Update version to 1.1.0

.github/workflows/release.yml

@@ -2,9 +2,9 @@ name: release
 
 on:
   workflow_dispatch:
-  push:
-    tags:
-      - 'v*'
+#  push:
+#    tags:
+#      - 'v*'
 
 jobs:
   build-and-release:

README.md

@@ -27,6 +27,40 @@ Ultimate camera streaming application with support RTSP, WebRTC, HomeKit, FFmpeg
 - [MediaSoup](https://mediasoup.org/) framework routing idea
 - HomeKit Accessory Protocol from [@brutella](https://github.com/brutella/hap)
 
+---
+
+* [Fast start](#fast-start)
+  * [go2rtc: Binary](#go2rtc-binary)
+  * [go2rtc: Home Assistant Add-on](#go2rtc-home-assistant-add-on)
+  * [go2rtc: Docker](#go2rtc-docker)
+* [Configuration](#configuration)
+  * [Module: Streams](#module-streams)
+    * [Source: RTSP](#source-rtsp)
+    * [Source: RTMP](#source-rtmp)
+    * [Source: HTTP](#source-http)
+    * [Source: FFmpeg](#source-ffmpeg)
+    * [Source: FFmpeg Device](#source-ffmpeg-device)
+    * [Source: Exec](#source-exec)
+    * [Source: Echo](#source-echo)
+    * [Source: HomeKit](#source-homekit)
+    * [Source: Ivideon](#source-ivideon)
+    * [Source: Hass](#source-hass)
+  * [Module: API](#module-api)
+  * [Module: RTSP](#module-rtsp)
+  * [Module: WebRTC](#module-webrtc)
+  * [Module: Ngrok](#module-ngrok)
+  * [Module: Hass](#module-hass)
+    * [From go2rtc to Hass](#from-go2rtc-to-hass)
+    * [From Hass to go2rtc](#from-hass-to-go2rtc)
+  * [Module: MP4](#module-mp4)
+  * [Module: MJPEG](#module-mjpeg)
+  * [Module: Log](#module-log)
+* [Security](#security)
+* [Codecs madness](#codecs-madness)
+* [Codecs negotiation](#codecs-negotiation)
+* [TIPS](#tips)
+* [FAQ](#faq)
+
 ## Fast start
 
 1. Download [binary](#go2rtc-binary) or use [Docker](#go2rtc-docker) or [Home Assistant Add-on](#go2rtc-home-assistant-add-on)
@@ -36,7 +70,6 @@ Ultimate camera streaming application with support RTSP, WebRTC, HomeKit, FFmpeg
 
 - add your [streams](#module-streams) to [config](#configuration) file
 - setup [external access](#module-webrtc) to webrtc
-- setup [external access](#module-ngrok) to web interface
 
 **Developers:**
 
@@ -74,14 +107,14 @@ Container [alexxit/go2rtc](https://hub.docker.com/r/alexxit/go2rtc) with support
 
 ## Configuration
 
-Create file `go2rtc.yaml`. go2rtc will search this file in current work dirrectory by default.
-
-- by default, you need to config only your `streams` links
+- by default go2rtc will search `go2rtc.yaml` in the current work dirrectory
 - `api` server will start on default **1984 port** (TCP)
 - `rtsp` server will start on default **8554 port** (TCP)
 - `webrtc` will use port **8555** (TCP/UDP) for connections
 - `ffmpeg` will use default transcoding options
 
+Configuration options and a complete list of settings can be found in [the wiki](https://github.com/AlexxIT/go2rtc/wiki/Configuration).
+
 Available modules:
 
 - [streams](#module-streams)
@@ -95,8 +128,6 @@ Available modules:
 - [hass](#module-hass) - Home Assistant integration
 - [log](#module-log) - logs config
 
-Full default config [example](https://github.com/AlexxIT/go2rtc/wiki/Configuration).
-
 ### Module: Streams
 
 **go2rtc** support different stream source types. You can config one or multiple links of any type as stream source.
@@ -359,6 +390,8 @@ go2rtc has simple HTML page (`stream.html`) with support params in URL:
 ```yaml
 api:
   listen: ":1984"    # default ":1984", HTTP API port ("" - disabled)
+  username: "admin"  # default "", Basic auth for WebUI
+  password: "pass"   # default "", Basic auth for WebUI
   base_path: "/rtc"  # default "", API prefix for serve on suburl (/api => /rtc/api)
   static_dir: "www"  # default "", folder for static files (custom web interface)
   origin: "*"        # default "", allow CORS requests (only * supported)
@@ -366,7 +399,7 @@ api:
 
 **PS:**
 
-- go2rtc doesn't provide HTTPS or password protection. Use [Nginx](https://nginx.org/) or [Ngrok](#module-ngrok) or [Home Assistant Add-on](#go2rtc-home-assistant-add-on) for this tasks
+- go2rtc doesn't provide HTTPS. Use [Nginx](https://nginx.org/) or [Ngrok](#module-ngrok) or [Home Assistant Add-on](#go2rtc-home-assistant-add-on) for this tasks
 - you can access microphone (for 2-way audio) only with HTTPS ([read more](https://stackoverflow.com/questions/52759992/how-to-access-camera-and-microphone-in-chrome-without-https))
 - MJPEG over WebSocket plays better than native MJPEG because Chrome [bug](https://bugs.chromium.org/p/chromium/issues/detail?id=527446)
 - MP4 over WebSocket was created only for Apple iOS because it doesn't support MSE and native MP4
@@ -384,9 +417,9 @@ Password protection always disabled for localhost calls (ex. FFmpeg or Hass on s
 
 ```yaml
 rtsp:
-  listen: ":8554"  # RTSP Server TCP port, default - 8554
-  username: admin  # optional, default - disabled
-  password: pass   # optional, default - disabled
+  listen: ":8554"    # RTSP Server TCP port, default - 8554
+  username: "admin"  # optional, default - disabled
+  password: "pass"   # optional, default - disabled
 ```
 
 ### Module: WebRTC
@@ -399,7 +432,7 @@ WebRTC usually works without problems in the local network. But external access
 
