store media IDs in description.Session

pkg/description/media.go

@@ -8,6 +8,7 @@ import (
 	"sort"
 	"strconv"
 	"strings"
+	"unicode"
 
 	psdp "github.com/pion/sdp/v3"
 
@@ -17,9 +18,23 @@ import (
 
 var smartRegexp = regexp.MustCompile("^([0-9]+) (.*?)/90000")
 
-func getControlAttribute(attributes []psdp.Attribute) string {
+func replaceSmartPayloadType(payloadType string, attributes []psdp.Attribute) string {
+	if payloadType == "smart/1/90000" {
+		for _, attr := range attributes {
+			if attr.Key == "rtpmap" {
+				sm := smartRegexp.FindStringSubmatch(attr.Value)
+				if sm != nil {
+					return sm[1]
+				}
+			}
+		}
+	}
+	return payloadType
+}
+
+func getAttribute(attributes []psdp.Attribute, key string) string {
 	for _, attr := range attributes {
-		if attr.Key == "control" {
+		if attr.Key == key {
 			return attr.Value
 		}
 	}
@@ -92,6 +107,15 @@ func sortedKeys(fmtp map[string]string) []string {
 	return keys
 }
 
+func isAlphaNumeric(v string) bool {
+	for _, r := range v {
+		if !unicode.IsLetter(r) && !unicode.IsNumber(r) {
+			return false
+		}
+	}
+	return true
+}
+
 // MediaDirection is the direction of a media stream.
 type MediaDirection string
 
@@ -118,7 +142,10 @@ type Media struct {
 	// Media type.
 	Type MediaType
 
-	// Direction of the stream.
+	// Media ID (optional).
+	ID string
+
+	// Direction of the stream (optional).
 	Direction MediaDirection
 
 	// Control attribute.
@@ -131,22 +158,18 @@ type Media struct {
 // Unmarshal decodes the media from the SDP format.
 func (m *Media) Unmarshal(md *psdp.MediaDescription) error {
 	m.Type = MediaType(md.MediaName.Media)
+
+	m.ID = getAttribute(md.Attributes, "mid")
+	if m.ID != "" && !isAlphaNumeric(m.ID) {
+		return fmt.Errorf("invalid mid: %v", m.ID)
+	}
+
 	m.Direction = getDirection(md.Attributes)
-	m.Control = getControlAttribute(md.Attributes)
+	m.Control = getAttribute(md.Attributes, "control")
 
 	m.Formats = nil
 	for _, payloadType := range md.MediaName.Formats {
-		if payloadType == "smart/1/90000" {
+		payloadType = replaceSmartPayloadType(payloadType, md.Attributes)
-			for _, attr := range md.Attributes {
-				if attr.Key == "rtpmap" {
-					sm := smartRegexp.FindStringSubmatch(attr.Value)
-					if sm != nil {
-						payloadType = sm[1]
-						break
-					}
-				}
-			}
-		}
 
 		tmp, err := strconv.ParseUint(payloadType, 10, 8)
 		if err != nil {
@@ -179,12 +202,13 @@ func (m Media) Marshal() *psdp.MediaDescription {
 			Media:  string(m.Type),
 			Protos: []string{"RTP", "AVP"},
 		},
-		Attributes: []psdp.Attribute{
+	}
-			{
+
-				Key:   "control",
+	if m.ID != "" {
-				Value: m.Control,
+		md.Attributes = append(md.Attributes, psdp.Attribute{
-			},
+			Key:   "mid",
-		},
+			Value: m.ID,
+		})
 	}
 
 	if m.Direction != "" {
@@ -193,6 +217,11 @@ func (m Media) Marshal() *psdp.MediaDescription {
 		})
 	}
 
+	md.Attributes = append(md.Attributes, psdp.Attribute{
+		Key:   "control",
+		Value: m.Control,
+	})
+
 	for _, forma := range m.Formats {
 		typ := strconv.FormatUint(uint64(forma.PayloadType()), 10)
 		md.MediaName.Formats = append(md.MediaName.Formats, typ)

pkg/description/session.go

@@ -9,6 +9,33 @@ import (
 	"github.com/bluenviron/gortsplib/v4/pkg/url"
 )
 
