feat: add zingmp3 extractor (#1280)

* feat: add zingmp3

* adhoc: temp commit

* register the extractor

---------

Co-authored-by: Xinzhao Xu <[email protected]>

.github/workflows/stream_zingmp3.yml

@@ -0,0 +1,31 @@
+name: zingmp3
+
+on:
+  push:
+    paths:
+      - "extractors/zingmp3/*.go"
+      - ".github/workflows/stream_zingmp3.yml"
+  pull_request:
+    paths:
+      - "extractors/zingmp3/*.go"
+      - ".github/workflows/stream_zingmp3.yml"
+  schedule:
+    # run ci weekly
+    - cron: "0 0 * * 0"
+
+jobs:
+  test:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        go: ["1.21"]
+        os: [ubuntu-latest]
+    name: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions/setup-go@v2
+        with:
+          go-version: ${{ matrix.go }}
+
+      - name: Test
+        run: go test -timeout 5m -race -coverpkg=./... -coverprofile=coverage.txt github.com/iawia002/lux/extractors/zingmp3

README.md

@@ -644,6 +644,7 @@ $ lux -j "https://www.bilibili.com/video/av20203945"
 | 知乎             | <https://zhihu.com>                                                       | ✓        |          |         |            |                  | [![zhihu](https://github.com/iawia002/lux/actions/workflows/stream_zhihu.yml/badge.svg)](https://github.com/iawia002/lux/actions/workflows/stream_zhihu.yml/)                     |
 | Rumble           | <https://rumble.com>                                                      | ✓        |          |         |            |                  | [![rumble](https://github.com/iawia002/lux/actions/workflows/stream_rumble.yml/badge.svg)](https://github.com/iawia002/lux/actions/workflows/stream_rumble.yml/)                  |
 | 小红书           | <https://xiaohongshu.com>                                                 | ✓        |          |         |            |                  | [![xiaohongshu](https://github.com/iawia002/lux/actions/workflows/stream_xiaohongshu.yml/badge.svg)](https://github.com/iawia002/lux/actions/workflows/stream_xiaohongshu.yml/)   |
+| Zing MP3         | <https://zingmp3.vn>                                                      | ✓        |          | ✓       |            |                  | [![zingmp3](https://github.com/iawia002/lux/actions/workflows/stream_zingmp3.yml/badge.svg)](https://github.com/iawia002/lux/actions/workflows/stream_zingmp3.yml/)               |
 
 
 ## Known issues

app/register.go

@@ -43,4 +43,5 @@ import (
 	_ "github.com/iawia002/lux/extractors/youku"
 	_ "github.com/iawia002/lux/extractors/youtube"
 	_ "github.com/iawia002/lux/extractors/zhihu"
+	_ "github.com/iawia002/lux/extractors/zingmp3"
 )