 ```yaml
 webrtc:
-  listen: ":8555" # address of your local server and port (TCP/UDP)
+  listen: ":8555"  # address of your local server and port (TCP/UDP)
 ```
 
 **Static public IP**

cmd/api/ws.go

@@ -81,7 +81,9 @@ func apiWS(w http.ResponseWriter, r *http.Request) {
 	for {
 		msg := new(Message)
 		if err = ws.ReadJSON(msg); err != nil {
-			log.Trace().Err(err).Caller().Send()
+			if !websocket.IsCloseError(err, websocket.CloseNoStatusReceived) {
+				log.Trace().Err(err).Caller().Send()
+			}
 			_ = ws.Close()
 			break
 		}

cmd/app/app.go

@@ -14,7 +14,7 @@ import (
 	"time"
 )
 
-var Version = "1.0.0"
+var Version = "1.1.0"
 var UserAgent = "go2rtc/" + Version
 
 var ConfigPath string

cmd/ffmpeg/device/device_darwin.go

@@ -15,11 +15,11 @@ func deviceInputSuffix(videoIdx, audioIdx int) string {
 	audio := findMedia(streamer.KindAudio, audioIdx)
 	switch {
 	case video != nil && audio != nil:
-		return `"` + video.Title + `:` + audio.Title + `"`
+		return `"` + video.MID + `:` + audio.MID + `"`
 	case video != nil:
-		return `"` + video.Title + `"`
+		return `"` + video.MID + `"`
 	case audio != nil:
-		return `"` + audio.Title + `"`
+		return `"` + audio.MID + `"`
 	}
 	return ""
 }
@@ -57,7 +57,5 @@ process:
 }
 
 func loadMedia(kind, name string) *streamer.Media {
-	return &streamer.Media{
-		Kind: kind, Title: name,
-	}
+	return &streamer.Media{Kind: kind, MID: name}
 }

cmd/ffmpeg/device/device_linux.go

@@ -13,7 +13,7 @@ const deviceInputPrefix = "-f v4l2"
 
 func deviceInputSuffix(videoIdx, audioIdx int) string {
 	video := findMedia(streamer.KindVideo, videoIdx)
-	return video.Title
+	return video.MID
 }
 
 func loadMedias() {
@@ -44,7 +44,5 @@ func loadMedia(kind, name string) *streamer.Media {
 		return nil
 	}
 
-	return &streamer.Media{
-		Kind: kind, Title: name,
-	}
+	return &streamer.Media{Kind: kind, MID: name}
 }

cmd/ffmpeg/device/device_windows.go

@@ -15,11 +15,11 @@ func deviceInputSuffix(videoIdx, audioIdx int) string {
 	audio := findMedia(streamer.KindAudio, audioIdx)
 	switch {
 	case video != nil && audio != nil:
-		return `video="` + video.Title + `":audio=` + audio.Title + `"`
+		return `video="` + video.MID + `":audio=` + audio.MID + `"`
 	case video != nil:
-		return `video="` + video.Title + `"`
+		return `video="` + video.MID + `"`
 	case audio != nil:
-		return `audio="` + audio.Title + `"`
+		return `audio="` + audio.MID + `"`
 	}
 	return ""
 }
@@ -53,7 +53,5 @@ func loadMedias() {
 }
 
 func loadMedia(kind, name string) *streamer.Media {
-	return &streamer.Media{
-		Kind: kind, Title: name,
-	}
+	return &streamer.Media{Kind: kind, MID: name}
 }

cmd/ffmpeg/ffmpeg.go

@@ -67,6 +67,7 @@ var defaults = map[string]string{
 	"pcma/48000": "-c:a pcm_alaw -ar:a 48000 -ac:a 1",
 	"aac":        "-c:a aac", // keep sample rate and channels
 	"aac/16000":  "-c:a aac -ar:a 16000 -ac:a 1",
+	"mp3":        "-c:a libmp3lame -q:a 8",
 
 	// hardware Intel and AMD on Linux
 	// better not to set `-async_depth:v 1` like for QSV, because framedrops
@@ -141,6 +142,8 @@ func parseArgs(s string) *Args {
 			s += "?video"
 		case args.audio > 0 && args.video == 0:
 			s += "?audio"
+		default:
+			s += "?video&audio"
 		}
 		args.input = strings.Replace(defaults["rtsp"], "{input}", s, 1)
 	} else if strings.HasPrefix(s, "device?") {

cmd/hls/hls.go

@@ -0,0 +1,261 @@
+package hls
+
+import (
+	"fmt"
+	"github.com/AlexxIT/go2rtc/cmd/api"
+	"github.com/AlexxIT/go2rtc/cmd/streams"
+	"github.com/AlexxIT/go2rtc/pkg/mp4"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/AlexxIT/go2rtc/pkg/ts"
+	"github.com/rs/zerolog/log"
+	"net/http"
+	"strconv"
+	"sync"
+	"time"
+)
+
+func Init() {
+	api.HandleFunc("api/stream.m3u8", handlerStream)
+	api.HandleFunc("api/hls/playlist.m3u8", handlerPlaylist)
+
+	// HLS (TS)
+	api.HandleFunc("api/hls/segment.ts", handlerSegmentTS)
+
+	// HLS (fMP4)
+	api.HandleFunc("api/hls/init.mp4", handlerInit)
+	api.HandleFunc("api/hls/segment.m4s", handlerSegmentMP4)
+}
+
+type Consumer interface {
+	streamer.Consumer
+	Init() ([]byte, error)
+	MimeCodecs() string
+	Start()
+}
+
+type Session struct {
+	cons     Consumer
+	playlist string
+	init     []byte
+	segment  []byte
+	seq      int
+	alive    *time.Timer
+	mu       sync.Mutex
+}
+
+const keepalive = 5 * time.Second
+
+var sessions = map[string]*Session{}
+
+func handlerStream(w http.ResponseWriter, r *http.Request) {
+	// CORS important for Chromecast
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	w.Header().Set("Content-Type", "application/vnd.apple.mpegurl")
+
+	if r.Method == "OPTIONS" {
+		w.Header().Set("Access-Control-Allow-Methods", "GET")
+		return
+	}
+
+	src := r.URL.Query().Get("src")
+	stream := streams.GetOrNew(src)
+	if stream == nil {
+		http.Error(w, api.StreamNotFound, http.StatusNotFound)
+		return
+	}
+
+	var cons Consumer
+
+	// use fMP4 with codecs filter and TS without
+	medias := mp4.ParseQuery(r.URL.Query())
+	if medias != nil {
+		cons = &mp4.Consumer{
+			RemoteAddr: r.RemoteAddr,
+			UserAgent:  r.UserAgent(),
+			Medias:     medias,
+		}
+	} else {
+		cons = &ts.Consumer{
+			RemoteAddr: r.RemoteAddr,
+			UserAgent:  r.UserAgent(),
+		}
+	}
+
+	session := &Session{cons: cons}
+
+	cons.Listen(func(msg interface{}) {
+		if data, ok := msg.([]byte); ok {
+			session.mu.Lock()
+			session.segment = append(session.segment, data...)
+			session.mu.Unlock()
+		}
+	})
+
+	if err := stream.AddConsumer(cons); err != nil {
+		log.Error().Err(err).Caller().Send()
+		return
+	}
+
+	session.alive = time.AfterFunc(keepalive, func() {
+		stream.RemoveConsumer(cons)
+	})
+	session.init, _ = cons.Init()
+
+	cons.Start()
+
+	sid := strconv.FormatInt(time.Now().UnixNano(), 10)
+
+	// two segments important for Chromecast
+	if medias != nil {
+		session.playlist = `#EXTM3U
+#EXT-X-VERSION:6
+#EXT-X-TARGETDURATION:1
+#EXT-X-MEDIA-SEQUENCE:%d
+#EXT-X-MAP:URI="init.mp4?id=` + sid + `"
+#EXTINF:0.500,
+segment.m4s?id=` + sid + `&n=%d
+#EXTINF:0.500,
+segment.m4s?id=` + sid + `&n=%d`
+	} else {
+		session.playlist = `#EXTM3U
+#EXT-X-VERSION:3
+#EXT-X-TARGETDURATION:1
+#EXT-X-MEDIA-SEQUENCE:%d
+#EXTINF:0.500,
+segment.ts?id=` + sid + `&n=%d
+#EXTINF:0.500,
+segment.ts?id=` + sid + `&n=%d`
+	}
+
+	sessions[sid] = session
+
+	// bandwidth important for Safari, codecs useful for smooth playback
+	data := []byte(`#EXTM3U
+#EXT-X-STREAM-INF:BANDWIDTH=1000000,CODECS="` + cons.MimeCodecs() + `"
+hls/playlist.m3u8?id=` + sid)
+
+	if _, err := w.Write(data); err != nil {
+		log.Error().Err(err).Caller().Send()
+	}
+}
+
+func handlerPlaylist(w http.ResponseWriter, r *http.Request) {
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	w.Header().Set("Content-Type", "application/vnd.apple.mpegurl")
+
+	if r.Method == "OPTIONS" {
+		w.Header().Set("Access-Control-Allow-Methods", "GET")
+		return
+	}
+
+	sid := r.URL.Query().Get("id")
+	session := sessions[sid]
+	if session == nil {
+		http.NotFound(w, r)
+		return
+	}
+
+	s := fmt.Sprintf(session.playlist, session.seq, session.seq, session.seq+1)
+
+	if _, err := w.Write([]byte(s)); err != nil {
+		log.Error().Err(err).Caller().Send()
+	}
+}
+
+func handlerSegmentTS(w http.ResponseWriter, r *http.Request) {
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	w.Header().Set("Content-Type", "video/mp2t")
+
+	if r.Method == "OPTIONS" {
+		w.Header().Set("Access-Control-Allow-Methods", "GET")
+		return
+	}
+
+	sid := r.URL.Query().Get("id")
+	session := sessions[sid]
+	if session == nil {
+		http.NotFound(w, r)
+		return
+	}
+
+	session.alive.Reset(keepalive)
+
+	var i byte
+	for len(session.segment) == 0 {
+		if i++; i > 10 {
+			http.NotFound(w, r)
+			return
+		}
+		time.Sleep(time.Millisecond * 100)
+	}
+
+	session.mu.Lock()
+	data := session.segment
+	// important to start new segment with init
+	session.segment = session.init
+	session.seq++
+	session.mu.Unlock()
+
+	if _, err := w.Write(data); err != nil {
+		log.Error().Err(err).Caller().Send()
+	}
+}
+
+func handlerInit(w http.ResponseWriter, r *http.Request) {
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	w.Header().Add("Content-Type", "video/mp4")
+
+	if r.Method == "OPTIONS" {
+		w.Header().Set("Access-Control-Allow-Methods", "GET")
+		return
+	}
+
+	sid := r.URL.Query().Get("id")
+	session := sessions[sid]
+	if session == nil {
+		http.NotFound(w, r)
+		return
+	}
+
+	if _, err := w.Write(session.init); err != nil {
+		log.Error().Err(err).Caller().Send()
+	}
+}
+
+func handlerSegmentMP4(w http.ResponseWriter, r *http.Request) {
+	w.Header().Set("Access-Control-Allow-Origin", "*")
+	w.Header().Add("Content-Type", "video/iso.segment")
+
+	if r.Method == "OPTIONS" {
+		w.Header().Set("Access-Control-Allow-Methods", "GET, OPTIONS")
+		return
+	}
+
+	sid := r.URL.Query().Get("id")
+	session := sessions[sid]
+	if session == nil {
+		http.NotFound(w, r)
+		return
+	}
+
+	session.alive.Reset(keepalive)
+
+	var i byte
+	for len(session.segment) == 0 {
+		if i++; i > 10 {
+			http.NotFound(w, r)
+			return
+		}
+		time.Sleep(time.Millisecond * 100)
+	}
+
+	session.mu.Lock()
+	data := session.segment
+	session.segment = nil
+	session.seq++
+	session.mu.Unlock()
+
+	if _, err := w.Write(data); err != nil {
+		log.Error().Err(err).Caller().Send()
+	}
+}

cmd/mp4/mp4.go

@@ -5,6 +5,7 @@ import (
 	"github.com/AlexxIT/go2rtc/cmd/app"
 	"github.com/AlexxIT/go2rtc/cmd/streams"
 	"github.com/AlexxIT/go2rtc/pkg/mp4"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
 	"github.com/rs/zerolog"
 	"net/http"
 	"strconv"
@@ -25,8 +26,14 @@ func Init() {
 var log zerolog.Logger
 
 func handlerKeyframe(w http.ResponseWriter, r *http.Request) {
-	if isChromeFirst(w, r) {
-		return
+	// Chrome 105 does two requests: without Range and with `Range: bytes=0-`
+	ua := r.UserAgent()
+	if strings.Contains(ua, " Chrome/") {
+		if r.Header.Values("Range") == nil {
+			w.Header().Set("Content-Type", "video/mp4")
+			w.WriteHeader(http.StatusOK)
+			return
+		}
 	}
 
 	src := r.URL.Query().Get("src")
@@ -67,7 +74,22 @@ func handlerKeyframe(w http.ResponseWriter, r *http.Request) {
 func handlerMP4(w http.ResponseWriter, r *http.Request) {
 	log.Trace().Msgf("[mp4] %s %+v", r.Method, r.Header)
 
-	if isChromeFirst(w, r) || isSafari(w, r) {
+	// Chrome has Safari in UA, so check first Chrome and later Safari
+	ua := r.UserAgent()
+	if strings.Contains(ua, " Chrome/") {
+		if r.Header.Values("Range") == nil {
+			w.Header().Set("Content-Type", "video/mp4")
+			w.WriteHeader(http.StatusOK)
+			return
+		}
+	} else if strings.Contains(ua, " Safari/") {
+		// auto redirect to HLS/fMP4 format, because Safari not support MP4 stream
+		url := "stream.m3u8?" + r.URL.RawQuery
+		if !r.URL.Query().Has("mp4") {
+			url += "&mp4"
+		}
+
+		http.Redirect(w, r, url, http.StatusMovedPermanently)
 		return
 	}
 
@@ -83,7 +105,9 @@ func handlerMP4(w http.ResponseWriter, r *http.Request) {
 	cons := &mp4.Consumer{
 		RemoteAddr: r.RemoteAddr,
 		UserAgent:  r.UserAgent(),
+		Medias:     streamer.ParseQuery(r.URL.Query()),
 	}
+
 	cons.Listen(func(msg interface{}) {
 		if data, ok := msg.([]byte); ok {
 			if _, err := w.Write(data); err != nil && exit != nil {
@@ -135,23 +159,3 @@ func handlerMP4(w http.ResponseWriter, r *http.Request) {
 		duration.Stop()
 	}
 }
-
-func isChromeFirst(w http.ResponseWriter, r *http.Request) bool {
-	// Chrome 105 does two requests: without Range and with `Range: bytes=0-`
-	if strings.Contains(r.UserAgent(), " Chrome/") {
-		if r.Header.Values("Range") == nil {
-			w.Header().Set("Content-Type", "video/mp4")
-			w.WriteHeader(http.StatusOK)
-			return true
-		}
-	}
-	return false
-}
-
-func isSafari(w http.ResponseWriter, r *http.Request) bool {
-	if r.Header.Get("Range") == "bytes=0-1" {
-		handlerKeyframe(w, r)
-		return true
-	}
-	return false
-}

cmd/mp4/ws.go

@@ -22,8 +22,6 @@ func handlerWSMSE(tr *api.Transport, msg *api.Message) error {
 		RemoteAddr: tr.Request.RemoteAddr,
 		UserAgent:  tr.Request.UserAgent(),
 	}
-	cons.UserAgent = tr.Request.UserAgent()
-	cons.RemoteAddr = tr.Request.RemoteAddr
 
 	if codecs, ok := msg.Value.(string); ok {
 		log.Trace().Str("codecs", codecs).Msgf("[mp4] new WS/MSE consumer")
@@ -108,15 +106,18 @@ func parseMedias(codecs string, parseAudio bool) (medias []*streamer.Media) {
 
