToLower(trimmed), "artgrid.io/clip/") {
+		return trimmed
+	}
+	parsed, err := url.Parse(trimmed)
+	if err != nil {
+		return trimmed
+	}
+	segments := strings.Split(strings.Trim(parsed.Path, "/"), "/")
+	slug := clipID
+	for idx, segment := range segments {
+		if segment == clipID && idx > 0 {
+			slug = segments[idx-1]
+			break
+		}
+	}
+	return "https://artgrid.io/clip/" + clipID + "/" + slug
+}
+
+func normalizeResultForCollector(source string, result SearchResult) SearchResult {
+	switch source {
+	case "Artgrid":
+		result.Link = canonicalizeArtgridLink(result.Link)
+		result.Source = "Artgrid"
+	case "Envato":
+		result.Source = "Envato"
+	case "Google Video":
+		result.Source = "Google Video"
+	}
+	return result
+}
+
 func collectURLs(body string) []string {
 	pattern := regexp.MustCompile(`https?:\/\/[^"'\\\s]+`)
 	matches := pattern.FindAllString(body, -1)