Send INIT as a single message.
Much simpler on both the client and server side.
This commit is contained in:
Luke Curley
parent
e9663accc6
commit
d7237c4926
|
|
@ -1,121 +0,0 @@
|
|||
import * as Message from "./message";
|
||||
import * as MP4 from "../mp4"
|
||||
import * as Stream from "../stream"
|
||||
import * as Util from "../util"
|
||||
|
||||
import Renderer from "./renderer"
|
||||
|
||||
export default class Decoder {
|
||||
// Store the init message for each track
|
||||
tracks: Map<string, Util.Deferred<Message.Init>>;
|
||||
decoder: AudioDecoder; // TODO one per track
|
||||
sync: Message.Sync;
|
||||
|
||||
constructor(config: Message.Config, renderer: Renderer) {
|
||||
this.tracks = new Map();
|
||||
|
||||
this.decoder = new AudioDecoder({
|
||||
output: renderer.emit.bind(renderer),
|
||||
error: console.warn,
|
||||
});
|
||||
}
|
||||
|
||||
init(msg: Message.Init) {
|
||||
let defer = this.tracks.get(msg.track);
|
||||
if (!defer) {
|
||||
defer = new Util.Deferred()
|
||||
this.tracks.set(msg.track, defer)
|
||||
}
|
||||
|
||||
if (msg.info.audioTracks.length != 1 || msg.info.videoTracks.length != 0) {
|
||||
throw new Error("Expected a single audio track")
|
||||
}
|
||||
|
||||
const track = msg.info.audioTracks[0]
|
||||
const audio = track.audio
|
||||
|
||||
defer.resolve(msg)
|
||||
}
|
||||
|
||||
async decode(msg: Message.Segment) {
|
||||
let track = this.tracks.get(msg.track);
|
||||
if (!track) {
|
||||
track = new Util.Deferred()
|
||||
this.tracks.set(msg.track, track)
|
||||
}
|
||||
|
||||
// Wait for the init segment to be fully received and parsed
|
||||
const init = await track.promise;
|
||||
const audio = init.info.audioTracks[0]
|
||||
|
||||
if (this.decoder.state == "unconfigured") {
|
||||
this.decoder.configure({
|
||||
codec: audio.codec,
|
||||
numberOfChannels: audio.audio.channel_count,
|
||||
sampleRate: audio.audio.sample_rate,
|
||||
})
|
||||
}
|
||||
|
||||
const input = MP4.New();
|
||||
|
||||
input.onSamples = (id: number, user: any, samples: MP4.Sample[]) => {
|
||||
for (let sample of samples) {
|
||||
// Convert to microseconds
|
||||
const timestamp = 1000 * 1000 * sample.dts / sample.timescale
|
||||
const duration = 1000 * 1000 * sample.duration / sample.timescale
|
||||
|
||||
// This assumes that timescale == sample rate
|
||||
this.decoder.decode(new EncodedAudioChunk({
|
||||
type: sample.is_sync ? "key" : "delta",
|
||||
data: sample.data,
|
||||
duration: duration,
|
||||
timestamp: timestamp,
|
||||
}))
|
||||
}
|
||||
}
|
||||
|
||||
input.onReady = (info: any) => {
|
||||
input.setExtractionOptions(info.tracks[0].id, {}, { nbSamples: 1 });
|
||||
input.start();
|
||||
}
|
||||
|
||||
// MP4box requires us to reparse the init segment unfortunately
|
||||
let offset = 0;
|
||||
|
||||
for (let raw of init.raw) {
|
||||
raw.fileStart = offset
|
||||
input.appendBuffer(raw)
|
||||
}
|
||||
|
||||
const stream = new Stream.Reader(msg.reader, msg.buffer)
|
||||
|
||||
/* TODO I'm not actually sure why this code doesn't work; something trips up the MP4 parser
|
||||
while (1) {
|
||||
const data = await stream.read()
|
||||