 	for _, name := range strings.Split(codecs, ",") {
 		switch name {
-		case "avc1.640029":
+		case mp4.MimeH264:
 			codec := &streamer.Codec{Name: streamer.CodecH264}
 			videos = append(videos, codec)
-		case "hvc1.1.6.L153.B0":
+		case mp4.MimeH265:
 			codec := &streamer.Codec{Name: streamer.CodecH265}
 			videos = append(videos, codec)
-		case "mp4a.40.2":
+		case mp4.MimeAAC:
 			codec := &streamer.Codec{Name: streamer.CodecAAC}
 			audios = append(audios, codec)
+		case mp4.MimeOpus:
+			codec := &streamer.Codec{Name: streamer.CodecOpus}
+			audios = append(audios, codec)
 		}
 	}
 

cmd/rtsp/rtsp.go

@@ -3,25 +3,29 @@ package rtsp
 import (
 	"github.com/AlexxIT/go2rtc/cmd/app"
 	"github.com/AlexxIT/go2rtc/cmd/streams"
+	"github.com/AlexxIT/go2rtc/pkg/mp4"
 	"github.com/AlexxIT/go2rtc/pkg/rtsp"
 	"github.com/AlexxIT/go2rtc/pkg/streamer"
 	"github.com/AlexxIT/go2rtc/pkg/tcp"
 	"github.com/rs/zerolog"
 	"net"
+	"net/url"
 	"strings"
 )
 
 func Init() {
 	var conf struct {
 		Mod struct {
-			Listen   string `yaml:"listen" json:"listen"`
-			Username string `yaml:"username" json:"-"`
-			Password string `yaml:"password" json:"-"`
+			Listen       string `yaml:"listen" json:"listen"`
+			Username     string `yaml:"username" json:"-"`
+			Password     string `yaml:"password" json:"-"`
+			DefaultQuery string `yaml:"default_query"`
 		} `yaml:"rtsp"`
 	}
 
 	// default config
 	conf.Mod.Listen = ":8554"
+	conf.Mod.DefaultQuery = "video&audio"
 
 	app.LoadConfig(&conf)
 	app.Info["rtsp"] = conf.Mod
@@ -49,6 +53,10 @@ func Init() {
 
 	log.Info().Str("addr", address).Msg("[rtsp] listen")
 
+	if query, err := url.ParseQuery(conf.Mod.DefaultQuery); err == nil {
+		defaultMedias = mp4.ParseQuery(query)
+	}
+
 	go func() {
 		for {
 			conn, err := ln.Accept()
@@ -78,6 +86,7 @@ var Port string
 
 var log zerolog.Logger
 var handlers []Handler
+var defaultMedias []*streamer.Media
 
 func rtspHandler(url string) (streamer.Producer, error) {
 	backchannel := true
@@ -164,7 +173,10 @@ func tcpHandler(conn *rtsp.Conn) {
 
 			conn.SessionName = app.UserAgent
 
-			initMedias(conn)
+			conn.Medias = mp4.ParseQuery(conn.URL.Query())
+			if conn.Medias == nil {
+				conn.Medias = defaultMedias
+			}
 
 			if err := stream.AddConsumer(conn); err != nil {
 				log.Warn().Err(err).Str("stream", name).Msg("[rtsp]")
@@ -228,45 +240,3 @@ func tcpHandler(conn *rtsp.Conn) {
 
 	_ = conn.Close()
 }
-
-func initMedias(conn *rtsp.Conn) {
-	// set media candidates from query list
-	for key, value := range conn.URL.Query() {
-		switch key {
-		case streamer.KindVideo, streamer.KindAudio:
-			for _, name := range value {
-				name = strings.ToUpper(name)
-
-				// check aliases
-				switch name {
-				case "COPY":
-					name = "" // pass empty codecs list
-				case "MJPEG":
-					name = streamer.CodecJPEG
-				case "AAC":
-					name = streamer.CodecAAC
-				}
-
-				media := &streamer.Media{
-					Kind: key, Direction: streamer.DirectionRecvonly,
-				}
-
-				// empty codecs match all codecs
-				if name != "" {
-					// empty clock rate and channels match any values
-					media.Codecs = []*streamer.Codec{{Name: name}}
-				}
-
-				conn.Medias = append(conn.Medias, media)
-			}
-		}
-	}
-
-	// set default media candidates if query is empty
-	if conn.Medias == nil {
-		conn.Medias = []*streamer.Media{
-			{Kind: streamer.KindVideo, Direction: streamer.DirectionRecvonly},
-			{Kind: streamer.KindAudio, Direction: streamer.DirectionRecvonly},
-		}
-	}
-}

cmd/streams/stream.go

@@ -91,7 +91,9 @@ func (s *Stream) AddConsumer(cons streamer.Consumer) (err error) {
 
 					consumer.tracks = append(consumer.tracks, consTrack)
 					producers = append(producers, prod)
-					break producers
+					if !consMedia.MatchAll() {
+						break producers
+					}
 				}
 			}
 		}

main.go

@@ -8,6 +8,7 @@ import (
 	"github.com/AlexxIT/go2rtc/cmd/exec"
 	"github.com/AlexxIT/go2rtc/cmd/ffmpeg"
 	"github.com/AlexxIT/go2rtc/cmd/hass"
+	"github.com/AlexxIT/go2rtc/cmd/hls"
 	"github.com/AlexxIT/go2rtc/cmd/homekit"
 	"github.com/AlexxIT/go2rtc/cmd/http"
 	"github.com/AlexxIT/go2rtc/cmd/ivideon"
@@ -26,27 +27,25 @@ import (
 
 func main() {
 	app.Init()     // init config and logs
+	api.Init()     // init HTTP API server
 	streams.Init() // load streams list
 
-	api.Init() // init HTTP API server
-
-	echo.Init()
-
 	rtsp.Init()   // add support RTSP client and RTSP server
 	rtmp.Init()   // add support RTMP client
 	exec.Init()   // add support exec scheme (depends on RTSP server)
 	ffmpeg.Init() // add support ffmpeg scheme (depends on exec scheme)
 	hass.Init()   // add support hass scheme
-
-	webrtc.Init()
-	mp4.Init()
-	mjpeg.Init()
-	http.Init()
+	echo.Init()
+	ivideon.Init()
 
 	srtp.Init()
 	homekit.Init()
 
-	ivideon.Init()
+	webrtc.Init()
+	mp4.Init()
+	hls.Init()
+	mjpeg.Init()
+	http.Init()
 
 	ngrok.Init()
 	debug.Init()

pkg/aac/rtp.go

@@ -17,9 +17,14 @@ func RTPDepay(track *streamer.Track) streamer.WrapperFunc {
 
 			//log.Printf("[RTP/AAC] units: %d, size: %4d, ts: %10d, %t", headersSize/2, len(packet.Payload), packet.Timestamp, packet.Marker)
 
+			data := packet.Payload[2+headersSize:]
+			if IsADTS(data) {
+				data = data[7:]
+			}
+
 			clone := *packet
 			clone.Version = RTPPacketVersionAAC
-			clone.Payload = packet.Payload[2+headersSize:]
+			clone.Payload = data
 			return push(&clone)
 		}
 	}
@@ -55,3 +60,7 @@ func RTPPay(mtu uint16) streamer.WrapperFunc {
 		}
 	}
 }
+
+func IsADTS(b []byte) bool {
+	return len(b) > 7 && b[0] == 0xFF && b[1]&0xF0 == 0xF0
+}

pkg/h264/rtp.go

@@ -1,6 +1,7 @@
 package h264
 
 import (
+	"bytes"
 	"encoding/binary"
 	"github.com/AlexxIT/go2rtc/pkg/streamer"
 	"github.com/pion/rtp"
@@ -27,7 +28,8 @@ func RTPDepay(track *streamer.Track) streamer.WrapperFunc {
 			}
 
 			// Fix TP-Link Tapo TC70: sends SPS and PPS with packet.Marker = true
-			if packet.Marker {
+			// Reolink Duo 2: sends SPS with Marker and PPS without
+			if packet.Marker && len(payload) < 128 {
 				switch NALUType(payload) {
 				case NALUTypeSPS, NALUTypePPS:
 					buf = append(buf, payload...)
@@ -68,6 +70,27 @@ func RTPDepay(track *streamer.Track) streamer.WrapperFunc {
 			if len(buf) > 0 {
 				payload = append(buf, payload...)
 				buf = buf[:0]
+			} else {
+				// some Chinese buggy cameras has single packet with SPS+PPS+IFrame separated by 00 00 00 01
+				// https://github.com/AlexxIT/WebRTC/issues/391
+				// https://github.com/AlexxIT/WebRTC/issues/392
+				for i := 0; i < len(payload); {
+					if i+4 >= len(payload) {
+						break
+					}
+
+					size := bytes.Index(payload[i+4:], []byte{0, 0, 0, 1})
+					if size < 0 {
+						if i == 0 {
+							break
+						}
+						size = len(payload) - (i + 4)
+					}
+
+					binary.BigEndian.PutUint32(payload[i:], uint32(size))
+
+					i += size + 4
+				}
 			}
 
 			//log.Printf("[AVC] %v, len: %d, ts: %10d, seq: %d", Types(payload), len(payload), packet.Timestamp, packet.SequenceNumber)

pkg/httpflv/flvio.go

@@ -0,0 +1,97 @@
+package httpflv
+
+import (
+	"fmt"
+	"github.com/deepch/vdk/format/flv/flvio"
+	"github.com/deepch/vdk/utils/bits/pio"
+	"io"
+)
+
+// TODO: rewrite all of this someday
+
+func ReadTag(r io.Reader, b []byte) (tag flvio.Tag, ts int32, err error) {
+	if _, err = io.ReadFull(r, b[:flvio.TagHeaderLength]); err != nil {
+		return
+	}
+	var datalen int
+	if tag, ts, datalen, err = flvio.ParseTagHeader(b); err != nil {
+		return
+	}
+
+	data := make([]byte, datalen)
+	if _, err = io.ReadFull(r, data); err != nil {
+		return
+	}
+
+	n, err := ParseHeader(&tag, data)
+	if err != nil {
+		return
+	}
+	tag.Data = data[n:]
+
+	if _, err = io.ReadFull(r, b[:4]); err != nil {
+		return
+	}
+	return
+}
+
+func ParseHeader(self *flvio.Tag, b []byte) (n int, err error) {
+	switch self.Type {
+	case flvio.TAG_AUDIO:
+		return audioParseHeader(self, b)
+
+	case flvio.TAG_VIDEO:
+		return videoParseHeader(self, b)
+	}
+
+	return
+}
+
+func audioParseHeader(tag *flvio.Tag, b []byte) (n int, err error) {
+	if len(b) < n+1 {
+		err = fmt.Errorf("audiodata: parse invalid")
+		return
+	}
+
+	flags := b[n]
+	n++
+	tag.SoundFormat = flags >> 4
+	tag.SoundRate = (flags >> 2) & 0x3
+	tag.SoundSize = (flags >> 1) & 0x1
+	tag.SoundType = flags & 0x1
+
+	switch tag.SoundFormat {
+	case flvio.SOUND_AAC:
+		if len(b) < n+1 {
+			err = fmt.Errorf("audiodata: parse invalid")
+			return
+		}
+		tag.AACPacketType = b[n]
+		n++
+	}
+
+	return
+}
+
+func videoParseHeader(tag *flvio.Tag, b []byte) (n int, err error) {
+	if len(b) < n+1 {
+		err = fmt.Errorf("videodata: parse invalid")
+		return
+	}
+	flags := b[n]
+	tag.FrameType = flags >> 4
+	tag.CodecID = flags & 0xf
+	n++
+
+	if len(b) < n+4 {
+		err = fmt.Errorf("videodata: parse invalid")
+		return
+	}
+	tag.AVCPacketType = b[n]
+	n++
+
+	tag.CompositionTime = pio.I24BE(b[n:])
+	n += 3
+
+	return
+}

pkg/httpflv/httpflv.go

@@ -128,8 +128,10 @@ func (c *Conn) Streams() ([]av.CodecData, error) {
 		return []av.CodecData{video, audio}, nil
 	} else if video != nil {
 		c.videoIdx = 0
+		c.audioIdx = -1
 		return []av.CodecData{video}, nil
 	} else if audio != nil {
+		c.videoIdx = -1
 		c.audioIdx = 0
 		return []av.CodecData{audio}, nil
 	}
@@ -139,17 +141,19 @@ func (c *Conn) Streams() ([]av.CodecData, error) {
 
 func (c *Conn) ReadPacket() (av.Packet, error) {
 	for {
-		tag, ts, err := flvio.ReadTag(c.reader, c.buf)
+		tag, ts, err := ReadTag(c.reader, c.buf)
 		if err != nil {
 			return av.Packet{}, err
 		}
 
 		switch tag.Type {
 		case flvio.TAG_VIDEO:
-			if tag.AVCPacketType != flvio.AVC_NALU {
+			if c.videoIdx < 0 || tag.AVCPacketType != flvio.AVC_NALU {
 				continue
 			}
 
+			//log.Printf("[FLV] %v, len: %d, ts: %10d", h264.Types(tag.Data), len(tag.Data), flvio.TsToTime(ts))
+
 			return av.Packet{
 				Idx:             c.videoIdx,
 				Data:            tag.Data,
@@ -159,7 +163,7 @@ func (c *Conn) ReadPacket() (av.Packet, error) {
 			}, nil
 
 		case flvio.TAG_AUDIO:
-			if tag.SoundFormat != flvio.SOUND_AAC || tag.AACPacketType != flvio.AAC_RAW {
+			if c.audioIdx < 0 || tag.SoundFormat != flvio.SOUND_AAC || tag.AACPacketType != flvio.AAC_RAW {
 				continue
 			}
 

pkg/iso/atoms.go

@@ -0,0 +1,318 @@
+package iso
+
+const (
+	Ftyp                        = "ftyp"
+	Moov                        = "moov"
+	MoovMvhd                    = "mvhd"
+	MoovTrak                    = "trak"
+	MoovTrakTkhd                = "tkhd"
+	MoovTrakMdia                = "mdia"
+	MoovTrakMdiaMdhd            = "mdhd"
+	MoovTrakMdiaHdlr            = "hdlr"
+	MoovTrakMdiaMinf            = "minf"
+	MoovTrakMdiaMinfVmhd        = "vmhd"
+	MoovTrakMdiaMinfSmhd        = "smhd"
+	MoovTrakMdiaMinfDinf        = "dinf"
+	MoovTrakMdiaMinfDinfDref    = "dref"
+	MoovTrakMdiaMinfDinfDrefUrl = "url "
+	MoovTrakMdiaMinfStbl        = "stbl"
+	MoovTrakMdiaMinfStblStsd    = "stsd"
+	MoovTrakMdiaMinfStblStts    = "stts"
+	MoovTrakMdiaMinfStblStsc    = "stsc"
+	MoovTrakMdiaMinfStblStsz    = "stsz"
+	MoovTrakMdiaMinfStblStco    = "stco"
+	MoovMvex                    = "mvex"
+	MoovMvexTrex                = "trex"
+	Moof                        = "moof"
+	MoofMfhd                    = "mfhd"
+	MoofTraf                    = "traf"
+	MoofTrafTfhd                = "tfhd"
+	MoofTrafTfdt                = "tfdt"
+	MoofTrafTrun                = "trun"
+	Mdat                        = "mdat"
+)
+
+func (m *Movie) WriteFileType() {
+	m.StartAtom(Ftyp)
+	m.WriteString("iso5")
+	m.WriteUint32(512)
+	m.WriteString("iso5")
+	m.WriteString("iso6")
+	m.WriteString("mp41")
+	m.EndAtom()
+}
+
+func (m *Movie) WriteMovieHeader() {
+	m.StartAtom(MoovMvhd)
+	m.Skip(1)           // version
+	m.Skip(3)           // flags
+	m.Skip(4)           // create time
+	m.Skip(4)           // modify time
+	m.WriteUint32(1000) // time scale
+	m.Skip(4)           // duration
+	m.WriteFloat32(1)   // preferred rate
+	m.WriteFloat16(1)   // preferred volume
+	m.Skip(10)          // reserved
+	m.WriteMatrix()
+	m.Skip(6 * 4)             // predefined?
+	m.WriteUint32(0xFFFFFFFF) // next track ID
+	m.EndAtom()
+}
+
+func (m *Movie) WriteTrackHeader(id uint32, width, height uint16) {
+	const (
+		TkhdTrackEnabled   = 0x0001
+		TkhdTrackInMovie   = 0x0002
+		TkhdTrackInPreview = 0x0004
+		TkhdTrackInPoster  = 0x0008
+	)
+
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-32963
