fix #47 修复快手解析

wujunwei928 · Oct 20, 2024 · e8bd8bf · e8bd8bf
1 parent dd16b1f
commit e8bd8bf
Showing 1 changed file with 25 additions and 44 deletions.
diff --git a/parser/kuaishou.go b/parser/kuaishou.go
@@ -1,9 +1,10 @@
 package parser
 
 import (
+	"bytes"
 	"errors"
 	"fmt"
-	"strings"
+	"regexp"
 
 	"github.com/go-resty/resty/v2"
 	"github.com/tidwall/gjson"
@@ -13,66 +14,46 @@ type kuaiShou struct{}
 
 func (k kuaiShou) parseShareUrl(shareUrl string) (*VideoParseInfo, error) {
 	client := resty.New()
-	// disable redirects in the HTTP client, get params before redirects
-	client.SetRedirectPolicy(resty.NoRedirectPolicy())
 	res, err := client.R().
 		SetHeader(HttpHeaderUserAgent, DefaultUserAgent).
 		SetHeader("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7").
 		Get(shareUrl)
-	// 非 resty.ErrAutoRedirectDisabled 错误时，返回错误
-	if !errors.Is(err, resty.ErrAutoRedirectDisabled) {
-		return nil, err
-	}
-
-	// 获取 cookies： did，didv
-	cookies := res.RawResponse.Cookies()
-	if len(cookies) <= 0 {
-		return nil, errors.New("get cookies from share url fail")
-	}
-
-	locationRes, err := res.RawResponse.Location()
 	if err != nil {
 		return nil, err
 	}
 
-	// 分享的中间跳转链接不太一样, 有些是 /fw/long-video , 有些 /fw/photo
-	referUri := strings.ReplaceAll(locationRes.String(), "v.m.chenzhongtech.com/fw/long-video", "m.gifshow.com/fw/photo")
-	referUri = strings.ReplaceAll(referUri, "v.m.chenzhongtech.com/fw/photo", "m.gifshow.com/fw/photo")
-	videoId := strings.ReplaceAll(strings.Trim(locationRes.Path, "/"), "fw/long-video/", "")
-	videoId = strings.ReplaceAll(videoId, "fw/photo/", "")
-	if len(videoId) <= 0 {
-		return nil, errors.New("parse video id from share url fail")
+	re := regexp.MustCompile(`window.INIT_STATE\s*=\s*(.*?)</script>`)
+	findRes := re.FindSubmatch(res.Body())
+	if len(findRes) < 2 {
+		return nil, errors.New("parse video json info from html fail")
 	}
+	jsonBytes := bytes.TrimSpace(findRes[1])
 
-	postData := map[string]interface{}{
-		"fid":               "0",
-		"shareResourceType": "PHOTO_OTHER",
-		"shareChannel":      "share_copylink",
-		"kpn":               "KUAISHOU",
-		"subBiz":            "BROWSE_SLIDE_PHOTO",
-		"env":               "SHARE_VIEWER_ENV_TX_TRICK",
-		"h5Domain":          "m.gifshow.com",
-		"photoId":           videoId,
-		"isLongVideo":       false,
+	var (
+		videoRes   gjson.Result
+		isFindInfo bool
+	)
+
+	for _, jsonItem := range gjson.ParseBytes(jsonBytes).Map() {
+		jsonItemMap := jsonItem.Map()
+		_, hasResult := jsonItemMap["result"]
+		_, hasPhoto := jsonItemMap["photo"]
+		if hasResult && hasPhoto {
+			videoRes = jsonItem
+			isFindInfo = true
+			break
+		}
 	}
-	videoRes, err := client.R().
-		SetHeader("Origin", "https://m.gifshow.com").
-		SetHeader(HttpHeaderReferer, strings.ReplaceAll(referUri, "m.gifshow.com/fw/photo", "m.gifshow.com/fw/photo")).
-		SetHeader(HttpHeaderContentType, "application/json").
-		SetHeader(HttpHeaderUserAgent, DefaultUserAgent).
-		SetCookies(cookies).
-		SetBody(postData).
-		Post("https://m.gifshow.com/rest/wd/photo/info?kpn=KUAISHOU&captchaToken=")
 
-	if err != nil {
-		return nil, err
+	if !isFindInfo {
+		return nil, errors.New("parse video json fail")
 	}
 
-	if resultCode := gjson.GetBytes(videoRes.Body(), "result").Int(); resultCode != 1 {
+	if resultCode := videoRes.Get("result").Int(); resultCode != 1 {
 		return nil, fmt.Errorf("获取作品信息失败:result=%d", resultCode)
 	}
 
-	data := gjson.GetBytes(videoRes.Body(), "photo")
+	data := videoRes.Get("photo")
 	avatar := data.Get("headUrl").String()
 	author := data.Get("userName").String()
 	title := data.Get("caption").String()