if (!data) break
|
||||
|
||||
input.appendBuffer(data)
|
||||
input.flush()
|
||||
}
|
||||
*/
|
||||
|
||||
// One day I'll figure it out; until then read one top-level atom at a time
|
||||
while (!await stream.done()) {
|
||||
const raw = await stream.peek(4)
|
||||
const size = new DataView(raw.buffer, raw.byteOffset, raw.byteLength).getUint32(0)
|
||||
const atom = await stream.bytes(size)
|
||||
|
||||
// Make a copy of the atom because mp4box only accepts an ArrayBuffer unfortunately
|
||||
let box = new Uint8Array(atom.byteLength);
|
||||
box.set(atom)
|
||||
|
||||
// and for some reason we need to modify the underlying ArrayBuffer with offset
|
||||
let buffer = box.buffer as MP4.ArrayBuffer
|
||||
buffer.fileStart = offset
|
||||
|
||||
// Parse the data
|
||||
offset = input.appendBuffer(buffer)
|
||||
input.flush()
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -1,8 +1,8 @@
|
|||
import * as Message from "./message"
|
||||
import Renderer from "./renderer"
|
||||
import Renderer from "../media/audio"
|
||||
import Decoder from "./decoder"
|
||||
|
||||
import { RingInit } from "./ring"
|
||||
import { RingInit } from "../media/ring"
|
||||
|
||||
// Abstracts the Worker and Worklet into a simpler API
|
||||
// This class must be created on the main thread due to AudioContext.
|
||||
|
|
|
|||
|
|
@ -1,85 +0,0 @@
|
|||
import * as Message from "./message"
|
||||
import { Ring } from "./ring"
|
||||
|
||||
export default class Renderer {
|
||||
ring: Ring;
|
||||
queue: Array<AudioData>;
|
||||
sync?: DOMHighResTimeStamp
|
||||
running: number;
|
||||
|
||||
constructor(config: Message.Config) {
|
||||
this.ring = new Ring(config.ring)
|
||||
this.queue = [];
|
||||
this.running = 0
|
||||
}
|
||||
|
||||
emit(frame: AudioData) {
|
||||
if (!this.sync) {
|
||||
// Save the frame as the sync point
|
||||
this.sync = 1000 * performance.now() - frame.timestamp
|
||||
}
|
||||
|
||||
// Insert the frame into the queue sorted by timestamp.
|
||||
if (this.queue.length > 0 && this.queue[this.queue.length-1].timestamp <= frame.timestamp) {
|
||||
// Fast path because we normally append to the end.
|
||||
this.queue.push(frame)
|
||||
} else {
|
||||
// Do a full binary search
|
||||
let low = 0
|
||||
let high = this.queue.length;
|
||||
|
||||
while (low < high) {
|
||||
var mid = (low + high) >>> 1;
|
||||
if (this.queue[mid].timestamp < frame.timestamp) low = mid + 1;
|
||||
else high = mid;
|
||||
}
|
||||
|
||||
this.queue.splice(low, 0, frame)
|
||||
}
|
||||
|
||||
if (!this.running) {
|
||||
// Wait for the next animation frame
|
||||
this.running = self.requestAnimationFrame(this.render.bind(this))
|
||||
}
|
||||
}
|
||||
|
||||
render() {
|
||||
// Determine the target timestamp.
|
||||
const target = 1000 * performance.now() - this.sync!
|
||||
|
||||
// Check if we should skip some frames
|
||||
while (this.queue.length) {
|
||||
const next = this.queue[0]
|
||||
if (next.timestamp >= target) {
|
||||
break
|
||||
}
|
||||
|
||||
console.warn("dropping audio")
|
||||
|
||||
this.queue.shift()
|
||||
next.close()
|
||||
}
|
||||
|
||||
// Push as many as we can to the ring buffer.