+	m.StartAtom(MoovTrakTkhd)
+	m.Skip(1) // version
+	m.WriteUint24(TkhdTrackEnabled | TkhdTrackInMovie)
+	m.Skip(4)         // create time
+	m.Skip(4)         // modify time
+	m.WriteUint32(id) // trackID
+	m.Skip(4)         // reserved
+	m.Skip(4)         // duration
+	m.Skip(8)         // reserved
+	m.Skip(2)         // layer
+	if width > 0 {
+		m.Skip(2)
+		m.Skip(2)
+	} else {
+		m.WriteUint16(1)  // alternate group
+		m.WriteFloat16(1) // volume
+	}
+	m.Skip(2) // reserved
+	m.WriteMatrix()
+	if width > 0 {
+		m.WriteFloat32(float64(width))
+		m.WriteFloat32(float64(height))
+	} else {
+		m.Skip(4)
+		m.Skip(4)
+	}
+	m.EndAtom()
+}
+
+func (m *Movie) WriteMediaHeader(timescale uint32) {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-32999
+	m.StartAtom(MoovTrakMdiaMdhd)
+	m.Skip(1)                // version
+	m.Skip(3)                // flags
+	m.Skip(4)                // creation time
+	m.Skip(4)                // modification time
+	m.WriteUint32(timescale) // timescale
+	m.Skip(4)                // duration
+	m.WriteUint16(0x55C4)    // language (Unspecified)
+	m.Skip(2)                // quality
+	m.EndAtom()
+}
+
+func (m *Movie) WriteMediaHandler(s, name string) {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-33004
+	m.StartAtom(MoovTrakMdiaHdlr)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4)
+	m.WriteString(s)    // handler type (4 byte!)
+	m.Skip(3 * 4)       // reserved
+	m.WriteString(name) // handler name (any len)
+	m.Skip(1)           // end string
+	m.EndAtom()
+}
+
+func (m *Movie) WriteVideoMediaInfo() {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-33012
+	m.StartAtom(MoovTrakMdiaMinfVmhd)
+	m.Skip(1)        // version
+	m.WriteUint24(1) // flags (You should always set this flag to 1)
+	m.Skip(2)        // graphics mode
+	m.Skip(3 * 2)    // op color
+	m.EndAtom()
+}
+
+func (m *Movie) WriteAudioMediaInfo() {
+	m.StartAtom(MoovTrakMdiaMinfSmhd)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4) // balance
+	m.EndAtom()
+}
+
+func (m *Movie) WriteDataInfo() {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-25680
+	m.StartAtom(MoovTrakMdiaMinfDinf)
+	m.StartAtom(MoovTrakMdiaMinfDinfDref)
+	m.Skip(1)        // version
+	m.Skip(3)        // flags
+	m.WriteUint32(1) // childrens
+
+	m.StartAtom(MoovTrakMdiaMinfDinfDrefUrl)
+	m.Skip(1)        // version
+	m.WriteUint24(1) // flags (self reference)
+	m.EndAtom()
+
+	m.EndAtom() // DREF
+	m.EndAtom() // DINF
+}
+
+func (m *Movie) WriteSampleTable(writeSampleDesc func()) {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap2/qtff2.html#//apple_ref/doc/uid/TP40000939-CH204-33040
+	m.StartAtom(MoovTrakMdiaMinfStbl)
+
+	m.StartAtom(MoovTrakMdiaMinfStblStsd)
+	m.Skip(1)        // version
+	m.Skip(3)        // flags
+	m.WriteUint32(1) // entry count
+	writeSampleDesc()
+	m.EndAtom()
+
+	m.StartAtom(MoovTrakMdiaMinfStblStts)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4) // entry count
+	m.EndAtom()
+
+	m.StartAtom(MoovTrakMdiaMinfStblStsc)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4) // entry count
+	m.EndAtom()
+
+	m.StartAtom(MoovTrakMdiaMinfStblStsz)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4) // sample size
+	m.Skip(4) // entry count
+	m.EndAtom()
+
+	m.StartAtom(MoovTrakMdiaMinfStblStco)
+	m.Skip(1) // version
+	m.Skip(3) // flags
+	m.Skip(4) // entry count
+	m.EndAtom()
+
+	m.EndAtom()
+}
+
+func (m *Movie) WriteTrackExtend(id uint32) {
+	m.StartAtom(MoovMvexTrex)
+	m.Skip(1)         // version
+	m.Skip(3)         // flags
+	m.WriteUint32(id) // trackID
+	m.WriteUint32(1)  // default sample description index
+	m.Skip(4)         // default sample duration
+	m.Skip(4)         // default sample size
+	m.Skip(4)         // default sample flags
+	m.EndAtom()
+}
+
+func (m *Movie) WriteVideoTrack(id uint32, codec string, timescale uint32, width, height uint16, conf []byte) {
+	m.StartAtom(MoovTrak)
+	m.WriteTrackHeader(id, width, height)
+
+	m.StartAtom(MoovTrakMdia)
+	m.WriteMediaHeader(timescale)
+	m.WriteMediaHandler("vide", "VideoHandler")
+
+	m.StartAtom(MoovTrakMdiaMinf)
+	m.WriteVideoMediaInfo()
+	m.WriteDataInfo()
+	m.WriteSampleTable(func() {
+		m.WriteVideo(codec, width, height, conf)
+	})
+	m.EndAtom() // MINF
+
+	m.EndAtom() // MDIA
+	m.EndAtom() // TRAK
+}
+
+func (m *Movie) WriteAudioTrack(id uint32, codec string, timescale uint32, channels uint16, conf []byte) {
+	m.StartAtom(MoovTrak)
+	m.WriteTrackHeader(id, 0, 0)
+
+	m.StartAtom(MoovTrakMdia)
+	m.WriteMediaHeader(timescale)
+	m.WriteMediaHandler("soun", "SoundHandler")
+
+	m.StartAtom(MoovTrakMdiaMinf)
+	m.WriteAudioMediaInfo()
+	m.WriteDataInfo()
+	m.WriteSampleTable(func() {
+		m.WriteAudio(codec, channels, timescale, conf)
+	})
+	m.EndAtom() // MINF
+
+	m.EndAtom() // MDIA
+	m.EndAtom() // TRAK
+}
+
+func (m *Movie) WriteMovieFragment(seq, tid, duration, size uint32, time uint64) {
+	m.StartAtom(Moof)
+
+	m.StartAtom(MoofMfhd)
+	m.Skip(1)          // version
+	m.Skip(3)          // flags
+	m.WriteUint32(seq) // sequence number
+	m.EndAtom()
+
+	m.StartAtom(MoofTraf)
+
+	const (
+		TfhdDefaultSampleDuration = 0x000008
+		TfhdDefaultSampleSize     = 0x000010
+		TfhdDefaultSampleFlags    = 0x000020
+		TfhdDefaultBaseIsMoof     = 0x020000
+	)
+
+	m.StartAtom(MoofTrafTfhd)
+	m.Skip(1) // version
+	m.WriteUint24(
+		TfhdDefaultSampleDuration |
+			TfhdDefaultSampleSize |
+			TfhdDefaultSampleFlags |
+			TfhdDefaultBaseIsMoof,
+	)
+	m.WriteUint32(tid)       // track id
+	m.WriteUint32(duration)  // default sample duration
+	m.WriteUint32(size)      // default sample size
+	m.WriteUint32(0x2000000) // default sample flags
+	m.EndAtom()
+
+	m.StartAtom(MoofTrafTfdt)
+	m.WriteBytes(1)     // version
+	m.Skip(3)           // flags
+	m.WriteUint64(time) // base media decode time
+	m.EndAtom()
+
+	const (
+		TrunDataOffset       = 0x000001
+		TrunFirstSampleFlags = 0x000004
+		TrunSampleDuration   = 0x0000100
+		TrunSampleSize       = 0x0000200
+		TrunSampleFlags      = 0x0000400
+		TrunSampleCTS        = 0x0000800
+	)
+
+	m.StartAtom(MoofTrafTrun)
+	m.Skip(1)                     // version
+	m.WriteUint24(TrunDataOffset) // flags
+	m.WriteUint32(1)              // sample count
+	// data offset: current pos + uint32 len + MDAT header len
+	m.WriteUint32(uint32(len(m.b)) + 4 + 8)
+	m.EndAtom() // TRUN
+
+	m.EndAtom() // TRAF
+
+	m.EndAtom() // MOOF
+}
+
+func (m *Movie) WriteData(b []byte) {
+	m.StartAtom(Mdat)
+	m.Write(b)
+	m.EndAtom()
+
+}

pkg/iso/codecs.go

@@ -0,0 +1,151 @@
+package iso
+
+import "github.com/AlexxIT/go2rtc/pkg/streamer"
+
+func (m *Movie) WriteVideo(codec string, width, height uint16, conf []byte) {
+	// https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/QTFFChap3/qtff3.html
+	switch codec {
+	case streamer.CodecH264:
+		m.StartAtom("avc1")
+	case streamer.CodecH265:
+		m.StartAtom("hev1")
+	default:
+		panic("unsupported iso video: " + codec)
+	}
+	m.Skip(6)
+	m.WriteUint16(1)      // data_reference_index
+	m.Skip(2)             // version
+	m.Skip(2)             // revision
+	m.Skip(4)             // vendor
+	m.Skip(4)             // temporal quality
+	m.Skip(4)             // spatial quality
+	m.WriteUint16(width)  // width
+	m.WriteUint16(height) // height
+	m.WriteFloat32(72)    // horizontal resolution
+	m.WriteFloat32(72)    // vertical resolution
+	m.Skip(4)             // reserved
+	m.WriteUint16(1)      // frame count
+	m.Skip(32)            // compressor name
+	m.WriteUint16(24)     // depth
+	m.WriteUint16(0xFFFF) // color table id (-1)
+
+	switch codec {
+	case streamer.CodecH264:
+		m.StartAtom("avcC")
+	case streamer.CodecH265:
+		m.StartAtom("hvcC")
+	}
+	m.Write(conf)
+	m.EndAtom() // AVCC
+
+	m.EndAtom() // AVC1
+}
+
+func (m *Movie) WriteAudio(codec string, channels uint16, sampleRate uint32, conf []byte) {
+	switch codec {
+	case streamer.CodecAAC, streamer.CodecMP3:
+		m.StartAtom("mp4a")
+	case streamer.CodecOpus:
+		m.StartAtom("Opus")
+	case streamer.CodecPCMU:
+		m.StartAtom("ulaw")
+	case streamer.CodecPCMA:
+		m.StartAtom("alaw")
+	default:
+		panic("unsupported iso audio: " + codec)
+	}
+	m.Skip(6)
+	m.WriteUint16(1)                    // data_reference_index
+	m.Skip(2)                           // version
+	m.Skip(2)                           // revision
+	m.Skip(4)                           // vendor
+	m.WriteUint16(channels)             // channel_count
+	m.WriteUint16(16)                   // sample_size
+	m.Skip(2)                           // compression id
+	m.Skip(2)                           // reserved
+	m.WriteFloat32(float64(sampleRate)) // sample_rate
+
+	switch codec {
+	case streamer.CodecAAC:
+		m.WriteEsdsAAC(conf)
+	case streamer.CodecMP3:
+		m.WriteEsdsMP3()
+	case streamer.CodecOpus:
+		// don't know what means this magic
+		m.StartAtom("dOps")
+		m.WriteBytes(0, 0x02, 0x01, 0x38, 0, 0, 0xBB, 0x80, 0, 0, 0)
+		m.EndAtom()
+	case streamer.CodecPCMU, streamer.CodecPCMA:
+		// don't know what means this magic
+		m.StartAtom("chan")
+		m.WriteBytes(0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 4, 0, 0, 0, 0)
+		m.EndAtom()
+	}
+
+	m.EndAtom() // MP4A/OPUS
+}
+
+func (m *Movie) WriteEsdsAAC(conf []byte) {
+	m.StartAtom("esds")
+	m.Skip(1) // version
+	m.Skip(3) // flags
+
+	// MP4ESDescrTag[3]:
+	// - MP4DecConfigDescrTag[4]:
+	//   - MP4DecSpecificDescrTag[5]: conf
+	// - Other[6]
+	const header = 5
+	const size3 = 3
+	const size4 = 13
+	size5 := byte(len(conf))
+	const size6 = 1
+
+	m.WriteBytes(3, 0x80, 0x80, 0x80, size3+header+size4+header+size5+header+size6)
+	m.Skip(2) // es id
+	m.Skip(1) // es flags
+
+	m.WriteBytes(4, 0x80, 0x80, 0x80, size4+header+size5)
+	m.WriteBytes(0x40) // object id
+	m.WriteBytes(0x15) // stream type
+	m.Skip(3)          // buffer size db
+	m.Skip(4)          // max bitraga
+	m.Skip(4)          // avg bitraga
+
+	m.WriteBytes(5, 0x80, 0x80, 0x80, size5)
+	m.Write(conf)
+
+	m.WriteBytes(6, 0x80, 0x80, 0x80, 1)
+	m.WriteBytes(2) // ?
+
+	m.EndAtom() // ESDS
+}
+
+func (m *Movie) WriteEsdsMP3() {
+	m.StartAtom("esds")
+	m.Skip(1) // version
+	m.Skip(3) // flags
+
+	// MP4ESDescrTag[3]:
+	// - MP4DecConfigDescrTag[4]:
+	// - Other[6]
+	const header = 5
+	const size3 = 3
+	const size4 = 13
+	const size6 = 1
+
+	m.WriteBytes(3, 0x80, 0x80, 0x80, size3+header+size4+header+size6)
+	m.Skip(2) // es id
+	m.Skip(1) // es flags
+
+	m.WriteBytes(4, 0x80, 0x80, 0x80, size4)
+	m.WriteBytes(0x6B) // object id
+	m.WriteBytes(0x15) // stream type
+	m.Skip(3)          // buffer size db
+	m.Skip(4)          // max bitraga
+	m.Skip(4)          // avg bitraga
+
+	m.WriteBytes(6, 0x80, 0x80, 0x80, 1)
+	m.WriteBytes(2) // ?
+
+	m.EndAtom() // ESDS
+}

pkg/iso/iso.go

@@ -0,0 +1,91 @@
+package iso
+
+import (
+	"encoding/binary"
+	"math"
+)
+
+type Movie struct {
+	b     []byte
+	start []int
+}
+
+func NewMovie(size int) *Movie {
+	return &Movie{b: make([]byte, 0, size)}
+}
+
+func (m *Movie) Bytes() []byte {
+	return m.b
+}
+
+func (m *Movie) StartAtom(name string) {
+	m.start = append(m.start, len(m.b))
+	m.b = append(m.b, 0, 0, 0, 0)
+	m.b = append(m.b, name...)
+}
+
+func (m *Movie) EndAtom() {
+	n := len(m.start) - 1
+
+	i := m.start[n]
+	size := uint32(len(m.b) - i)
+	binary.BigEndian.PutUint32(m.b[i:], size)
+
+	m.start = m.start[:n]
+}
+
+func (m *Movie) Write(b []byte) {
+	m.b = append(m.b, b...)
+}
+
+func (m *Movie) WriteBytes(b ...byte) {
+	m.b = append(m.b, b...)
+}
+
+func (m *Movie) WriteString(s string) {
+	m.b = append(m.b, s...)
+}
+
+func (m *Movie) Skip(n int) {
+	m.b = append(m.b, make([]byte, n)...)
+}
+
+func (m *Movie) WriteUint16(v uint16) {
+	m.b = append(m.b, byte(v>>8), byte(v))
+}
+
+func (m *Movie) WriteUint24(v uint32) {
+	m.b = append(m.b, byte(v>>16), byte(v>>8), byte(v))
+}
+
+func (m *Movie) WriteUint32(v uint32) {
+	m.b = append(m.b, byte(v>>24), byte(v>>16), byte(v>>8), byte(v))
+}
+
+func (m *Movie) WriteUint64(v uint64) {
+	m.b = append(m.b, byte(v>>56), byte(v>>48), byte(v>>40), byte(v>>32), byte(v>>24), byte(v>>16), byte(v>>8), byte(v))
+}
+
+func (m *Movie) WriteFloat16(f float64) {
+	i, f := math.Modf(f)
+	f *= 256
+	m.b = append(m.b, byte(i), byte(f))
+}
+
+func (m *Movie) WriteFloat32(f float64) {
+	i, f := math.Modf(f)
+	f *= 65536
+	m.b = append(m.b, byte(uint16(i)>>8), byte(i), byte(uint16(f)>>8), byte(f))