extractors/zingmp3/zingmp3.go

@@ -0,0 +1,220 @@
+package zingmp3
+
+import (
+	"crypto/hmac"
+	"crypto/sha256"
+	"crypto/sha512"
+	"encoding/hex"
+	"fmt"
+	"maps"
+	"net/http"
+	neturl "net/url"
+	"regexp"
+	"sort"
+
+	"github.com/buger/jsonparser"
+	"github.com/pkg/errors"
+
+	"github.com/iawia002/lux/extractors"
+	"github.com/iawia002/lux/request"
+	"github.com/iawia002/lux/utils"
+)
+
+func init() {
+	zingmp3Extractor := New()
+	extractors.Register("zingmp3", zingmp3Extractor)
+	extractors.Register("zing", zingmp3Extractor)
+}
+
+type extractor struct{}
+
+// New returns a zingmp3 extractor.
+func New() extractors.Extractor {
+	return &extractor{}
+}
+
+type params map[string]string
+
+var ApiSlugs = map[string]string{
+	"bai-hat":        "/api/v2/page/get/song",
+	"embed":          "/api/v2/page/get/song",
+	"video-clip":     "/api/v2/page/get/video",
+	"lyric":          "/api/v2/lyric/get/lyric",
+	"song-streaming": "/api/v2/song/get/streaming",
+}
+
+const Domain = "https://zingmp3.vn"
+
+// Extract is the main function to extract the data.
+func (e *extractor) Extract(url string, option extractors.Options) ([]*extractors.Data, error) {
+	urlRegExp := regexp.MustCompile(`https?://(?:mp3\.zing|zingmp3)\.vn/(?P<type>(?:bai-hat|video-clip|embed))/[^/?#]+/(?P<id>\w+)(?:\.html|\?)`)
+	urlMatcher := urlRegExp.FindStringSubmatch(url)
+	if len(urlMatcher) == 0 {
+		return nil, errors.WithStack(extractors.ErrURLParseFailed)
+	}
+	urlType := urlMatcher[1]
+	id := urlMatcher[2]
+	if err := updatingCookies(); err != nil {
+		return nil, errors.WithStack(err)
+	}
+	data := callApi(urlType, params{"id": id})
+	title, _ := jsonparser.GetString(data, "title")
+	var contentType extractors.DataType
+	var source []byte
+	if urlType == "video-clip" {
+		source, _, _, _ = jsonparser.Get(data, "streaming")
+		api := fmt.Sprintf(`http://api.mp3.zing.vn/api/mobile/video/getvideoinfo?requestdata={"id":"%s"}`, id)
+		res, _ := request.Get(api, api, nil)
+		newSource, _, _, _ := jsonparser.Get([]byte(res), "source")
+		source, _ = jsonparser.Set(source, newSource, "mp4")
+		contentType = extractors.DataTypeVideo
+	} else {
+		contentType = extractors.DataTypeAudio
+		source = callApi("song-streaming", params{"id": id})
+	}
+	streams := make(map[string]*extractors.Stream)
+	if err := jsonparser.ObjectEach(source, func(k []byte, v []byte, dataType jsonparser.ValueType, offset int) error {
+		key := string(k)
+		value := string(v)
+		if value == "" || value == "VIP" {
+			return nil
+		}
+
+		// Handle for audio
+		if key != "mp4" && key != "hls" {
+			size, _ := request.Size(value, url)
+			urlData := &extractors.Part{
+				URL:  value,
+				Ext:  "mp3",
+				Size: size,
+			}
+			streams["default"] = &extractors.Stream{
+				Parts: []*extractors.Part{urlData},
+			}
+			return nil
+		}
+
+		// Handle for video
+		return jsonparser.ObjectEach(v, func(kSource []byte, vSource []byte, _ jsonparser.ValueType, _ int) error {
+			resolution := string(kSource)
+			videoUrl := string(vSource)
+			if resolution == "" {
+				return nil
+			}
+			if resolution == "hls" {
+				urls, _ := utils.M3u8URLs(videoUrl)
+				parts := make([]*extractors.Part, 0)
+				for _, u := range urls {
+					parts = append(parts, &extractors.Part{
+						URL: u,
+						Ext: "ts",
+					})
+				}
+				streams[resolution] = &extractors.Stream{
+					ID:      resolution,
+					Parts:   parts,
+					NeedMux: false,
+				}
+				return nil
+			}
+			size, _ := request.Size(videoUrl, url)
+			streams[fmt.Sprintf("mp4-%s", resolution)] = &extractors.Stream{
+				Parts: []*extractors.Part{{
+					URL:  videoUrl,
+					Ext:  "mp4",
+					Size: size,
+				}},
+			}
+			return nil
+		})
+	}); err != nil {
+		return nil, errors.WithStack(err)
+	}
+
+	return []*extractors.Data{
+		{
+			Site:    "Zing MP3 zingmp3.vn",
+			Title:   title,
+			Type:    contentType,
+			Streams: streams,
+			URL:     url,
+		},
+	}, nil
+}
+
+func callApi(urlType string, p params) []byte {
+	api := generateApi(urlType, p)
+	res, _ := request.GetByte(api, api, nil)
+	data, _, _, _ := jsonparser.Get(res, "data")
+	return data
+}
+
+func updatingCookies() error {
+	// For the first time. We need to call the temp API to get cookies and set cookies to for next request
+	// But sometime zingmp3 doesn't return cookies. We need to retry get and set cookies again (only allow 5 time)
+	for i := 0; i < 5; i++ {
+		api := generateApi("bai-hat", params{"id": ""})
+		res, err := request.Request(http.MethodGet, api, nil, nil)
+		if err != nil {
+			return err
+		}
+		cookies := ""
+		for _, value := range res.Cookies() {
+			cookies += value.String()
+		}
+		res.Body.Close() // nolint
+		if cookies != "" {
+			request.SetOptions(request.Options{
+				Cookie: cookies,
+			})
+			return nil
+		}
+	}
+	return nil
+}
+
+func generateApi(urlType string, p params) string {
+	slugApi := ApiSlugs[urlType]
+	maps.Copy(p, params{"ctime": "1"})
+
+	sortedParams := sortedParams(p)
+	sig := generateSig(slugApi, sortedParams)
+	maps.Copy(sortedParams, params{
+		"apiKey": "X5BM3w8N7MKozC0B85o4KMlzLZKhV00y",
+		"sig":    sig,
+	})
+
+	urlParams := neturl.Values{}
+	for key, value := range sortedParams {
+		urlParams.Add(key, value)
+	}
+	return fmt.Sprintf("%s%s?%s", Domain, slugApi, urlParams.Encode())
+}
+
+func generateSig(slugApi string, p params) string {
+	str := ""
+	for key, value := range p {
+		str += fmt.Sprintf("%s=%s", key, value)
+	}
+	h := sha256.New()
+	h.Write([]byte(str))
+	sha256Value := hex.EncodeToString(h.Sum(nil))
+	var passwordBytes = []byte(fmt.Sprintf("%s%s", slugApi, sha256Value))
+	salt := []byte("acOrvUS15XRW2o9JksiK1KgQ6Vbds8ZW")
+	hmacHashed := hmac.New(sha512.New, salt)
+	hmacHashed.Write(passwordBytes)
+	return hex.EncodeToString(hmacHashed.Sum(nil))
+}
+
+func sortedParams(p params) params {
+	keys := make([]string, 0, len(p))
+	for k := range p {
+		keys = append(keys, k)
+	}
+	sort.Strings(keys)
+	sortedParams := params{}
+	for _, k := range keys {
+		sortedParams[k] = p[k]
+	}
+	return sortedParams
+}

