感谢 Issues 859提供的抖音思路 (#924)

* 感谢 Issues 859提供的抖音思路

* 补上漏掉的文件，去掉打印输出

Co-authored-by: wyndem <wyndem@qq.com>

extractors/douyin/douyin.go

@@ -3,19 +3,12 @@ package douyin
 import (
 	"encoding/json"
 	"errors"
-	"fmt"
-
 	"github.com/iawia002/annie/extractors/types"
 	"github.com/iawia002/annie/request"
 	"github.com/iawia002/annie/utils"
+	"strings"
 )
 
-type data struct {
-	ItemList []struct {
-		Desc string `json:"desc"`
-	} `json:"item_list"`
-}
-
 type extractor struct{}
 
 // New returns a youtube extractor.
@@ -26,17 +19,22 @@ func New() types.Extractor {
 // Extract is the main function to extract the data.
 func (e *extractor) Extract(url string, option types.Options) ([]*types.Data, error) {
 	var err error
-	html, err := request.Get(url, url, nil)
-	if err != nil {
-		return nil, err
+	itemIds := utils.MatchOneOf(url, `/video/(\d+)`)
+	if len(itemIds) == 0 {
+		return nil, errors.New("unable to get video ID")
 	}
 
-	realURLs := utils.MatchOneOf(html, `playAddr: "(.+?)"`)
-	if realURLs == nil || len(realURLs) < 2 {
+	if itemIds == nil || len(itemIds) < 2 {
 		return nil, types.ErrURLParseFailed
 	}
-	realURL := realURLs[1]
-
+	itemId := itemIds[len(itemIds)-1]
+	jsonData, err := request.Get("https://www.iesdouyin.com/web/api/v2/aweme/iteminfo/?item_ids="+itemId, url, nil)
+	var douyin douyinData
+	err = json.Unmarshal([]byte(jsonData), &douyin)
+	if err != nil {
+		return nil, err
+	}
+	realURL := strings.Replace(douyin.ItemList[0].Video.PlayAddr.URLList[0], "playwm", "play", -1)
 	size, err := request.Size(realURL, url)
 	if err != nil {
 		return nil, err
@@ -53,35 +51,10 @@ func (e *extractor) Extract(url string, option types.Options) ([]*types.Data, er
 		},
 	}
 
-	videoIDs := utils.MatchOneOf(url, `/video/(\d+)`)
-	if len(videoIDs) == 0 {
-		return nil, errors.New("unable to get video ID")
-	}
-	videoID := videoIDs[1]
-
-	dytks := utils.MatchOneOf(html, `dytk: "(.+?)"`)
-	if len(dytks) == 0 {
-		return nil, errors.New("unable to get dytk info")
-	}
-	dytk := dytks[1]
-
-	apiDataString, err := request.Get(
-		fmt.Sprintf("https://www.douyin.com/web/api/v2/aweme/iteminfo/?item_ids=%s&dytk=%s", videoID, dytk),
-		url, nil,
-	)
-	if err != nil {
-		return nil, err
-	}
-
-	var apiData data
-	if err = json.Unmarshal([]byte(apiDataString), &apiData); err != nil {
-		return nil, err
-	}
-
 	return []*types.Data{
 		{
 			Site:    "抖音 douyin.com",
-			Title:   apiData.ItemList[0].Desc,
+			Title:   douyin.ItemList[0].Desc,
 			Type:    types.DataTypeVideo,
 			Streams: streams,
 			URL:     url,

extractors/douyin/douyin_test.go

@@ -15,8 +15,8 @@ func TestDownload(t *testing.T) {
 		{
 			name: "normal test",
 			args: test.Args{
-				URL:   "https://www.douyin.com/share/video/6557825773007277319/?mid=6557826301539912456",
-				Title: "跟特效师一起学跳舞，看变形金刚擎天柱怎么跳，你也来试试！@抖音小助手",
+				URL:   "https://www.douyin.com/video/6967223681286278436?previous_page=main_page&tab_name=home",
+				Title: "是爱情，让父子相认#陈翔六点半  #关于爱情",
 			},
 		},
 	}