+}
+
+func (m *Movie) WriteMatrix() {
+	m.WriteUint32(0x00010000)
+	m.Skip(4)
+	m.Skip(4)
+	m.Skip(4)
+	m.WriteUint32(0x00010000)
+	m.Skip(4)
+	m.Skip(4)
+	m.Skip(4)
+	m.WriteUint32(0x40000000)
+}

pkg/mp4/README.md

@@ -1,19 +1,30 @@
+## Fragmented MP4
+
+```
+ffmpeg -i "rtsp://..." -movflags +frag_keyframe+separate_moof+default_base_moof+empty_moov -frag_duration 1 -c copy -t 5 sample.mp4
+```
+
+- movflags frag_keyframe 
+  Start a new fragment at each video keyframe.
+- frag_duration duration
+  Create fragments that are duration microseconds long.
+- movflags separate_moof
+  Write a separate moof (movie fragment) atom for each track.
+- movflags default_base_moof
+  Similarly to the omit_tfhd_offset, this flag avoids writing the absolute base_data_offset field in tfhd atoms, but does so by using the new default-base-is-moof flag instead.
+
+https://ffmpeg.org/ffmpeg-formats.html#Options-13
+
 ## HEVC
 
-Browser     | avc1 | hvc1 | hev1
-------------|------|------|---
-Mac Chrome  | +    | -    | +
-Mac Safari  | +    | +    | -
-iOS 15?     | +    | +    | -
-Mac Firefox | +    | -    | -
-iOS 12      | +    | -    | -
-Android 13  | +    | -    | -
-
-```
-ffmpeg -i input-hev1.mp4 -c:v copy -tag:v hvc1 -c:a copy output-hvc1.mp4
-Stream #0:0(eng): Video: hevc (Main) (hev1 / 0x31766568), yuv420p(tv, progressive), 720x404, 164 kb/s, 29.97 fps,
-Stream #0:0(eng): Video: hevc (Main) (hvc1 / 0x31637668), yuv420p(tv, progressive), 720x404, 164 kb/s, 29.97 fps,
-```
+| Browser     | avc1 | hvc1 | hev1 |
+|-------------|------|------|------|
+ | Mac Chrome  | +    | -    | +    |
+ | Mac Safari  | +    | +    | -    |
+ | iOS 15?     | +    | +    | -    |
+ | Mac Firefox | +    | -    | -    |
+ | iOS 12      | +    | -    | -    |
+ | Android 13  | +    | -    | -    |
 
 ## Useful links
 

pkg/mp4/consumer.go

@@ -24,6 +24,16 @@ type Consumer struct {
 	send uint32
 }
 
+// ParseQuery - like usual parse, but with mp4 param handler
+func ParseQuery(query map[string][]string) []*streamer.Media {
+	if query["mp4"] != nil {
+		cons := Consumer{}
+		return cons.GetMedias()
+	}
+
+	return streamer.ParseQuery(query)
+}
+
 const (
 	waitNone byte = iota
 	waitKeyframe
@@ -140,14 +150,33 @@ func (c *Consumer) AddTrack(media *streamer.Media, track *streamer.Track) *strea
 			push = wrapper(push)
 		}
 
+		return track.Bind(push)
+
+	case streamer.CodecOpus, streamer.CodecMP3, streamer.CodecPCMU, streamer.CodecPCMA:
+		push := func(packet *rtp.Packet) error {
+			if c.wait != waitNone {
+				return nil
+			}
+
+			buf := c.muxer.Marshal(trackID, packet)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			return nil
+		}
+
 		return track.Bind(push)
 	}
 
 	panic("unsupported codec")
 }
 
+func (c *Consumer) MimeCodecs() string {
+	return c.muxer.MimeCodecs(c.codecs)
+}
+
 func (c *Consumer) MimeType() string {
-	return c.muxer.MimeType(c.codecs)
+	return `video/mp4; codecs="` + c.MimeCodecs() + `"`
 }
 
 func (c *Consumer) Init() ([]byte, error) {

pkg/mp4/muxer.go

@@ -1,17 +1,13 @@
 package mp4
 
 import (
-	"encoding/binary"
 	"encoding/hex"
 	"github.com/AlexxIT/go2rtc/pkg/h264"
 	"github.com/AlexxIT/go2rtc/pkg/h265"
+	"github.com/AlexxIT/go2rtc/pkg/iso"
 	"github.com/AlexxIT/go2rtc/pkg/streamer"
-	"github.com/deepch/vdk/av"
 	"github.com/deepch/vdk/codec/h264parser"
 	"github.com/deepch/vdk/codec/h265parser"
-	"github.com/deepch/vdk/format/fmp4/fmp4io"
-	"github.com/deepch/vdk/format/mp4/mp4io"
-	"github.com/deepch/vdk/format/mp4f/mp4fio"
 	"github.com/pion/rtp"
 )
 
@@ -21,8 +17,15 @@ type Muxer struct {
 	pts       []uint32
 }
 
-func (m *Muxer) MimeType(codecs []*streamer.Codec) string {
-	s := `video/mp4; codecs="`
+const (
+	MimeH264 = "avc1.640029"
+	MimeH265 = "hvc1.1.6.L153.B0"
+	MimeAAC  = "mp4a.40.2"
+	MimeOpus = "opus"
+)
+
+func (m *Muxer) MimeCodecs(codecs []*streamer.Codec) string {
+	var s string
 
 	for i, codec := range codecs {
 		if i > 0 {
@@ -35,17 +38,23 @@ func (m *Muxer) MimeType(codecs []*streamer.Codec) string {
 		case streamer.CodecH265:
 			// H.265 profile=main level=5.1
 			// hvc1 - supported in Safari, hev1 - doesn't, both supported in Chrome
-			s += "hvc1.1.6.L153.B0"
+			s += MimeH265
 		case streamer.CodecAAC:
-			s += "mp4a.40.2"
+			s += MimeAAC
+		case streamer.CodecOpus:
+			s += MimeOpus
 		}
 	}
 
-	return s + `"`
+	return s
 }
 
 func (m *Muxer) GetInit(codecs []*streamer.Codec) ([]byte, error) {
-	moov := MOOV()
+	mv := iso.NewMovie(1024)
+	mv.WriteFileType()
+
+	mv.StartAtom(iso.Moov)
+	mv.WriteMovieHeader()
 
 	for i, codec := range codecs {
 		switch codec.Name {
@@ -62,35 +71,11 @@ func (m *Muxer) GetInit(codecs []*streamer.Codec) ([]byte, error) {
 				return nil, err
 			}
 
-			width := codecData.Width()
-			height := codecData.Height()
-
-			trak := TRAK(i + 1)
-			trak.Header.TrackWidth = float64(width)
-			trak.Header.TrackHeight = float64(height)
-			trak.Media.Header.TimeScale = int32(codec.ClockRate)
-			trak.Media.Handler = &mp4io.HandlerRefer{
-				SubType: [4]byte{'v', 'i', 'd', 'e'},
-				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
-			}
-			trak.Media.Info.Video = &mp4io.VideoMediaInfo{
-				Flags: 0x000001,
-			}
-			trak.Media.Info.Sample.SampleDesc.AVC1Desc = &mp4io.AVC1Desc{
-				DataRefIdx:           1,
-				HorizontalResolution: 72,
-				VorizontalResolution: 72,
-				Width:                int16(width),
-				Height:               int16(height),
-				FrameCount:           1,
-				Depth:                24,
-				ColorTableId:         -1,
-				Conf: &mp4io.AVC1Conf{
-					Data: codecData.AVCDecoderConfRecordBytes(),
-				},
-			}
-
-			moov.Tracks = append(moov.Tracks, trak)
+			mv.WriteVideoTrack(
+				uint32(i+1), codec.Name, codec.ClockRate,
+				uint16(codecData.Width()), uint16(codecData.Height()),
+				codecData.AVCDecoderConfRecordBytes(),
+			)
 
 		case streamer.CodecH265:
 			vps, sps, pps := h265.GetParameterSet(codec.FmtpLine)
@@ -106,35 +91,11 @@ func (m *Muxer) GetInit(codecs []*streamer.Codec) ([]byte, error) {
 				return nil, err
 			}
 
-			width := codecData.Width()
-			height := codecData.Height()
-
-			trak := TRAK(i + 1)
-			trak.Header.TrackWidth = float64(width)
-			trak.Header.TrackHeight = float64(height)
-			trak.Media.Header.TimeScale = int32(codec.ClockRate)
-			trak.Media.Handler = &mp4io.HandlerRefer{
-				SubType: [4]byte{'v', 'i', 'd', 'e'},
-				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
-			}
-			trak.Media.Info.Video = &mp4io.VideoMediaInfo{
-				Flags: 0x000001,
-			}
-			trak.Media.Info.Sample.SampleDesc.HV1Desc = &mp4io.HV1Desc{
-				DataRefIdx:           1,
-				HorizontalResolution: 72,
-				VorizontalResolution: 72,
-				Width:                int16(width),
-				Height:               int16(height),
-				FrameCount:           1,
-				Depth:                24,
-				ColorTableId:         -1,
-				Conf: &mp4io.HV1Conf{
-					Data: codecData.AVCDecoderConfRecordBytes(),
-				},
-			}
-
-			moov.Tracks = append(moov.Tracks, trak)
+			mv.WriteVideoTrack(
+				uint32(i+1), codec.Name, codec.ClockRate,
+				uint16(codecData.Width()), uint16(codecData.Height()),
+				codecData.AVCDecoderConfRecordBytes(),
+			)
 
 		case streamer.CodecAAC:
 			s := streamer.Between(codec.FmtpLine, "config=", ";")
@@ -143,44 +104,29 @@ func (m *Muxer) GetInit(codecs []*streamer.Codec) ([]byte, error) {
 				return nil, err
 			}
 
-			trak := TRAK(i + 1)
-			trak.Header.AlternateGroup = 1
-			trak.Header.Duration = 0
-			trak.Header.Volume = 1
-			trak.Media.Header.TimeScale = int32(codec.ClockRate)
+			mv.WriteAudioTrack(
+				uint32(i+1), codec.Name, codec.ClockRate, codec.Channels, b,
+			)
 
-			trak.Media.Handler = &mp4io.HandlerRefer{
-				SubType: [4]byte{'s', 'o', 'u', 'n'},
-				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
-			}
-			trak.Media.Info.Sound = &mp4io.SoundMediaInfo{}
-
-			trak.Media.Info.Sample.SampleDesc.MP4ADesc = &mp4io.MP4ADesc{
-				DataRefIdx:       1,
-				NumberOfChannels: int16(codec.Channels),
-				SampleSize:       int16(av.FLTP.BytesPerSample() * 4),
-				SampleRate:       float64(codec.ClockRate),
-				Unknowns:         []mp4io.Atom{ESDS(b)},
-			}
-
-			moov.Tracks = append(moov.Tracks, trak)
+		case streamer.CodecOpus, streamer.CodecMP3, streamer.CodecPCMU, streamer.CodecPCMA:
+			mv.WriteAudioTrack(
+				uint32(i+1), codec.Name, codec.ClockRate, codec.Channels, nil,
+			)
 		}
 
-		trex := &mp4io.TrackExtend{
-			TrackId:               uint32(i + 1),
-			DefaultSampleDescIdx:  1,
-			DefaultSampleDuration: 0,
-		}
-		moov.MovieExtend.Tracks = append(moov.MovieExtend.Tracks, trex)
-
 		m.pts = append(m.pts, 0)
 		m.dts = append(m.dts, 0)
 	}
 
-	data := make([]byte, moov.Len())
-	moov.Marshal(data)
+	mv.StartAtom(iso.MoovMvex)
+	for i := range codecs {
+		mv.WriteTrackExtend(uint32(i + 1))
+	}
+	mv.EndAtom() // MVEX
 
-	return append(FTYP(), data...), nil
+	mv.EndAtom() // MOOV
+
+	return mv.Bytes(), nil
 }
 
 func (m *Muxer) Reset() {
@@ -192,65 +138,28 @@ func (m *Muxer) Reset() {
 }
 
 func (m *Muxer) Marshal(trackID byte, packet *rtp.Packet) []byte {
-	run := &mp4fio.TrackFragRun{
-		Flags:            0x000b05,
-		FirstSampleFlags: uint32(fmp4io.SampleNoDependencies),
-		DataOffset:       0,
-		Entries:          []mp4io.TrackFragRunEntry{},
-	}
-
-	moof := &mp4fio.MovieFrag{
-		Header: &mp4fio.MovieFragHeader{
-			Seqnum: m.fragIndex + 1,
-		},
-		Tracks: []*mp4fio.TrackFrag{
-			{
-				Header: &mp4fio.TrackFragHeader{
-					Data: []byte{0x00, 0x02, 0x00, 0x20, 0x00, 0x00, 0x00, trackID + 1, 0x01, 0x01, 0x00, 0x00},
-				},
-				DecodeTime: &mp4fio.TrackFragDecodeTime{
-					Version: 1,
-					Flags:   0,
-					Time:    m.dts[trackID],
-				},
-				Run: run,
-			},
-		},
-	}
-
-	entry := mp4io.TrackFragRunEntry{
-		Size: uint32(len(packet.Payload)),
-	}
-
-	newTime := packet.Timestamp
-	if m.pts[trackID] > 0 {
-		entry.Duration = newTime - m.pts[trackID]
-		m.dts[trackID] += uint64(entry.Duration)
-	} else {
-		// important, or Safari will fail with first frame
-		entry.Duration = 1
-	}
-	m.pts[trackID] = newTime
-
-	// important before moof.Len()
-	run.Entries = append(run.Entries, entry)
-
-	moofLen := moof.Len()
-	mdatLen := 8 + len(packet.Payload)
-
-	// important after moof.Len()
-	run.DataOffset = uint32(moofLen + 8)
-
-	buf := make([]byte, moofLen+mdatLen)
-	moof.Marshal(buf)
-
-	binary.BigEndian.PutUint32(buf[moofLen:], uint32(mdatLen))
-	copy(buf[moofLen+4:], "mdat")
-	copy(buf[moofLen+8:], packet.Payload)
+	// important before increment
+	time := m.dts[trackID]
 
 	m.fragIndex++
 
-	//m.total += moofLen + mdatLen
+	var duration uint32
+	newTime := packet.Timestamp
+	if m.pts[trackID] > 0 {
+		duration = newTime - m.pts[trackID]
+		m.dts[trackID] += uint64(duration)
+	} else {
+		// important, or Safari will fail with first frame
+		duration = 1
+	}
+	m.pts[trackID] = newTime
 
-	return buf
+	mv := iso.NewMovie(1024 + len(packet.Payload))
+	mv.WriteMovieFragment(
+		m.fragIndex, uint32(trackID+1), duration,
+		uint32(len(packet.Payload)), time,
+	)
+	mv.WriteData(packet.Payload)
+
+	return mv.Bytes()
 }

pkg/mp4/segment.go

@@ -50,7 +50,7 @@ func (c *Segment) AddTrack(media *streamer.Media, track *streamer.Track) *stream
 		return nil
 	}
 
-	c.MimeType = muxer.MimeType(codecs)
+	c.MimeType = `video/mp4; codecs="` + muxer.MimeCodecs(codecs) + `"`
 
 	switch track.Codec.Name {
 	case streamer.CodecH264:

pkg/mp4/v1/consumer.go

@@ -1,6 +1,8 @@
-package mp4f
+package mp4
 
 import (
+	"encoding/hex"
+	"github.com/AlexxIT/go2rtc/pkg/aac"
 	"github.com/AlexxIT/go2rtc/pkg/h264"
 	"github.com/AlexxIT/go2rtc/pkg/streamer"
 	"github.com/deepch/vdk/av"
@@ -14,6 +16,7 @@ import (
 type Consumer struct {
 	streamer.Element
 
+	Medias     []*streamer.Media
 	UserAgent  string
 	RemoteAddr string
 
@@ -26,6 +29,10 @@ type Consumer struct {
 }
 
 func (c *Consumer) GetMedias() []*streamer.Media {
+	if c.Medias != nil {
+		return c.Medias
+	}
+
 	return []*streamer.Media{
 		{
 			Kind:      streamer.KindVideo,
@@ -96,7 +103,17 @@ func (c *Consumer) AddTrack(media *streamer.Media, track *streamer.Track) *strea
 		return track.Bind(push)
 
 	case streamer.CodecAAC:
-		stream, _ := aacparser.NewCodecDataFromMPEG4AudioConfigBytes([]byte{20, 8})
+		s := streamer.Between(codec.FmtpLine, "config=", ";")