|
||||
while (this.queue.length) {
|
||||
let frame = this.queue[0]
|
||||
let ok = this.ring.write(frame)
|
||||
if (!ok) {
|
||||
break
|
||||
}
|
||||
|
||||
frame.close()
|
||||
this.queue.shift()
|
||||
}
|
||||
|
||||
if (this.queue.length) {
|
||||
this.running = self.requestAnimationFrame(this.render.bind(this))
|
||||
} else {
|
||||
this.running = 0
|
||||
}
|
||||
}
|
||||
|
||||
play(play: Message.Play) {
|
||||
this.ring.reset()
|
||||
}
|
||||
}
|
||||
|
|
@ -1,26 +0,0 @@
|
|||
import Decoder from "./decoder"
|
||||
import Renderer from "./renderer"
|
||||
|
||||
import * as Message from "./message"
|
||||
|
||||
let decoder: Decoder
|
||||
let renderer: Renderer;
|
||||
|
||||
self.addEventListener('message', (e: MessageEvent) => {
|
||||
if (e.data.config) {
|
||||
renderer = new Renderer(e.data.config)
|
||||
decoder = new Decoder(e.data.config, renderer)
|
||||
}
|
||||
|
||||
if (e.data.init) {
|
||||
decoder.init(e.data.init)
|
||||
}
|
||||
|
||||
if (e.data.segment) {
|
||||
decoder.decode(e.data.segment)
|
||||
}
|
||||
|
||||
if (e.data.play) {
|
||||
renderer.play(e.data.play)
|
||||
}
|
||||
})
|
||||
|
|
@ -0,0 +1,160 @@
|
|||
import * as Message from "./message";
|
||||
import * as MP4 from "../mp4"
|
||||
import * as Stream from "../stream"
|
||||
import * as Util from "../util"
|
||||
|
||||
import Renderer from "./renderer"
|
||||
|
||||
export default class Decoder {
|
||||
init: MP4.InitParser;
|
||||
decoders: Map<number, AudioDecoder | VideoDecoder>;
|
||||
renderer: Renderer;
|
||||
|
||||
constructor(renderer: Renderer) {
|
||||
this.init = new MP4.InitParser();
|
||||
this.decoders = new Map();
|
||||
this.renderer = renderer;
|
||||
}
|
||||
|
||||
async receiveInit(msg: Message.Init) {
|
||||
let stream = new Stream.Reader(msg.reader, msg.buffer);
|
||||
while (1) {
|
||||
const data = await stream.read()
|
||||
if (!data) break
|
||||
|
||||
this.init.push(data)
|
||||
}
|
||||
|
||||
// TODO make sure the init segment is fully received
|
||||
}
|
||||
|
||||
async receiveSegment(msg: Message.Segment) {
|
||||
// Wait for the init segment to be fully received and parsed
|
||||
const info = await this.init.info
|
||||
const input = MP4.New();
|
||||
|
||||
input.onSamples = this.onSamples.bind(this);
|
||||
input.onReady = (info: any) => {
|
||||
// Extract all of the tracks, because we don't know if it's audio or video.
|
||||
for (let track of info.tracks) {
|
||||
input.setExtractionOptions(track.id, track, { nbSamples: 1 });
|
||||
}
|
||||
|
||||
input.start();
|
||||
}
|
||||
|
||||
// MP4box requires us to reparse the init segment unfortunately
|
||||
let offset = 0;
|
||||
|
||||
for (let raw of this.init.raw) {
|
||||
raw.fileStart = offset
|
||||
offset = input.appendBuffer(raw)
|
||||
}
|
||||
|
||||
const stream = new Stream.Reader(msg.reader, msg.buffer)
|
||||
|
||||
// For whatever reason, mp4box doesn't work until you read an atom at a time.
|
||||
while (!await stream.done()) {
|
||||
const raw = await stream.peek(4)
|
||||
|
||||
// TODO this doesn't support when size = 0 (until EOF) or size = 1 (extended size)
|
||||
const size = new DataView(raw.buffer, raw.byteOffset, raw.byteLength).getUint32(0)
|
||||
const atom = await stream.bytes(size)
|
||||
|
||||
// Make a copy of the atom because mp4box only accepts an ArrayBuffer unfortunately
|
||||
let box = new Uint8Array(atom.byteLength);
|
||||
box.set(atom)
|
||||
|
||||
// and for some reason we need to modify the underlying ArrayBuffer with offset
|
||||
let buffer = box.buffer as MP4.ArrayBuffer
|
||||
buffer.fileStart = offset
|
||||
|
||||
// Parse the data
|
||||
offset = input.appendBuffer(buffer)
|
||||
input.flush()
|
||||
}
|
||||
}
|
||||
|
||||
onSamples(track_id: number, track: MP4.Track, samples: MP4.Sample[]) {
|
||||
let decoder = this.decoders.get(track_id);
|
||||
|
||||
if (!decoder) {
|
||||
// We need a sample to initalize the video decoder, because of mp4box limitations.