extractors/zingmp3/zingmp3_test.go

@@ -0,0 +1,44 @@
+package zingmp3
+
+import (
+	"testing"
+
+	"github.com/iawia002/lux/extractors"
+	"github.com/iawia002/lux/test"
+)
+
+func TestDownload(t *testing.T) {
+	tests := []struct {
+		name string
+		args test.Args
+	}{
+		{
+			name: "Host is mp3.zing.vn",
+			args: test.Args{
+				URL:   "https://mp3.zing.vn/bai-hat/Xa-Mai-Xa-Bao-Thy/ZWZB9WAB.html",
+				Title: "Xa Mãi Xa",
+			},
+		},
+		{
+			name: "Host is zingmp3.vn",
+			args: test.Args{
+				URL:   "https://zingmp3.vn/bai-hat/SOLO-JENNIE/ZW9FID6Z.html",
+				Title: "SOLO",
+			},
+		},
+		{
+			name: "Video clip",
+			args: test.Args{
+				URL:   "https://zingmp3.vn/video-clip/Suong-Hoa-Dua-Loi-K-ICM-RYO/ZO8ZF7C7.html",
+				Title: "Sương Hoa Đưa Lối",
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			data, err := New().Extract(tt.args.URL, extractors.Options{})
+			test.CheckError(t, err)
+			test.Check(t, tt.args, data[0])
+		})
+	}
+}

go.mod

@@ -5,6 +5,7 @@ go 1.21
 require (
 	github.com/MercuryEngineering/CookieMonster v0.0.0-20180304172713-1584578b3403
 	github.com/PuerkitoBio/goquery v1.8.0
+	github.com/buger/jsonparser v1.1.1
 	github.com/cheggaaa/pb/v3 v3.0.8
 	github.com/dop251/goja v0.0.0-20230304130813-e2f543bf4b4c
 	github.com/fatih/color v1.13.0

go.sum

@@ -11,6 +11,8 @@ github.com/bitly/go-simplejson v0.5.0 h1:6IH+V8/tVMab511d5bn4M7EwGXZf9Hj6i2xSwkN
 github.com/bitly/go-simplejson v0.5.0/go.mod h1:cXHtHw4XUPsvGaxgjIAn8PhEWG9NfngEKAMDJEczWVA=
 github.com/bmizerany/assert v0.0.0-20160611221934-b7ed37b82869 h1:DDGfHa7BWjL4YnC6+E63dPcxHo2sUxDIu8g3QgEJdRY=
 github.com/bmizerany/assert v0.0.0-20160611221934-b7ed37b82869/go.mod h1:Ekp36dRnpXw/yCqJaO+ZrUyxD+3VXMFFr56k5XYrpB4=
+github.com/buger/jsonparser v1.1.1 h1:2PnMjfWD7wBILjqQbt530v576A/cAbQvEW9gGIpYMUs=
+github.com/buger/jsonparser v1.1.1/go.mod h1:6RYKKt7H4d4+iWqouImQ9R2FZql3VbhNgx27UK13J/0=
 github.com/cheggaaa/pb/v3 v3.0.8 h1:bC8oemdChbke2FHIIGy9mn4DPJ2caZYQnfbRqwmdCoA=
 github.com/cheggaaa/pb/v3 v3.0.8/go.mod h1:UICbiLec/XO6Hw6k+BHEtHeQFzzBH4i2/qk/ow1EJTA=
 github.com/chzyer/logex v1.1.10/go.mod h1:+Ywpsq7O8HXn0nuIou7OrIPyXbp3wmkHB+jjWRnGsAI=

utils/utils.go

@@ -62,7 +62,7 @@ func Domain(url string) string {
 	domainPattern := `([a-z0-9][-a-z0-9]{0,62})\.` +
 		`(com\.cn|com\.hk|` +
 		`cn|com|net|edu|gov|biz|org|info|pro|name|xxx|xyz|be|` +
-		`me|top|cc|tv|tt)`
+		`me|top|cc|tv|tt|vn)`
 	domain := MatchOneOf(url, domainPattern)
 	if domain != nil {
 		return domain[1]