+
+		b, err := hex.DecodeString(s)
+		if err != nil {
+			return nil
+		}
+
+		stream, err := aacparser.NewCodecDataFromMPEG4AudioConfigBytes(b)
+		if err != nil {
+			return nil
+		}
 
 		c.mimeType += ",mp4a.40.2"
 		c.streams = append(c.streams, stream)
@@ -126,6 +143,11 @@ func (c *Consumer) AddTrack(media *streamer.Media, track *streamer.Track) *strea
 			return nil
 		}
 
+		if codec.IsRTP() {
+			wrapper := aac.RTPDepay(track)
+			push = wrapper(push)
+		}
+
 		return track.Bind(push)
 	}
 

pkg/mp4/v2/consumer.go

@@ -0,0 +1,174 @@
+package mp4
+
+import (
+	"encoding/json"
+	"github.com/AlexxIT/go2rtc/pkg/aac"
+	"github.com/AlexxIT/go2rtc/pkg/h264"
+	"github.com/AlexxIT/go2rtc/pkg/h265"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/pion/rtp"
+	"sync/atomic"
+)
+
+type Consumer struct {
+	streamer.Element
+
+	Medias     []*streamer.Media
+	UserAgent  string
+	RemoteAddr string
+
+	muxer  *Muxer
+	codecs []*streamer.Codec
+	wait   byte
+
+	send uint32
+}
+
+const (
+	waitNone byte = iota
+	waitKeyframe
+	waitInit
+)
+
+func (c *Consumer) GetMedias() []*streamer.Media {
+	if c.Medias != nil {
+		return c.Medias
+	}
+
+	// default medias
+	return []*streamer.Media{
+		{
+			Kind:      streamer.KindVideo,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecH264},
+				{Name: streamer.CodecH265},
+			},
+		},
+		{
+			Kind:      streamer.KindAudio,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecAAC},
+			},
+		},
+	}
+}
+
+func (c *Consumer) AddTrack(media *streamer.Media, track *streamer.Track) *streamer.Track {
+	trackID := byte(len(c.codecs))
+	c.codecs = append(c.codecs, track.Codec)
+
+	codec := track.Codec
+	switch codec.Name {
+	case streamer.CodecH264:
+		c.wait = waitInit
+
+		push := func(packet *rtp.Packet) error {
+			if packet.Version != h264.RTPPacketVersionAVC {
+				return nil
+			}
+
+			if c.wait != waitNone {
+				if c.wait == waitInit || !h264.IsKeyframe(packet.Payload) {
+					return nil
+				}
+				c.wait = waitNone
+			}
+
+			buf := c.muxer.Marshal(trackID, packet)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			return nil
+		}
+
+		var wrapper streamer.WrapperFunc
+		if codec.IsRTP() {
+			wrapper = h264.RTPDepay(track)
+		} else {
+			wrapper = h264.RepairAVC(track)
+		}
+		push = wrapper(push)
+
+		return track.Bind(push)
+
+	case streamer.CodecH265:
+		c.wait = waitInit
+
+		push := func(packet *rtp.Packet) error {
+			if packet.Version != h264.RTPPacketVersionAVC {
+				return nil
+			}
+
+			if c.wait != waitNone {
+				if c.wait == waitInit || !h265.IsKeyframe(packet.Payload) {
+					return nil
+				}
+				c.wait = waitNone
+			}
+
+			buf := c.muxer.Marshal(trackID, packet)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			return nil
+		}
+
+		if codec.IsRTP() {
+			wrapper := h265.RTPDepay(track)
+			push = wrapper(push)
+		}
+
+		return track.Bind(push)
+
+	case streamer.CodecAAC:
+		push := func(packet *rtp.Packet) error {
+			if c.wait != waitNone {
+				return nil
+			}
+
+			buf := c.muxer.Marshal(trackID, packet)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			return nil
+		}
+
+		if codec.IsRTP() {
+			wrapper := aac.RTPDepay(track)
+			push = wrapper(push)
+		}
+
+		return track.Bind(push)
+	}
+
+	panic("unsupported codec")
+}
+
+func (c *Consumer) MimeType() string {
+	return c.muxer.MimeType(c.codecs)
+}
+
+func (c *Consumer) Init() ([]byte, error) {
+	c.muxer = &Muxer{}
+	return c.muxer.GetInit(c.codecs)
+}
+
+func (c *Consumer) Start() {
+	if c.wait == waitInit {
+		c.wait = waitKeyframe
+	}
+}
+
+//
+
+func (c *Consumer) MarshalJSON() ([]byte, error) {
+	info := &streamer.Info{
+		Type:       "MP4 client",
+		RemoteAddr: c.RemoteAddr,
+		UserAgent:  c.UserAgent,
+		Send:       atomic.LoadUint32(&c.send),
+	}
+	return json.Marshal(info)
+}

pkg/mp4/v2/muxer.go

@@ -0,0 +1,256 @@
+package mp4
+
+import (
+	"encoding/binary"
+	"encoding/hex"
+	"github.com/AlexxIT/go2rtc/pkg/h264"
+	"github.com/AlexxIT/go2rtc/pkg/h265"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/deepch/vdk/av"
+	"github.com/deepch/vdk/codec/h264parser"
+	"github.com/deepch/vdk/codec/h265parser"
+	"github.com/deepch/vdk/format/fmp4/fmp4io"
+	"github.com/deepch/vdk/format/mp4/mp4io"
+	"github.com/deepch/vdk/format/mp4f/mp4fio"
+	"github.com/pion/rtp"
+)
+
+type Muxer struct {
+	fragIndex uint32
+	dts       []uint64
+	pts       []uint32
+}
+
+func (m *Muxer) MimeType(codecs []*streamer.Codec) string {
+	s := `video/mp4; codecs="`
+
+	for i, codec := range codecs {
+		if i > 0 {
+			s += ","
+		}
+
+		switch codec.Name {
+		case streamer.CodecH264:
+			s += "avc1." + h264.GetProfileLevelID(codec.FmtpLine)
+		case streamer.CodecH265:
+			// H.265 profile=main level=5.1
+			// hvc1 - supported in Safari, hev1 - doesn't, both supported in Chrome
+			s += "hvc1.1.6.L153.B0"
+		case streamer.CodecAAC:
+			s += "mp4a.40.2"
+		}
+	}
+
+	return s + `"`
+}
+
+func (m *Muxer) GetInit(codecs []*streamer.Codec) ([]byte, error) {
+	moov := MOOV()
+
+	for i, codec := range codecs {
+		switch codec.Name {
+		case streamer.CodecH264:
+			sps, pps := h264.GetParameterSet(codec.FmtpLine)
+			if sps == nil {
+				// some dummy SPS and PPS not a problem
+				sps = []byte{0x67, 0x42, 0x00, 0x0a, 0xf8, 0x41, 0xa2}
+				pps = []byte{0x68, 0xce, 0x38, 0x80}
+			}
+
+			codecData, err := h264parser.NewCodecDataFromSPSAndPPS(sps, pps)
+			if err != nil {
+				return nil, err
+			}
+
+			width := codecData.Width()
+			height := codecData.Height()
+
+			trak := TRAK(i + 1)
+			trak.Header.TrackWidth = float64(width)
+			trak.Header.TrackHeight = float64(height)
+			trak.Media.Header.TimeScale = int32(codec.ClockRate)
+			trak.Media.Handler = &mp4io.HandlerRefer{
+				SubType: [4]byte{'v', 'i', 'd', 'e'},
+				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
+			}
+			trak.Media.Info.Video = &mp4io.VideoMediaInfo{
+				Flags: 0x000001,
+			}
+			trak.Media.Info.Sample.SampleDesc.AVC1Desc = &mp4io.AVC1Desc{
+				DataRefIdx:           1,
+				HorizontalResolution: 72,
+				VorizontalResolution: 72,
+				Width:                int16(width),
+				Height:               int16(height),
+				FrameCount:           1,
+				Depth:                24,
+				ColorTableId:         -1,
+				Conf: &mp4io.AVC1Conf{
+					Data: codecData.AVCDecoderConfRecordBytes(),
+				},
+			}
+
+			moov.Tracks = append(moov.Tracks, trak)
+
+		case streamer.CodecH265:
+			vps, sps, pps := h265.GetParameterSet(codec.FmtpLine)
+			if sps == nil {
+				// some dummy SPS and PPS not a problem
+				vps = []byte{0x40, 0x01, 0x0c, 0x01, 0xff, 0xff, 0x01, 0x40, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x99, 0xac, 0x09}
+				sps = []byte{0x42, 0x01, 0x01, 0x01, 0x40, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x99, 0xa0, 0x01, 0x40, 0x20, 0x05, 0xa1, 0xfe, 0x5a, 0xee, 0x46, 0xc1, 0xae, 0x55, 0x04}
+				pps = []byte{0x44, 0x01, 0xc0, 0x73, 0xc0, 0x4c, 0x90}
+			}
+
+			codecData, err := h265parser.NewCodecDataFromVPSAndSPSAndPPS(vps, sps, pps)
+			if err != nil {
+				return nil, err
+			}
+
+			width := codecData.Width()
+			height := codecData.Height()
+
+			trak := TRAK(i + 1)
+			trak.Header.TrackWidth = float64(width)
+			trak.Header.TrackHeight = float64(height)
+			trak.Media.Header.TimeScale = int32(codec.ClockRate)
+			trak.Media.Handler = &mp4io.HandlerRefer{
+				SubType: [4]byte{'v', 'i', 'd', 'e'},
+				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
+			}
+			trak.Media.Info.Video = &mp4io.VideoMediaInfo{
+				Flags: 0x000001,
+			}
+			trak.Media.Info.Sample.SampleDesc.HV1Desc = &mp4io.HV1Desc{
+				DataRefIdx:           1,
+				HorizontalResolution: 72,
+				VorizontalResolution: 72,
+				Width:                int16(width),
+				Height:               int16(height),
+				FrameCount:           1,
+				Depth:                24,
+				ColorTableId:         -1,
+				Conf: &mp4io.HV1Conf{
+					Data: codecData.AVCDecoderConfRecordBytes(),
+				},
+			}
+
+			moov.Tracks = append(moov.Tracks, trak)
+
+		case streamer.CodecAAC:
+			s := streamer.Between(codec.FmtpLine, "config=", ";")
+			b, err := hex.DecodeString(s)
+			if err != nil {
+				return nil, err
+			}
+
+			trak := TRAK(i + 1)
+			trak.Header.AlternateGroup = 1
+			trak.Header.Duration = 0
+			trak.Header.Volume = 1
+			trak.Media.Header.TimeScale = int32(codec.ClockRate)
+
+			trak.Media.Handler = &mp4io.HandlerRefer{
+				SubType: [4]byte{'s', 'o', 'u', 'n'},
+				Name:    []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 'm', 'a', 'i', 'n', 0},
+			}
+			trak.Media.Info.Sound = &mp4io.SoundMediaInfo{}
+
+			trak.Media.Info.Sample.SampleDesc.MP4ADesc = &mp4io.MP4ADesc{
+				DataRefIdx:       1,
+				NumberOfChannels: int16(codec.Channels),
+				SampleSize:       int16(av.FLTP.BytesPerSample() * 4),
+				SampleRate:       float64(codec.ClockRate),
+				Unknowns:         []mp4io.Atom{ESDS(b)},
+			}
+
+			moov.Tracks = append(moov.Tracks, trak)
+		}
+
+		trex := &mp4io.TrackExtend{
+			TrackId:               uint32(i + 1),
+			DefaultSampleDescIdx:  1,
+			DefaultSampleDuration: 0,
+		}
+		moov.MovieExtend.Tracks = append(moov.MovieExtend.Tracks, trex)
+
+		m.pts = append(m.pts, 0)
+		m.dts = append(m.dts, 0)
+	}
+
+	data := make([]byte, moov.Len())
+	moov.Marshal(data)
+
+	return append(FTYP(), data...), nil
+}
+
+func (m *Muxer) Reset() {
+	m.fragIndex = 0
+	for i := range m.dts {
+		m.dts[i] = 0
+		m.pts[i] = 0
+	}
+}
+
+func (m *Muxer) Marshal(trackID byte, packet *rtp.Packet) []byte {
+	run := &mp4fio.TrackFragRun{
+		Flags:            0x000b05,
+		FirstSampleFlags: uint32(fmp4io.SampleNoDependencies),
+		DataOffset:       0,
+		Entries:          []mp4io.TrackFragRunEntry{},
+	}
+
+	moof := &mp4fio.MovieFrag{
+		Header: &mp4fio.MovieFragHeader{
+			Seqnum: m.fragIndex + 1,
+		},
+		Tracks: []*mp4fio.TrackFrag{
+			{
+				Header: &mp4fio.TrackFragHeader{
+					Data: []byte{0x00, 0x02, 0x00, 0x20, 0x00, 0x00, 0x00, trackID + 1, 0x01, 0x01, 0x00, 0x00},
+				},
+				DecodeTime: &mp4fio.TrackFragDecodeTime{
+					Version: 1,
+					Flags:   0,
+					Time:    m.dts[trackID],
+				},
+				Run: run,
+			},
+		},
+	}
+
+	entry := mp4io.TrackFragRunEntry{
+		Size: uint32(len(packet.Payload)),
+	}
+
+	newTime := packet.Timestamp
+	if m.pts[trackID] > 0 {
+		entry.Duration = newTime - m.pts[trackID]
+		m.dts[trackID] += uint64(entry.Duration)
+	} else {
+		// important, or Safari will fail with first frame
+		entry.Duration = 1
+	}
+	m.pts[trackID] = newTime
+
+	// important before moof.Len()
+	run.Entries = append(run.Entries, entry)
+
+	moofLen := moof.Len()
+	mdatLen := 8 + len(packet.Payload)
+
+	// important after moof.Len()
+	run.DataOffset = uint32(moofLen + 8)
+
+	buf := make([]byte, moofLen+mdatLen)
+	moof.Marshal(buf)
+
+	binary.BigEndian.PutUint32(buf[moofLen:], uint32(mdatLen))
+	copy(buf[moofLen+4:], "mdat")
+	copy(buf[moofLen+8:], packet.Payload)
+
+	m.fragIndex++
+
+	//m.total += moofLen + mdatLen
+
+	return buf
+}

pkg/mp4/v2/segment.go

@@ -0,0 +1,143 @@
+package mp4
+
+import (
+	"encoding/json"
+	"github.com/AlexxIT/go2rtc/pkg/h264"
+	"github.com/AlexxIT/go2rtc/pkg/h265"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/pion/rtp"
+	"sync/atomic"
+)
+
+type Segment struct {
+	streamer.Element
+
+	Medias     []*streamer.Media
+	UserAgent  string
+	RemoteAddr string
+
+	MimeType     string
+	OnlyKeyframe bool
+
+	send uint32
+}
+
+func (c *Segment) GetMedias() []*streamer.Media {
+	if c.Medias != nil {
+		return c.Medias
+	}
+
+	// default medias
+	return []*streamer.Media{
+		{
+			Kind:      streamer.KindVideo,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecH264},
+				{Name: streamer.CodecH265},
+			},
+		},
+	}
+}
+
+func (c *Segment) AddTrack(media *streamer.Media, track *streamer.Track) *streamer.Track {
+	muxer := &Muxer{}
+
+	codecs := []*streamer.Codec{track.Codec}
+
+	init, err := muxer.GetInit(codecs)
+	if err != nil {
+		return nil
+	}
+
+	c.MimeType = muxer.MimeType(codecs)
+
+	switch track.Codec.Name {
+	case streamer.CodecH264:
+		var push streamer.WriterFunc
+
+		if c.OnlyKeyframe {
+			push = func(packet *rtp.Packet) error {
+				if !h264.IsKeyframe(packet.Payload) {
+					return nil
+				}
+
+				buf := muxer.Marshal(0, packet)
+				atomic.AddUint32(&c.send, uint32(len(buf)))
+				c.Fire(append(init, buf...))