|
||||
let sample = samples[0];
|
||||
|
||||
if (MP4.isVideoTrack(track)) {
|
||||
// Configure the decoder using the AVC box for H.264
|
||||
// TODO it should be easy to support other codecs, just need to know the right boxes.
|
||||
const avcc = sample.description.avcC;
|
||||
if (!avcc) throw new Error("TODO only h264 is supported");
|
||||
|
||||
const description = new MP4.Stream(new Uint8Array(avcc.size), 0, false)
|
||||
avcc.write(description)
|
||||
|
||||
const videoDecoder = new VideoDecoder({
|
||||
output: this.renderer.push.bind(this.renderer),
|
||||
error: console.warn,
|
||||
});
|
||||
|
||||
videoDecoder.configure({
|
||||
codec: track.codec,
|
||||
codedHeight: track.video.height,
|
||||
codedWidth: track.video.width,
|
||||
description: description.buffer?.slice(8),
|
||||
// optimizeForLatency: true
|
||||
})
|
||||
|
||||
decoder = videoDecoder
|
||||
} else if (MP4.isAudioTrack(track)) {
|
||||
const audioDecoder = new AudioDecoder({
|
||||
output: this.renderer.push.bind(this.renderer),
|
||||
error: console.warn,
|
||||
});
|
||||
|
||||
audioDecoder.configure({
|
||||
codec: track.codec,
|
||||
numberOfChannels: track.audio.channel_count,
|
||||
sampleRate: track.audio.sample_rate,
|
||||
})
|
||||
|
||||
decoder = audioDecoder
|
||||
} else {
|
||||
throw new Error("unknown track type")
|
||||
}
|
||||
|
||||
this.decoders.set(track_id, decoder)
|
||||
}
|
||||
|
||||
for (let sample of samples) {
|
||||
// Convert to microseconds
|
||||
const timestamp = 1000 * 1000 * sample.dts / sample.timescale
|
||||
const duration = 1000 * 1000 * sample.duration / sample.timescale
|
||||
|
||||
if (isAudioDecoder(decoder)) {
|
||||
decoder.decode(new EncodedAudioChunk({
|
||||
type: sample.is_sync ? "key" : "delta",
|
||||
data: sample.data,
|
||||
duration: duration,
|
||||
timestamp: timestamp,
|
||||
}))
|
||||
} else if (isVideoDecoder(decoder)) {
|
||||
decoder.decode(new EncodedVideoChunk({
|
||||
type: sample.is_sync ? "key" : "delta",
|
||||
data: sample.data,
|
||||
duration: duration,
|
||||
timestamp: timestamp,
|
||||
}))
|
||||
} else {
|
||||
throw new Error("unknown decoder type")
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
function isAudioDecoder(decoder: AudioDecoder | VideoDecoder): decoder is AudioDecoder {
|
||||
return decoder instanceof AudioDecoder
|
||||
}
|
||||
|
||||
function isVideoDecoder(decoder: AudioDecoder | VideoDecoder): decoder is VideoDecoder {
|
||||
return decoder instanceof VideoDecoder
|
||||
}
|
||||
|
|
@ -0,0 +1,82 @@
|
|||
import * as Message from "./message"
|
||||
import { RingInit } from "./ring"
|
||||
|
||||
// Abstracts the Worker and Worklet into a simpler API
|
||||
// This class must be created on the main thread due to AudioContext.
