RSS Detection GO

rss.go

func DetectRSSLink(ctx context.Context, URL string, body string) []string {
	doc, err := html.Parse(strings.NewReader(body))
	var candidates []string
	if err != nil {
		return candidates
	}
	links := FindAll(doc, "link")
	for _, link := range links {
		if GetAttr(link, "rel") == "alternate" {
			attrHref := GetAttr(link, "href")
			attrType := GetAttr(link, "type")
			if len(attrHref) > 0 && len(attrType) > 0 {
				if strings.Contains(attrType, "feed") || strings.Contains(attrType, "rss") || strings.Contains(attrType, "atom") {
					resURL, err := resolveURL(ctx, URL, attrHref)
					if err == nil {
						candidates = append(candidates, resURL)
					}
				}
			}
		}
	}
	aTags := FindAll(doc, "a")
	for _, aTag := range aTags {
		href := GetAttr(aTag, "href")
		if len(href) > 0 && (strings.Contains(href, "xml") || strings.Contains(href, "feed") || strings.Contains(href, "rss")) || strings.Contains(href, "atom") {
			resURL, err := resolveURL(ctx, URL, href)
			if err == nil {
				candidates = append(candidates, resURL)
			}
		}
	}
	if len(candidates) > 1 {
		sort.Slice(candidates, func(i int, j int) bool {
			return getScore(candidates[i], URL) < getScore(candidates[j], URL)
		})
	}
	return candidates
}