kkscrap-go/controllers/scraper/cafe24/parser.go

package cafe24

import (
	"fmt"
	"kkscrap-go/controllers/scraper/common"
	util "kkscrap-go/controllers/scraper/util"
	"kkscrap-go/model"
	"strings"

	"github.com/PuerkitoBio/goquery"
)

func Parse(body string, item *model.ItemInfo) {
	item.Language = common.GetLanguage(body)

	doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))
	util.CheckError(err)
	//s := doc.Find("meta[property=\"og:url\"]")
	doc.Find("link ~ meta").Each(func(i int, s *goquery.Selection) {
		v, _ := s.Attr("property")
		if v == "og:description" {
			item.ShortDesc = s.AttrOr("content", "")
		} else if v == "og:title" {
			item.ItemName = s.AttrOr("content", "")
		} else if v == "og:image" {
			item.Images = append(item.Images, s.AttrOr("content", ""))
		} else if v == "product:price:amount" {
			item.SalesPrice = util.GetFloat32(s.AttrOr("content", ""))
		} else if v == "product:price:currency" {
			item.Currency = s.AttrOr("content", "")
		}
	})

	item.Images = getImages(body)
	item.OriginDesc = getDesc(body)
	item.TextDesc = getTextDesc(body)

	item.Emails = common.GetEmails(body)
	return
}

func getImages(body string) (images []string) {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))

	util.CheckError(err)
	s := doc.Find(".ThumbImage")
	s.Each(func(i int, selection *goquery.Selection) {
		if src, existed := selection.Attr("src"); existed {
			if strings.HasPrefix(src, "https:") {
				images = append(images, src)
			} else {
				images = append(images, fmt.Sprintf("https:%v", src))
			}
		}
	})

	return
}

func getDesc(body string) string {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))
	html := []string{}

	util.CheckError(err)
	s := doc.Find("div.xans-product #prdInfo")
	s.Each(func(i int, selection *goquery.Selection) {
		// log.Println(selection.Html())
		if !selection.HasClass("menu") {
			html = append(html, selection.Text())
		}
	})
	return strings.Join(html, "/n")
}

func getTextDesc(body string) string {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(body))
	util.CheckError(err)
	text := doc.Find("div.xans-product #prdDetail span p")
	html, _ := text.Html()
	return html
}