+func atLeastOneHasMID(medias []*Media) bool {
+	for _, media := range medias {
+		if media.ID != "" {
+			return true
+		}
+	}
+	return false
+}
+
+func atLeastOneDoesntHaveMID(medias []*Media) bool {
+	for _, media := range medias {
+		if media.ID == "" {
+			return true
+		}
+	}
+	return false
+}
+
+func hasMediaWithID(medias []*Media, id string) bool {
+	for _, media := range medias {
+		if media.ID == id {
+			return true
+		}
+	}
+	return false
+}
+
 // Session is the description of a RTSP stream.
 type Session struct {
 	// base URL of the stream (read only).
@@ -48,9 +75,18 @@ func (d *Session) Unmarshal(ssd *sdp.SessionDescription) error {
 		if err != nil {
 			return fmt.Errorf("media %d is invalid: %v", i+1, err)
 		}
+
+		if m.ID != "" && hasMediaWithID(d.Medias[:i], m.ID) {
+			return fmt.Errorf("duplicate media IDs")
+		}
+
 		d.Medias[i] = &m
 	}
 
+	if atLeastOneHasMID(d.Medias) && atLeastOneDoesntHaveMID(d.Medias) {
+		return fmt.Errorf("media IDs sent partially")
+	}
+
 	return nil
 }
 

