goquery 単語から兄弟要素を取得

type WebSpider struct {
	url      string
	keyword  string
	document *goquery.Document
}

func NewWebSpider(url string, keyword string) *WebSpider {
	return &WebSpider{
		url:      url,
		keyword:  keyword,
		document: nil,
	}
}

func (ws *WebSpider) Run() {
	ws.document, _ = ws.buildDocument()
	route := ws.findRoutesFromKeyWord()[ws.keyword]
	ws.document.Find(strings.Join(route, " ")).Each(func(i int, s *goquery.Selection) {
		fmt.Println(s.Text())
	})
}

func (ws *WebSpider) buildDocument() (*goquery.Document, error) {
	response, err := http.Get(ws.url)
	if err != nil {
		log.Fatal(err)
		return &goquery.Document{}, err
	}

	document, err := goquery.NewDocumentFromReader(response.Body)
	if err != nil {
		log.Fatal(err)
		return &goquery.Document{}, err
	}

	return document, nil
}

// Routesと複数形の理由はキーワードに正規表現を使ったときを想定して。
func (ws *WebSpider) findRoutesFromKeyWord() map[string][]string {
	routes := make(map[string][]string, 0)
	body := ws.document.Find("body")
	body.Find("*").Each(func(i int, s *goquery.Selection) {
		text := s.Clone().Children().Remove().End().Text()
		if text == ws.keyword {
			routes[text] = ws.findRouteFromNode(s)
		}
	})
	return routes
}

func (ws *WebSpider) findRouteFromNode(node *goquery.Selection) []string {
	route := make([]string, 0)
	currentNode := node
	for {
		route = append(route, goquery.NodeName(currentNode))
		if goquery.NodeName(currentNode) == "html" {
			break
		}
		currentNode = currentNode.Parent()
	}
	// reverse
	for i, j := 0, len(route)-1; i < j; i, j = i+1, j-1 {
		route[i], route[j] = route[j], route[i]
	}
	return route
}

func GoquerySample() {
	ws := NewWebSpider(
		"https://ja.wikipedia.org/wiki/%E5%9B%BD%E3%81%AE%E4%B8%80%E8%A6%A7",
		"日本国",
	)
	ws.Run()
}