|
||||
export default class Media {
|
||||
context: AudioContext;
|
||||
worker: Worker;
|
||||
worklet: Promise<AudioWorkletNode>;
|
||||
|
||||
constructor(videoConfig: Message.VideoConfig) {
|
||||
// Assume 44.1kHz and two audio channels
|
||||
const audioConfig = {
|
||||
sampleRate: 44100,
|
||||
ring: new RingInit(2, 4410), // 100ms at 44.1khz
|
||||
}
|
||||
|
||||
const config = {
|
||||
audio: audioConfig,
|
||||
video: videoConfig,
|
||||
}
|
||||
|
||||
this.context = new AudioContext({
|
||||
latencyHint: "interactive",
|
||||
sampleRate: config.audio.sampleRate,
|
||||
})
|
||||
|
||||
|
||||
this.worker = this.setupWorker(config)
|
||||
this.worklet = this.setupWorklet(config)
|
||||
}
|
||||
|
||||
init(init: Message.Init) {
|
||||
this.worker.postMessage({ init }, [ init.buffer.buffer, init.reader ])
|
||||
}
|
||||
|
||||
segment(segment: Message.Segment) {
|
||||
this.worker.postMessage({ segment }, [ segment.buffer.buffer, segment.reader ])
|
||||
}
|
||||
|
||||
play(play: Message.Play) {
|
||||
this.context.resume()
|
||||
//this.worker.postMessage({ play })
|
||||
}
|
||||
|
||||
private setupWorker(config: Message.Config): Worker {
|
||||
const url = new URL('worker.ts', import.meta.url)
|
||||
|
||||
const worker = new Worker(url, {
|
||||
type: "module",
|
||||
name: "media",
|
||||
})
|
||||
|
||||
worker.postMessage({ config }, [ config.video.canvas ])
|
||||
|
||||
return worker
|
||||
}
|
||||
|
||||
private async setupWorklet(config: Message.Config): Promise<AudioWorkletNode> {
|
||||
// Load the worklet source code.
|
||||
const url = new URL('worklet.ts', import.meta.url)
|
||||
await this.context.audioWorklet.addModule(url)
|
||||
|
||||
const volume = this.context.createGain()
|
||||
volume.gain.value = 2.0;
|
||||
|
||||
// Create a worklet
|
||||
const worklet = new AudioWorkletNode(this.context, 'renderer');
|
||||
worklet.onprocessorerror = (e: Event) => {
|
||||
console.error("Audio worklet error:", e)
|
||||
};
|
||||
|
||||
worklet.port.postMessage({ config })
|
||||
|
||||
// Connect the worklet to the volume node and then to the speakers
|
||||
worklet.connect(volume)
|
||||
volume.connect(this.context.destination)
|
||||
|
||||
return worklet
|
||||
}
|
||||
|
||||
}
|
||||
|
|
@ -1,28 +1,29 @@
|
|||
import * as MP4 from "../mp4"
|
||||
import { RingInit } from "./ring"
|
||||
import { RingInit } from "../media/ring"
|
||||
|
||||
export interface Config {
|
||||
audio: AudioConfig;
|
||||
video: VideoConfig;
|
||||
}
|
||||
|
||||
export interface VideoConfig {
|
||||
canvas: OffscreenCanvas;
|
||||
}
|
||||
|
||||
export interface AudioConfig {
|
||||
// audio stuff
|
||||
sampleRate: number;
|
||||
ring: RingInit;
|
||||
}
|
||||
|
||||
export interface Init {
|
||||
track: string;
|
||||
info: MP4.Info;
|
||||
raw: MP4.ArrayBuffer[];
|
||||
}
|
||||
|
||||
export interface Segment {
|
||||
track: string;
|
||||
buffer: Uint8Array; // unread buffered data
|
||||
reader: ReadableStream; // unread unbuffered data
|
||||
}
|
||||
|
||||
// Audio tells video when the given timestamp should be rendered.