pkg/description/session_test.go

@@ -51,8 +51,8 @@ var casesSession = []struct {
 			"a=rtpmap:97 H264/90000\r\n" +
 			"a=fmtp:97 packetization-mode=1; profile-level-id=640028; sprop-parameter-sets=Z2QAKKy0A8ARPyo=,aO4Bniw=\r\n" +
 			"m=audio 0 RTP/AVP 0\r\n" +
-			"a=control:rtsp://10.0.100.50/profile5/media.smp/trackID=a\r\n" +
 			"a=recvonly\r\n" +
+			"a=control:rtsp://10.0.100.50/profile5/media.smp/trackID=a\r\n" +
 			"a=rtpmap:0 PCMU/8000\r\n" +
 			"m=application 0 RTP/AVP 107\r\n" +
 			"a=control\r\n",
@@ -121,8 +121,8 @@ var casesSession = []struct {
 			"a=rtpmap:97 H264/90000\r\n" +
 			"a=fmtp:97 packetization-mode=1; profile-level-id=640028; sprop-parameter-sets=Z2QAKKy0A8ARPyo=,aO4Bniw=\r\n" +
 			"m=audio 0 RTP/AVP 0\r\n" +
-			"a=control:trackID=2\r\n" +
 			"a=recvonly\r\n" +
+			"a=control:trackID=2\r\n" +
 			"a=rtpmap:0 PCMU/8000\r\n" +
 			"m=application 0 RTP/AVP 107\r\n" +
 			"a=control\r\n",
@@ -264,8 +264,9 @@ var casesSession = []struct {
 			"c=IN IP4 0.0.0.0\r\n" +
 			"t=0 0\r\n" +
 			"m=audio 0 RTP/AVP 111 103 104 9 102 0 8 106 105 13 110 112 113 126\r\n" +
-			"a=control\r\n" +
+			"a=mid:audio\r\n" +
 			"a=sendonly\r\n" +
+			"a=control\r\n" +
 			"a=rtpmap:111 opus/48000/2\r\n" +
 			"a=fmtp:111 sprop-stereo=0\r\n" +
 			"a=rtpmap:103 ISAC/16000\r\n" +
@@ -282,8 +283,9 @@ var casesSession = []struct {
 			"a=rtpmap:113 telephone-event/16000\r\n" +
 			"a=rtpmap:126 telephone-event/8000\r\n" +
 			"m=video 0 RTP/AVP 96 97 98 99 100 101 127 124 125\r\n" +
-			"a=control\r\n" +
+			"a=mid:video\r\n" +
 			"a=sendonly\r\n" +
+			"a=control\r\n" +
 			"a=rtpmap:96 VP8/90000\r\n" +
 			"a=rtpmap:97 rtx/90000\r\n" +
 			"a=fmtp:97 apt=96\r\n" +
@@ -301,6 +303,7 @@ var casesSession = []struct {
 			Title: ``,
 			Medias: []*Media{
 				{
+					ID:        "audio",
 					Type:      MediaTypeAudio,
 					Direction: MediaDirectionSendonly,
 					Formats: []format.Format{
@@ -368,6 +371,7 @@ var casesSession = []struct {
 					},
 				},
 				{
+					ID:        "video",
 					Type:      MediaTypeVideo,
 					Direction: MediaDirectionSendonly,
 					Formats: []format.Format{
@@ -496,16 +500,16 @@ var casesSession = []struct {
 			"c=IN IP4 0.0.0.0\r\n" +
 			"t=0 0\r\n" +
 			"m=video 0 RTP/AVP 26\r\n" +
-			"a=control:rtsp://192.168.0.1/video\r\n" +
 			"a=recvonly\r\n" +
+			"a=control:rtsp://192.168.0.1/video\r\n" +
 			"a=rtpmap:26 JPEG/90000\r\n" +
 			"m=audio 0 RTP/AVP 0\r\n" +
-			"a=control:rtsp://192.168.0.1/audio\r\n" +
 			"a=recvonly\r\n" +
+			"a=control:rtsp://192.168.0.1/audio\r\n" +
 			"a=rtpmap:0 PCMU/8000\r\n" +
 			"m=audio 0 RTP/AVP 0\r\n" +
-			"a=control:rtsp://192.168.0.1/audioback\r\n" +
 			"a=sendonly\r\n" +
+			"a=control:rtsp://192.168.0.1/audioback\r\n" +
 			"a=rtpmap:0 PCMU/8000\r\n",
 		Session{
 			Title: `RTSP Session with audiobackchannel`,
@@ -710,6 +714,27 @@ func FuzzSessionUnmarshalErrors(f *testing.F) {
 		"a=fmtp:97 profile-level-id=1;mode=AAC-hbr;sizelength=13;indexlength=3;indexdeltalength=3;config=zzz1210\r\n" +
 		"a=control:streamid=1\r\n")
 
+	f.Add("v=0\r\n" +
+		"o=- 4158123474391860926 2 IN IP4 127.0.0.1\r\n" +
+		"s=-\r\n" +
+		"t=0 0\r\n" +
+		"m=video 42504 RTP/AVP 96\r\n" +
+		"a=rtpmap:96 H264/90000\r\n" +
+		"a=fmtp:96 packetization-mode=1\r\n" +
+		"m=audio 0 RTP/AVP/TCP 0\r\n" +
+		"a=mid:2\r\n")
+
+	f.Add("v=0\r\n" +
+		"o=- 4158123474391860926 2 IN IP4 127.0.0.1\r\n" +
+		"s=-\r\n" +
+		"t=0 0\r\n" +
+		"m=video 42504 RTP/AVP 96\r\n" +
+		"a=mid:2\r\n" +
+		"a=rtpmap:96 H264/90000\r\n" +
+		"a=fmtp:96 packetization-mode=1\r\n" +
+		"m=audio 0 RTP/AVP/TCP 0\r\n" +
+		"a=mid:2\r\n")
+
 	f.Fuzz(func(t *testing.T, enc string) {
 		var sd sdp.SessionDescription
 		err := sd.Unmarshal([]byte(enc))

pkg/description/testdata/fuzz/FuzzSessionUnmarshalErrors/b0432aab46b15405

@@ -0,0 +1,2 @@
+go test fuzz v1
+string("v=0\ns=0\nm=video 0 AVP 0\na=mid: ")