+
+				return nil
+			}
+		} else {
+			var buf []byte
+
+			push = func(packet *rtp.Packet) error {
+				if h264.IsKeyframe(packet.Payload) {
+					// fist frame - send only IFrame
+					// other frames - send IFrame and all PFrames
+					if buf == nil {
+						buf = append(buf, init...)
+						b := muxer.Marshal(0, packet)
+						buf = append(buf, b...)
+					}
+
+					atomic.AddUint32(&c.send, uint32(len(buf)))
+					c.Fire(buf)
+
+					buf = buf[:0]
+					buf = append(buf, init...)
+					muxer.Reset()
+				}
+
+				if buf != nil {
+					b := muxer.Marshal(0, packet)
+					buf = append(buf, b...)
+				}
+
+				return nil
+			}
+		}
+
+		var wrapper streamer.WrapperFunc
+		if track.Codec.IsRTP() {
+			wrapper = h264.RTPDepay(track)
+		} else {
+			wrapper = h264.RepairAVC(track)
+		}
+		push = wrapper(push)
+
+		return track.Bind(push)
+
+	case streamer.CodecH265:
+		push := func(packet *rtp.Packet) error {
+			if !h265.IsKeyframe(packet.Payload) {
+				return nil
+			}
+
+			buf := muxer.Marshal(0, packet)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(append(init, buf...))
+
+			return nil
+		}
+
+		if track.Codec.IsRTP() {
+			wrapper := h265.RTPDepay(track)
+			push = wrapper(push)
+		}
+
+		return track.Bind(push)
+	}
+
+	panic("unsupported codec")
+}
+
+func (c *Segment) MarshalJSON() ([]byte, error) {
+	info := &streamer.Info{
+		Type:       "WS/MP4 client",
+		RemoteAddr: c.RemoteAddr,
+		UserAgent:  c.UserAgent,
+		Send:       atomic.LoadUint32(&c.send),
+	}
+	return json.Marshal(info)
+}

pkg/rtsp/conn.go

@@ -2,7 +2,6 @@ package rtsp
 
 import (
 	"bufio"
-	"bytes"
 	"crypto/tls"
 	"encoding/binary"
 	"errors"
@@ -281,7 +280,7 @@ func (c *Conn) Options() error {
 	}
 
 	if val := res.Header.Get("Content-Base"); val != "" {
-		c.URL, err = url.Parse(val)
+		c.URL, err = urlParse(val)
 		if err != nil {
 			return err
 		}
@@ -311,7 +310,7 @@ func (c *Conn) Describe() error {
 	}
 
 	if val := res.Header.Get("Content-Base"); val != "" {
-		c.URL, err = url.Parse(val)
+		c.URL, err = urlParse(val)
 		if err != nil {
 			return err
 		}
@@ -348,7 +347,7 @@ func (c *Conn) Describe() error {
 
 func (c *Conn) Setup() error {
 	for _, media := range c.Medias {
-		_, err := c.SetupMedia(media, media.Codecs[0])
+		_, err := c.SetupMedia(media, media.Codecs[0], true)
 		if err != nil {
 			return err
 		}
@@ -357,11 +356,12 @@ func (c *Conn) Setup() error {
 	return nil
 }
 
-func (c *Conn) SetupMedia(
-	media *streamer.Media, codec *streamer.Codec,
-) (*streamer.Track, error) {
-	c.stateMu.Lock()
-	defer c.stateMu.Unlock()
+func (c *Conn) SetupMedia(media *streamer.Media, codec *streamer.Codec, first bool) (*streamer.Track, error) {
+	// TODO: rewrite recoonection and first flag
+	if first {
+		c.stateMu.Lock()
+		defer c.stateMu.Unlock()
+	}
 
 	if c.state != StateConn && c.state != StateSetup {
 		return nil, fmt.Errorf("RTSP SETUP from wrong state: %s", c.state)
@@ -380,7 +380,7 @@ func (c *Conn) SetupMedia(
 		}
 		rawURL += media.Control
 	}
-	trackURL, err := url.Parse(rawURL)
+	trackURL, err := urlParse(rawURL)
 	if err != nil {
 		return nil, err
 	}
@@ -413,7 +413,7 @@ func (c *Conn) SetupMedia(
 
 			for _, newMedia := range c.Medias {
 				if newMedia.Control == media.Control {
-					return c.SetupMedia(newMedia, newMedia.Codecs[0])
+					return c.SetupMedia(newMedia, newMedia.Codecs[0], false)
 				}
 			}
 		}
@@ -882,42 +882,3 @@ func (c *Conn) bindTrack(
 
 	return track.Bind(push)
 }
-
-type RTCP struct {
-	Channel byte
-	Header  rtcp.Header
-	Packets []rtcp.Packet
-}
-
-const sdpHeader = `v=0
-o=- 0 0 IN IP4 0.0.0.0
-s=-
-t=0 0`
-
-func UnmarshalSDP(rawSDP []byte) ([]*streamer.Media, error) {
-	medias, err := streamer.UnmarshalSDP(rawSDP)
-	if err != nil {
-		// fix SDP header for some cameras
-		i := bytes.Index(rawSDP, []byte("\nm="))
-		if i > 0 {
-			rawSDP = append([]byte(sdpHeader), rawSDP[i:]...)
-			medias, err = streamer.UnmarshalSDP(rawSDP)
-		}
-		if err != nil {
-			return nil, err
-		}
-	}
-
-	// fix bug in ONVIF spec
-	// https://www.onvif.org/specs/stream/ONVIF-Streaming-Spec-v241.pdf
-	for _, media := range medias {
-		switch media.Direction {
-		case streamer.DirectionRecvonly, "":
-			media.Direction = streamer.DirectionSendonly
-		case streamer.DirectionSendonly:
-			media.Direction = streamer.DirectionRecvonly
-		}
-	}
-
-	return medias, nil
-}

pkg/rtsp/helpers.go

@@ -0,0 +1,68 @@
+package rtsp
+
+import (
+	"bytes"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/pion/rtcp"
+	"net/url"
+	"strings"
+)
+
+type RTCP struct {
+	Channel byte
+	Header  rtcp.Header
+	Packets []rtcp.Packet
+}
+
+const sdpHeader = `v=0
+o=- 0 0 IN IP4 0.0.0.0
+s=-
+t=0 0`
+
+func UnmarshalSDP(rawSDP []byte) ([]*streamer.Media, error) {
+	medias, err := streamer.UnmarshalSDP(rawSDP)
+	if err != nil {
+		// fix SDP header for some cameras
+		i := bytes.Index(rawSDP, []byte("\nm="))
+		if i > 0 {
+			rawSDP = append([]byte(sdpHeader), rawSDP[i:]...)
+			medias, err = streamer.UnmarshalSDP(rawSDP)
+		}
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	// fix bug in ONVIF spec
+	// https://www.onvif.org/specs/stream/ONVIF-Streaming-Spec-v241.pdf
+	for _, media := range medias {
+		switch media.Direction {
+		case streamer.DirectionRecvonly, "":
+			media.Direction = streamer.DirectionSendonly
+		case streamer.DirectionSendonly:
+			media.Direction = streamer.DirectionRecvonly
+		}
+	}
+
+	return medias, nil
+}
+
+// urlParse fix bugs:
+// 1. Content-Base: rtsp://::ffff:192.168.1.123/onvif/profile.1/
+// 2. Content-Base: rtsp://rtsp://turret2-cam.lan:554/stream1/
+func urlParse(rawURL string) (*url.URL, error) {
+	if strings.HasPrefix(rawURL, "rtsp://rtsp://") {
+		rawURL = rawURL[7:]
+	}
+
+	u, err := url.Parse(rawURL)
+	if err != nil && strings.HasSuffix(err.Error(), "after host") {
+		if i1 := strings.Index(rawURL, "://"); i1 > 0 {
+			if i2 := strings.IndexByte(rawURL[i1+3:], '/'); i2 > 0 {
+				return urlParse(rawURL[:i1+3+i2] + ":" + rawURL[i1+3+i2:])
+			}
+		}
+	}
+
+	return u, err
+}

pkg/rtsp/rtsp_test.go

@@ -0,0 +1,20 @@
+package rtsp
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+)
+
+func TestURLParse(t *testing.T) {
+	// https://github.com/AlexxIT/WebRTC/issues/395
+	base := "rtsp://::ffff:192.168.1.123/onvif/profile.1/"
+	u, err := urlParse(base)
+	assert.Empty(t, err)
+	assert.Equal(t, "::ffff:192.168.1.123:", u.Host)
+
+	// https://github.com/AlexxIT/go2rtc/issues/208
+	base = "rtsp://rtsp://turret2-cam.lan:554/stream1/"
+	u, err = urlParse(base)
+	assert.Empty(t, err)
+	assert.Equal(t, "turret2-cam.lan:554", u.Host)
+}

pkg/rtsp/streamer.go

@@ -9,7 +9,26 @@ import (
 // Element Producer
 
 func (c *Conn) GetMedias() []*streamer.Media {
-	return c.Medias
+	if c.Medias != nil {
+		return c.Medias
+	}
+
+	return []*streamer.Media{
+		{
+			Kind:      streamer.KindVideo,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecAll},
+			},
+		},
+		{
+			Kind:      streamer.KindAudio,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecAll},
+			},
+		},
+	}
 }
 
 func (c *Conn) GetTrack(media *streamer.Media, codec *streamer.Codec) *streamer.Track {
@@ -26,7 +45,7 @@ func (c *Conn) GetTrack(media *streamer.Media, codec *streamer.Codec) *streamer.
 		return streamer.NewTrack(codec, media.Direction)
 	}
 
-	track, err := c.SetupMedia(media, codec)
+	track, err := c.SetupMedia(media, codec, true)
 	if err != nil {
 		return nil
 	}
@@ -63,11 +82,20 @@ func (c *Conn) AddTrack(media *streamer.Media, track *streamer.Track) *streamer.
 		codec := track.Codec.Clone()
 		codec.PayloadType = uint8(96 + i)
 
-		for i, m := range c.Medias {
-			if m == media {
-				media.Codecs = []*streamer.Codec{codec}
-				c.Medias[i] = media
-				break
+		if media.MatchAll() {
+			// fill consumer medias list
+			c.Medias = append(c.Medias, &streamer.Media{
+				Kind: media.Kind, Direction: media.Direction,
+				Codecs: []*streamer.Codec{codec},
+			})
+		} else {
+			// find consumer media and replace codec with right one
+			for i, m := range c.Medias {
+				if m == media {
+					media.Codecs = []*streamer.Codec{codec}
+					c.Medias[i] = media
+					break
+				}
 			}
 		}
 

pkg/streamer/media.go

@@ -33,9 +33,12 @@ const (
 	CodecAAC  = "MPEG4-GENERIC"
 	CodecOpus = "OPUS" // payloadType: 111
 	CodecG722 = "G722"
-	CodecMPA  = "MPA" // payload: 14
+	CodecMP3  = "MPA" // payload: 14, aka MPEG-1 Layer III
 
 	CodecELD = "ELD" // AAC-ELD
+
+	CodecAll = "ALL"
+	CodecAny = "ANY"
 )
 
 const PayloadTypeRAW byte = 255
@@ -44,7 +47,7 @@ func GetKind(name string) string {
 	switch name {
 	case CodecH264, CodecH265, CodecVP8, CodecVP9, CodecAV1, CodecJPEG:
 		return KindVideo
-	case CodecPCMU, CodecPCMA, CodecAAC, CodecOpus, CodecG722, CodecMPA, CodecELD:
+	case CodecPCMU, CodecPCMA, CodecAAC, CodecOpus, CodecG722, CodecMP3, CodecELD:
 		return KindAudio
 	}
 	return ""
@@ -60,7 +63,6 @@ type Media struct {
 
 	MID     string `json:"mid,omitempty"`     // TODO: fixme?
 	Control string `json:"control,omitempty"` // TODO: fixme?
-	Title   string `json:"title,omitempty"`   // TODO: fixme?
 }
 
 func (m *Media) String() string {
@@ -112,10 +114,6 @@ func (m *Media) MatchMedia(media *Media) *Codec {
 	}
 
 	for _, localCodec := range m.Codecs {
-		if media.Codecs == nil {
-			return localCodec
-		}
-
 		for _, remoteCodec := range media.Codecs {
 			if localCodec.Match(remoteCodec) {
 				return localCodec
@@ -125,6 +123,10 @@ func (m *Media) MatchMedia(media *Media) *Codec {
 	return nil
 }
 
+func (m *Media) MatchAll() bool {
+	return len(m.Codecs) > 0 && m.Codecs[0].Name == CodecAll
+}
+
 // Codec take best from:
 // - deepch/vdk/av.CodecData
 // - pion/webrtc.RTPCodecCapability
@@ -154,6 +156,11 @@ func (c *Codec) Clone() *Codec {
 }
 
 func (c *Codec) Match(codec *Codec) bool {
+	switch codec.Name {
+	case CodecAll, CodecAny:
+		return true
+	}
+
 	return c.Name == codec.Name &&
 		(c.ClockRate == codec.ClockRate || codec.ClockRate == 0) &&
 		(c.Channels == codec.Channels || codec.Channels == 0)
@@ -286,7 +293,7 @@ func UnmarshalCodec(md *sdp.MediaDescription, payloadType string) *Codec {
 			c.Name = CodecPCMA
 			c.ClockRate = 8000
 		case "14":
-			c.Name = CodecMPA
+			c.Name = CodecMP3
 			c.ClockRate = 44100
 		case "26":
 			c.Name = CodecJPEG
@@ -299,6 +306,40 @@ func UnmarshalCodec(md *sdp.MediaDescription, payloadType string) *Codec {
 	return c
 }
 
+func ParseQuery(query map[string][]string) (medias []*Media) {
+	// set media candidates from query list
+	for key, values := range query {
+		switch key {
+		case KindVideo, KindAudio:
+			for _, value := range values {
+				media := &Media{Kind: key, Direction: DirectionRecvonly}
+
+				for _, name := range strings.Split(value, ",") {
+					name = strings.ToUpper(name)
+
+					// check aliases
+					switch name {
+					case "", "COPY":
+						name = CodecAny
+					case "MJPEG":
+						name = CodecJPEG
+					case "AAC":
+						name = CodecAAC
+					case "MP3":
+						name = CodecMP3
+					}
+
+					media.Codecs = append(media.Codecs, &Codec{Name: name})
+				}
+
+				medias = append(medias, media)
+			}
+		}
+	}
+
+	return
+}
+
 func atoi(s string) (i int) {
 	i, _ = strconv.Atoi(s)
 	return

pkg/streamer/media_test.go

@@ -3,6 +3,7 @@ package streamer
 import (
 	"github.com/pion/sdp/v3"
 	"github.com/stretchr/testify/assert"
+	"net/url"
 	"testing"
 )
 
@@ -21,3 +22,21 @@ func TestSDP(t *testing.T) {
 	err = sd.Unmarshal(data)
 	assert.Empty(t, err)
 }
+
+func TestParseQuery(t *testing.T) {
+	u, _ := url.Parse("rtsp://localhost:8554/camera1")
+	medias := ParseQuery(u.Query())
+	assert.Nil(t, medias)
+
+	for _, rawULR := range []string{
+		"rtsp://localhost:8554/camera1?video",
+		"rtsp://localhost:8554/camera1?video=copy",
+		"rtsp://localhost:8554/camera1?video=any",
+	} {
+		u, _ = url.Parse(rawULR)
+		medias = ParseQuery(u.Query())
+		assert.Equal(t, []*Media{
+			{Kind: KindVideo, Direction: DirectionRecvonly, Codecs: []*Codec{{Name: CodecAny}}},
+		}, medias)
+	}
+}

pkg/ts/ts.go

@@ -0,0 +1,200 @@
+package ts
+
+import (
+	"bytes"
+	"encoding/hex"
+	"github.com/AlexxIT/go2rtc/pkg/aac"
+	"github.com/AlexxIT/go2rtc/pkg/h264"
+	"github.com/AlexxIT/go2rtc/pkg/streamer"
+	"github.com/deepch/vdk/av"
+	"github.com/deepch/vdk/codec/aacparser"