extractors/douyin/types.go

@@ -0,0 +1,128 @@
+package douyin
+
+type douyinData struct {
+	StatusCode int `json:"status_code"`
+	ItemList   []struct {
+		CommentList interface{}   `json:"comment_list"`
+		GroupID     int64         `json:"group_id"`
+		TextExtra   []interface{} `json:"text_extra"`
+		ImageInfos  interface{}   `json:"image_infos"`
+		AwemeID     string        `json:"aweme_id"`
+		ShareInfo   struct {
+			ShareWeiboDesc string `json:"share_weibo_desc"`
+			ShareDesc      string `json:"share_desc"`
+			ShareTitle     string `json:"share_title"`
+		} `json:"share_info"`
+		IsPreview int         `json:"is_preview"`
+		Images    interface{} `json:"images"`
+		RiskInfos struct {
+			Type    int    `json:"type"`
+			Content string `json:"content"`
+			Warn    bool   `json:"warn"`
+		} `json:"risk_infos"`
+		VideoText    interface{} `json:"video_text"`
+		LabelTopText interface{} `json:"label_top_text"`
+		Author       struct {
+			Nickname    string `json:"nickname"`
+			AvatarThumb struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"avatar_thumb"`
+			AvatarMedium struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"avatar_medium"`
+			PlatformSyncInfo interface{} `json:"platform_sync_info"`
+			PolicyVersion    interface{} `json:"policy_version"`
+			UID              string      `json:"uid"`
+			ShortID          string      `json:"short_id"`
+			Signature        string      `json:"signature"`
+			AvatarLarger     struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"avatar_larger"`
+			UniqueID        string      `json:"unique_id"`
+			FollowersDetail interface{} `json:"followers_detail"`
+			Geofencing      interface{} `json:"geofencing"`
+			TypeLabel       interface{} `json:"type_label"`
+		} `json:"author"`
+		Music struct {
+			CoverMedium struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"cover_medium"`
+			CoverThumb struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"cover_thumb"`
+			Duration int    `json:"duration"`
+			Status   int    `json:"status"`
+			Author   string `json:"author"`
+			Mid      string `json:"mid"`
+			Title    string `json:"title"`
+			CoverHd  struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"cover_hd"`
+			CoverLarge struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"cover_large"`
+			PlayURL struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"play_url"`
+			Position interface{} `json:"position"`
+			ID       int64       `json:"id"`
+		} `json:"music"`
+		ChaList interface{} `json:"cha_list"`
+		Video   struct {
+			PlayAddr struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"play_addr"`
+			Width        int    `json:"width"`
+			Ratio        string `json:"ratio"`
+			HasWatermark bool   `json:"has_watermark"`
+			Duration     int    `json:"duration"`
+			Vid          string `json:"vid"`
+			Cover        struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"cover"`
+			Height       int `json:"height"`
+			DynamicCover struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"dynamic_cover"`
+			OriginCover struct {
+				URI     string   `json:"uri"`
+				URLList []string `json:"url_list"`
+			} `json:"origin_cover"`
+			BitRate interface{} `json:"bit_rate"`
+		} `json:"video"`
+		ShareURL     string      `json:"share_url"`
+		AuthorUserID int64       `json:"author_user_id"`
+		Geofencing   interface{} `json:"geofencing"`
+		Promotions   interface{} `json:"promotions"`
+		LongVideo    interface{} `json:"long_video"`
+		ForwardID    string      `json:"forward_id"`
+		Desc         string      `json:"desc"`
+		CreateTime   int         `json:"create_time"`
+		Statistics   struct {
+			CommentCount int    `json:"comment_count"`
+			DiggCount    int    `json:"digg_count"`
+			PlayCount    int    `json:"play_count"`
+			ShareCount   int    `json:"share_count"`
+			AwemeID      string `json:"aweme_id"`
+		} `json:"statistics"`
+		VideoLabels  interface{} `json:"video_labels"`
+		Duration     int         `json:"duration"`
+		AwemeType    int         `json:"aweme_type"`
+		IsLiveReplay bool        `json:"is_live_replay"`
+	} `json:"item_list"`
+	Extra struct {
+		Now   int64  `json:"now"`
+		Logid string `json:"logid"`
+	} `json:"extra"`
+}