新增虎扑视频下载 (#986)

* feat: Add Hupu Download

* feat: readme

* feat: fix note

* fix: package sort

* fix: code style

* feat: delete unused struct

Co-authored-by: 瓜瓜 <wangjunwei03@corp.netease.com>

README.md

@@ -624,6 +624,7 @@ $ annie -j "https://www.bilibili.com/video/av20203945"
 | AcFun      | <https://www.acfun.cn>       | ✓        |          | ✓          |                  |
 | Eporner    | <https://eporner.com>        | ✓        |          |            |                  |
 | StreamTape | <https://streamtape.com>     | ✓        |          |            |                  |
+| 虎扑       | https://hupu.com             | ✓        |          |            |                  |
 
 
 ## Known issues

extractors/extractors.go

@@ -13,6 +13,7 @@ import (
 	"github.com/iawia002/annie/extractors/facebook"
 	"github.com/iawia002/annie/extractors/geekbang"
 	"github.com/iawia002/annie/extractors/haokan"
+	"github.com/iawia002/annie/extractors/hupu"
 	"github.com/iawia002/annie/extractors/instagram"
 	"github.com/iawia002/annie/extractors/iqiyi"
 	"github.com/iawia002/annie/extractors/mgtv"
@@ -81,6 +82,7 @@ func init() {
 		"eporner":    eporner.New(),
 		"streamtape": stExtractor,
 		"streamta":   stExtractor, // streamta.pe
+		"hupu":       hupu.New(),
 	}
 }
 

extractors/hupu/hupu.go

@@ -0,0 +1,64 @@
+package hupu
+
+import (
+	"github.com/iawia002/annie/extractors/types"
+	"github.com/iawia002/annie/request"
+	"github.com/iawia002/annie/utils"
+)
+
+type extractor struct{}
+
+// New returns a hupu extractor.
+func New() types.Extractor {
+	return &extractor{}
+}
+
+func (e *extractor) Extract(url string, option types.Options) ([]*types.Data, error) {
+	html, err := request.Get(url, url, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	var title string
+	titleDesc := utils.MatchOneOf(html, `<span class="post-user-comp-info-bottom-title">(.+?)</span>`)
+	if len(titleDesc) > 1 {
+		title = titleDesc[1]
+	} else {
+		title = "hupu video"
+	}
+
+	var videoUrl string
+	urlDesc := utils.MatchOneOf(html, `<video src="(.+?)" controls="" poster=(.+?)></video>`)
+	if len(urlDesc) > 1 {
+		videoUrl = urlDesc[1]
+	} else {
+		return nil, types.ErrURLParseFailed
+	}
+
+	size, err := request.Size(videoUrl, url)
+	if err != nil {
+		return nil, err
+	}
+	urlData := &types.Part{
+		URL:  videoUrl,
+		Size: size,
+		Ext:  "mp4",
+	}
+	quality := "normal"
+	streams := map[string]*types.Stream{
+		quality: {
+			Parts:   []*types.Part{urlData},
+			Size:    size,
+			Quality: quality,
+		},
+	}
+	return []*types.Data{
+		{
+			Site:    "虎扑 hupu.com",
+			Title:   title,
+			Type:    types.DataTypeVideo,
+			Streams: streams,
+			URL:     url,
+		},
+	}, nil
+}

extractors/hupu/hupu_test.go

@@ -0,0 +1,28 @@
+package hupu
+
+import (
+	"testing"
+
+	"github.com/iawia002/annie/extractors/types"
+	"github.com/iawia002/annie/test"
+)
+
+func TestHupu(t *testing.T) {
+	tests := []struct {
+		name string
+		args test.Args
+	}{
+		{
+			name: "normal test",
+			args: test.Args{
+				URL:   "https://bbs.hupu.com/47401018.html?is_reflow=1&cid=84752419&bddid=56KXU5QUJH4VGM26SFPTYTKNI5CFNJMX736TIZ52DXLGUAAMBJVA01&puid=16522089&client=8577E496-4D9B-4E5C-A9DB-A8EF5C1956D2",
+				Title: "结局引起舒适",
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			New().Extract(tt.args.URL, types.Options{})
+		})
+	}
+}