+	"github.com/deepch/vdk/codec/h264parser"
+	"github.com/deepch/vdk/format/ts"
+	"github.com/pion/rtp"
+	"sync/atomic"
+	"time"
+)
+
+type Consumer struct {
+	streamer.Element
+
+	UserAgent  string
+	RemoteAddr string
+
+	buf      *bytes.Buffer
+	muxer    *ts.Muxer
+	mimeType string
+	streams  []av.CodecData
+	start    bool
+	init     []byte
+
+	send uint32
+}
+
+func (c *Consumer) GetMedias() []*streamer.Media {
+	return []*streamer.Media{
+		{
+			Kind:      streamer.KindVideo,
+			Direction: streamer.DirectionRecvonly,
+			Codecs: []*streamer.Codec{
+				{Name: streamer.CodecH264},
+			},
+		},
+		//{
+		//	Kind:      streamer.KindAudio,
+		//	Direction: streamer.DirectionRecvonly,
+		//	Codecs: []*streamer.Codec{
+		//		{Name: streamer.CodecAAC},
+		//	},
+		//},
+	}
+}
+
+func (c *Consumer) AddTrack(media *streamer.Media, track *streamer.Track) *streamer.Track {
+	codec := track.Codec
+	trackID := int8(len(c.streams))
+
+	switch codec.Name {
+	case streamer.CodecH264:
+		sps, pps := h264.GetParameterSet(codec.FmtpLine)
+		stream, err := h264parser.NewCodecDataFromSPSAndPPS(sps, pps)
+		if err != nil {
+			return nil
+		}
+
+		if len(c.mimeType) > 0 {
+			c.mimeType += ","
+		}
+
+		// TODO: fixme
+		// some devices won't play high level
+		if stream.RecordInfo.AVCLevelIndication <= 0x29 {
+			c.mimeType += "avc1." + h264.GetProfileLevelID(codec.FmtpLine)
+		} else {
+			c.mimeType += "avc1.640029"
+		}
+
+		c.streams = append(c.streams, stream)
+
+		pkt := av.Packet{Idx: trackID, CompositionTime: time.Millisecond}
+
+		ts2time := time.Second / time.Duration(codec.ClockRate)
+
+		push := func(packet *rtp.Packet) error {
+			if packet.Version != h264.RTPPacketVersionAVC {
+				return nil
+			}
+
+			if !c.start {
+				return nil
+			}
+
+			pkt.Data = packet.Payload
+			newTime := time.Duration(packet.Timestamp) * ts2time
+			if pkt.Time > 0 {
+				pkt.Duration = newTime - pkt.Time
+			}
+			pkt.Time = newTime
+
+			if err = c.muxer.WritePacket(pkt); err != nil {
+				return err
+			}
+
+			// clone bytes from buffer, so next packet won't overwrite it
+			buf := append([]byte{}, c.buf.Bytes()...)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			c.buf.Reset()
+
+			return nil
+		}
+
+		if codec.IsRTP() {
+			wrapper := h264.RTPDepay(track)
+			push = wrapper(push)
+		}
+
+		return track.Bind(push)
+
+	case streamer.CodecAAC:
+		s := streamer.Between(codec.FmtpLine, "config=", ";")
+
+		b, err := hex.DecodeString(s)
+		if err != nil {
+			return nil
+		}
+
+		stream, err := aacparser.NewCodecDataFromMPEG4AudioConfigBytes(b)
+		if err != nil {
+			return nil
+		}
+
+		if len(c.mimeType) > 0 {
+			c.mimeType += ","
+		}
+
+		c.mimeType += "mp4a.40.2"
+		c.streams = append(c.streams, stream)
+
+		pkt := av.Packet{Idx: trackID, CompositionTime: time.Millisecond}
+
+		ts2time := time.Second / time.Duration(codec.ClockRate)
+
+		push := func(packet *rtp.Packet) error {
+			if !c.start {
+				return nil
+			}
+
+			pkt.Data = packet.Payload
+			newTime := time.Duration(packet.Timestamp) * ts2time
+			if pkt.Time > 0 {
+				pkt.Duration = newTime - pkt.Time
+			}
+			pkt.Time = newTime
+
+			if err := c.muxer.WritePacket(pkt); err != nil {
+				return err
+			}
+
+			// clone bytes from buffer, so next packet won't overwrite it
+			buf := append([]byte{}, c.buf.Bytes()...)
+			atomic.AddUint32(&c.send, uint32(len(buf)))
+			c.Fire(buf)
+
+			c.buf.Reset()
+
+			return nil
+		}
+
+		if codec.IsRTP() {
+			wrapper := aac.RTPDepay(track)
+			push = wrapper(push)
+		}
+
+		return track.Bind(push)
+	}
+
+	panic("unsupported codec")
+}
+
+func (c *Consumer) MimeCodecs() string {
+	return c.mimeType
+}
+
+func (c *Consumer) Init() ([]byte, error) {
+	c.buf = bytes.NewBuffer(nil)
+	c.muxer = ts.NewMuxer(c.buf)
+
+	// first packet will be with header, it's ok
+	if err := c.muxer.WriteHeader(c.streams); err != nil {
+		return nil, err
+	}
+	data := append([]byte{}, c.buf.Bytes()...)
+
+	return data, nil
+}
+
+func (c *Consumer) Start() {
+	c.start = true
+}

www/README.md

@@ -56,3 +56,5 @@ pc.ontrack = ev => {
 - https://web.dev/i18n/en/fast-playback-with-preload/#manual_buffering
 - https://developer.mozilla.org/en-US/docs/Web/API/Media_Source_Extensions_API
 - https://chromium.googlesource.com/external/w3c/web-platform-tests/+/refs/heads/master/media-source/mediasource-is-type-supported.html
+- https://googlechrome.github.io/samples/media/sourcebuffer-changetype.html
+- https://chromestatus.com/feature/5100845653819392

www/codecs.html

@@ -53,7 +53,7 @@
     const video = document.createElement("video");
     out.innerText += "video.canPlayType\n";
     types.forEach(type => {
-        out.innerText += type + "=" + (video.canPlayType(type) ? "true" : "false") + "\n";
+        out.innerText += `${type} = ${MediaSource.isTypeSupported(type)} / ${video.canPlayType(type)}\n`;
     })
 
 </script>

www/index.html

@@ -89,9 +89,7 @@
     const templates = [
         '<a href="stream.html?src={name}">stream</a>',
         '<a href="webrtc.html?src={name}">2-way-aud</a>',
-        '<a href="api/stream.mp4?src={name}">mp4</a>',
-        '<a href="api/stream.mjpeg?src={name}">mjpeg</a>',
-        '<a href="api/streams?src={name}">info</a>',
+        '<a href="links.html?src={name}">links</a>',
         '<a href="#" data-name="{name}">delete</a>',
     ];
 
@@ -138,15 +136,17 @@
 
             for (const [name, value] of Object.entries(data)) {
                 const online = value && value.consumers ? value.consumers.length : 0;
+                const src = encodeURIComponent(name);
                 const links = templates.map(link => {
-                    return link.replace("{name}", encodeURIComponent(name));
+                    return link.replace("{name}", src);
                 }).join(" ");
 
                 const tr = document.createElement("tr");
                 tr.dataset["id"] = name;
                 tr.innerHTML =
                     `<td><label><input type="checkbox" name="${name}">${name}</label></td>` +
-                    `<td>${online}</td><td>${links}</td>`;
+                    `<td><a href="api/streams?src=${src}">${online} / info</a></td>` +
+                    `<td>${links}</td>`;
                 tbody.appendChild(tr);
             }
         });
@@ -156,17 +156,9 @@
     fetch(url, {cache: 'no-cache'}).then(r => r.json()).then(data => {
         const info = document.querySelector(".info");
         info.innerText = `Version: ${data.version}, Config: ${data.config_path}`;
-
-        try {
-            const host = data.host.match(/^[^:]+/)[0];
-            const port = data.rtsp.listen.match(/[0-9]+$/)[0];
-            templates.splice(4, 0, `<a href="rtsp://${host}:${port}/{name}">rtsp</a>`);
-        } catch (e) {
-            templates.splice(4, 0, `<a href="rtsp://${location.host}:8554/{name}">rtsp</a>`);
-        }
-
-        reload();
     });
+
+    reload();
 </script>
 </body>
 </html>

www/links.html

@@ -0,0 +1,89 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>go2rtc - links</title>
+    <meta name="viewport" content="width=device-width, user-scalable=yes, initial-scale=1, maximum-scale=1">
+    <meta http-equiv="X-UA-Compatible" content="ie=edge">
+    <style>
+        body {
+            font-family: Arial, Helvetica, sans-serif;
+        }
+
+        body {
+            margin: 0;
+            padding: 0;
+            display: flex;
+            flex-direction: column;
+        }
+
+        html, body, #config {
+            width: 100%;
+            height: 100%;
+        }
+
+        div {
+            padding: 10px;
+        }
+
+        div > li {
+            list-style-type: none;
+            padding-left: 10px;
+            position: relative;
+        }
+
+        div > li:before {
+            content: "-";
+            position: absolute;
+            left: 0;
+        }
+    </style>
+</head>
+<body>
+<script src="main.js"></script>
+<div id="links"></div>
+<script>
+    const params = new URLSearchParams(location.search);
+    const src = params.get("src");
+
+    const links = document.querySelector("#links");
+    links.innerHTML = `
+        <h2>Any codec in source</h2>
+        <li><a href="stream.html?src=${src}">stream.html</a> with auto-select mode / browsers: all / codecs: H264, H265*, MJPEG, JPEG, AAC, PCMU, PCMA, OPUS</li>
+        <li><a href="api/streams?src=${src}">info.json</a> page with active connections</li>
+    `;
+
+    const url = new URL('api', location.href);
+    fetch(url, {cache: 'no-cache'}).then(r => r.json()).then(data => {
+        let rtsp = location.host + ':8554';
+        try {
+            const host = data.host.match(/^[^:]+/)[0];
+            const port = data.rtsp.listen.match(/[0-9]+$/)[0];
+            rtsp = `${host}:${port}`;
+        } catch (e) {
+        }
+
+        const links = document.querySelector("#links");
+        links.innerHTML += `
+            <li><a href="rtsp://${rtsp}/${src}">rtsp</a> with only one video and one audio / codecs: any</li>
+            <li><a href="rtsp://${rtsp}/${src}?mp4">rtsp</a> for MP4 recording (Hass or Frigate) / codecs: H264, H265, AAC</li>
+            <li><a href="rtsp://${rtsp}/${src}?video=all&audio=all">rtsp</a> with all tracks / codecs: any</li>
+
+            <h2>H264/H265 source</h2>
+            <li><a href="stream.html?src=${src}&mode=webrtc">stream.html</a> WebRTC stream / browsers: all / codecs: H264, PCMU, PCMA, OPUS / +H265 in Safari</li>
+            <li><a href="stream.html?src=${src}&mode=mse">stream.html</a> MSE stream / browsers: Chrome, Firefox, Safari Mac/iPad / codecs: H264, H265*, AAC / +OPUS in Chrome and Firefox</li>
+            <li><a href="api/stream.mp4?src=${src}">stream.mp4</a> MP4 stream with AAC audio / browsers: Chrome, Firefox / codecs: H264, H265*, AAC</li>
+            <li><a href="api/stream.mp4?src=${src}&video=h264,h265&audio=aac,opus,mp3,pcma,pcmu">stream.mp4</a> MP4 stream with any audio / browsers: Chrome / codecs: H264, H265*, AAC, OPUS, MP3, PCMU, PCMA</li>
+            <li><a href="api/frame.mp4?src=${src}">frame.mp4</a> snapshot in MP4-format / browsers: all / codecs: H264, H265*</li>
+            <li><a href="api/stream.m3u8?src=${src}">stream.m3u8</a> HLS/TS / browsers: Safari all, Chrome Android / codecs: H264</li>
+            <li><a href="api/stream.m3u8?src=${src}&mp4">stream.m3u8</a> HLS/fMP4 / browsers: Safari all, Chrome Android / codecs: H264, H265*, AAC</li>
+            <li><a href="webrtc.html?src=${src}">webrtc.html</a> with two-way audio for supported cameras / browsers: all / codecs: H264, PCMU, PCMA, OPUS</li>
+
+            <h2>MJPEG source</h2>
+            <li><a href="stream.html?src=${src}&mode=mjpeg">stream.html</a> with MJPEG mode / browsers: all / codecs: MJPEG, JPEG</li>
+            <li><a href="api/stream.mjpeg?src=${src}">stream.mjpeg</a> MJPEG stream / browsers: all / codecs: MJPEG, JPEG</li>
+            <li><a href="api/frame.jpeg?src=${src}">frame.jpeg</a> snapshot in JPEG-format / browsers: all / codecs: MJPEG, JPEG</li>
+        `;
+    });
+</script>
+</body>
+</html>

www/video-rtc.js

@@ -26,8 +26,7 @@ export class VideoRTC extends HTMLElement {
             "hvc1.1.6.L153.B0", // H.265 main 5.1 (Chromecast Ultra)
             "mp4a.40.2",        // AAC LC
             "mp4a.40.5",        // AAC HE
-            "mp4a.69",          // MP3
-            "mp4a.6B",          // MP3
+            "opus",             // OPUS Chrome
         ];
 
         /**
@@ -395,11 +394,11 @@ export class VideoRTC extends HTMLElement {
                         bufLen = 0;
                         sb.appendBuffer(data);
                     } else if (sb.buffered && sb.buffered.length) {
-                        const end = sb.buffered.end(sb.buffered.length - 1) - 5;
+                        const end = sb.buffered.end(sb.buffered.length - 1) - 15;
                         const start = sb.buffered.start(0);
                         if (end > start) {
                             sb.remove(start, end);
-                            ms.setLiveSeekableRange(end, end + 5);
+                            ms.setLiveSeekableRange(end, end + 15);
                         }
                         // console.debug("VideoRTC.buffered", start, end);
                     }