kkscrap-go/controllers/scraper/tct-main.go

package controllers_scraper

import (
	"encoding/json"
	"fmt"
	"kkscrap-go/controllers/scraper/cafe24"
	"kkscrap-go/controllers/scraper/godo"
	"kkscrap-go/controllers/scraper/magento"
	"kkscrap-go/controllers/scraper/shopify"
	"kkscrap-go/locals"

	"kkscrap-go/controllers/scraper/wordpress"
	"kkscrap-go/controllers/scraper/young"
	"kkscrap-go/model"
	"net/http"
	"net/url"
	"regexp"
	"strings"

	util "kkscrap-go/controllers/scraper/util"

	// "golang.org/x/crypto/bcrypt"

	"github.com/labstack/echo"
)

type SolutionTypeGetReq struct {
	Url string
}

func SolutionTypeGet(c echo.Context) error {

	v := c.Get("receiver").(SolutionTypeGetReq)

	retv := &struct {
		SolutionType string
		ThemeType    string
	}{}

	// itemInfo, err := parse(v.ItemUrl)
	// if err != nil {
	// 	return c.String(604, "ertvwerawqfd-ItemUrl Parse failed: "+err.Error())
	// }
	fmt.Println(v.Url)
	retv.SolutionType = "Wordpress"
	retv.ThemeType = "Avada"
	// ret, _ := json.MarshalIndent(itemInfo, "", "\t")
	ret, _ := json.Marshal(retv)
	return c.JSONBlob(http.StatusOK, ret)
}

type ProductPageGetReq struct {
	SolutionType string
	ThemeType    string
	Products     []ProductUri
}

type ProductUri struct {
	Uri string
}

// 오리지널 웹사이트 전체를 업테이트 하는 경우 Uri를 하나씩 보내면 비효율적이므로 하나의 배치로
// 묶어서 요청할 수 있도록 한다. 주로 product-page-get를 쓰고 item-url-scrap은 deprecate 예정임.

func ProductPageGet(c echo.Context) error {

	v := c.Get("receiver").(ProductPageGetReq)

	var vRet locals.ProductPage // Row(개별레코드)->Page(Row의 집합)->Book(Page의 집합)의 개념
	for _, row := range v.Products {

		// (1)Url 의 HTML를 2번 가져오는데 아래와 같이 1번만 가져와서 처리할 수 있도록 수정요.
		// kkk := htmlGet(row.Uri)
		if v.SolutionType == "" {
			// v.SolutionType, v.ThemeType = solType(&kkk)
		}

		prodInfo, err := parse(row.Uri)
		if err != nil {
			prodInfo.ItemNick = "Parsing Failed"
		}
		// 개별 prodInfo가 계속 추가될 수 있도록 코드를 변경요.
		// vRet.ProductPage = append(vRet.ProductPage, *prodInfo)
	}

	// ret, _ := json.MarshalIndent(itemInfo, "", "\t")
	ret, _ := json.Marshal(vRet)
	return c.JSONBlob(http.StatusOK, ret)
}

type ItemUrlScrapReq struct {
	ItemUrl string
}

func ItemUrlScrap(c echo.Context) error {

	v := c.Get("receiver").(ItemUrlScrapReq)

	// retv := &struct {
	// 	model.ItemInfo
	// }{}

	itemInfo, err := parse(v.ItemUrl)
	if err != nil {
		return c.String(604, "ertvwerawqfd-ItemUrl Parse failed: "+err.Error())
	}
	ret, _ := json.MarshalIndent(itemInfo, "", "\t")
	// fmt.Println(string(data))

	// ret, _ := json.Marshal(itemInfo)
	return c.JSONBlob(http.StatusOK, ret)
}

var regexpTitle *regexp.Regexp

func init() {
	regexpTitle, _ = regexp.Compile("<title>(.*)</title>")
}

func getTitle(body string) string {
	ss := regexpTitle.FindAllStringSubmatch(body, 1)
	if len(ss) == 1 {
		return ss[0][1]
	}
	return ""
}

func parse(uri string) (ret model.ItemInfo, err error) {
	t, err := getSolutionType(uri)
	if err != nil {
		return
	}
	ret = parseSolution(t, uri)
	return
}

func parseSolution(t model.SolutionType, uri string) (ret model.ItemInfo) {
	ret.SolutionName = t
	u, err := url.Parse(uri)
	if err != nil {
		return
	}
	ret.DomainName = u.Host
	ret.DomainURI = uri

	if t == model.SolutionTypeWooCommerce {
		wordpress.Parse(uri, &ret)
	} else if t == model.SolutionTypeShopify {
		shopify.Parse(uri, &ret)
	} else if t == model.SolutionTypeMagento {
		magento.Parse(uri, &ret)
	} else if t == model.SolutionTypeCafe24 {
		cafe24.Parse(uri, &ret)
	} else if t == model.SolutionTypeGodo {
		godo.Parse(uri, &ret)
	} else if t == model.SolutionTypeYoung {
		young.Parse(uri, &ret)
	} else if t == model.SolutionTypeOthers {
		magento.Parse(uri, &ret)
	}
	return
}

func getSolutionType(url string) (t model.SolutionType, reterr error) {
	body, err := util.Get(url)
	if err != nil {
		reterr = err
		return
	}
	if strings.Contains(body, "window.CAFE24") {
		t = model.SolutionTypeCafe24
	} else if strings.Contains(body, "woocommerce-page") {
		t = model.SolutionTypeWooCommerce
	} else if strings.Contains(body, "고도몰5") {
		t = model.SolutionTypeGodo
	} else if strings.Contains(body, "cdn.shopify.com") {
		t = model.SolutionTypeShopify
	} else if strings.Contains(body, "magento") {
		t = model.SolutionTypeMagento
	} else if strings.Contains(body, "it_id=") {
		t = model.SolutionTypeYoung
	} else {
		t = model.SolutionTypeOthers
		//reterr = errors.New("no found solution type")
	}

	//ioutil.WriteFile(string(t) + ".html", []byte(body), 644)
	return
}