|
||||
export interface Sync {
|
||||
origin: number;
|
||||
clock: DOMHighResTimeStamp;
|
||||
timestamp: number;
|
||||
export interface Segment {
|
||||
buffer: Uint8Array; // unread buffered data
|
||||
reader: ReadableStream; // unread unbuffered data
|
||||
}
|
||||
|
||||
export interface Play {
|
||||
|
|
@ -0,0 +1,138 @@
|
|||
import * as Message from "./message";
|
||||
import { Ring } from "./ring"
|
||||
|
||||
export default class Renderer {
|
||||
audioRing: Ring;
|
||||
audioQueue: Array<AudioData>;
|
||||
|
||||
videoCanvas: OffscreenCanvas;
|
||||
videoQueue: Array<VideoFrame>;
|
||||
|
||||
render: number; // non-zero if requestAnimationFrame has been called
|
||||
sync?: DOMHighResTimeStamp; // the wall clock value for timestamp 0, in microseconds
|
||||
last?: number; // the timestamp of the last rendered frame, in microseconds
|
||||
|
||||
constructor(config: Message.Config) {
|
||||
this.audioRing = new Ring(config.audio.ring);
|
||||
this.audioQueue = [];
|
||||
|
||||
this.videoCanvas = config.video.canvas;
|
||||
this.videoQueue = [];
|
||||
|
||||
this.render = 0;
|
||||
}
|
||||
|
||||
push(frame: AudioData | VideoFrame) {
|
||||
if (!this.sync) {
|
||||
// Save the frame as the sync point
|
||||
this.sync = 1000 * performance.now() - frame.timestamp
|
||||
}
|
||||
|
||||
// Drop any old frames
|
||||
if (this.last && frame.timestamp <= this.last) {
|
||||
frame.close()
|
||||
return
|
||||
}
|
||||
|
||||
let queue
|
||||
if (isAudioData(frame)) {
|
||||
queue = this.audioQueue;
|
||||
} else if (isVideoFrame(frame)) {
|
||||
queue = this.videoQueue;
|
||||
} else {
|
||||
throw new Error("unknown frame type")
|
||||
}
|
||||
|
||||
// Insert the frame into the queue sorted by timestamp.
|
||||
if (queue.length > 0 && queue[queue.length-1].timestamp <= frame.timestamp) {
|
||||
// Fast path because we normally append to the end.
|
||||
queue.push(frame as any)
|
||||
} else {
|
||||
// Do a full binary search
|
||||
let low = 0
|
||||
let high = queue.length;
|
||||
|
||||
while (low < high) {
|
||||
var mid = (low + high) >>> 1;
|
||||
if (queue[mid].timestamp < frame.timestamp) low = mid + 1;
|
||||
else high = mid;
|
||||
}
|
||||
|
||||
queue.splice(low, 0, frame as any)
|
||||
}
|
||||
|
||||
// Queue up to render the next frame.
|
||||
if (!this.render) {
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
}
|
||||
}
|
||||
|
||||
draw(now: DOMHighResTimeStamp) {
|
||||
// Determine the target timestamp.
|
||||
const target = 1000 * now - this.sync!
|
||||
|
||||
this.drawAudio(now, target)
|
||||
this.drawVideo(now, target)
|
||||
|
||||
if (this.audioQueue.length || this.videoQueue.length) {
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
} else {
|
||||
this.render = 0
|
||||
}
|
||||
}
|
||||
|
||||
drawAudio(now: DOMHighResTimeStamp, target: DOMHighResTimeStamp) {
|
||||
// Check if we should skip some frames
|
||||
while (this.audioQueue.length) {
|
||||
const next = this.audioQueue[0]
|
||||
if (next.timestamp >= target) {
|
||||
let ok = this.audioRing.write(next)
|
||||
if (!ok) {
|
||||
// No more space in the ring
|
||||
break
|
||||
}
|
||||
} else {
|
||||
console.warn("dropping audio")
|
||||
}
|
||||
|
||||
next.close()
|
||||
this.audioQueue.shift()
|
||||
}
|
||||
}
|
||||
|
||||
drawVideo(now: DOMHighResTimeStamp, target: DOMHighResTimeStamp) {
|
||||
if (this.videoQueue.length == 0) return;
|
||||
|
||||
let frame = this.videoQueue[0];
|
||||
if (frame.timestamp >= target) {
|
||||
// nothing to render yet, wait for the next animation frame
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
return
|
||||
}
|
||||
|
||||
this.videoQueue.shift();
|
||||
|
||||
// Check if we should skip some frames
|
||||
while (this.videoQueue.length) {
|
||||
const next = this.videoQueue[0]
|
||||
if (next.timestamp > target) break
|
||||
|
||||
frame.close()
|
||||
frame = this.videoQueue.shift()!;
|
||||
}
|
||||
|
||||
const ctx = this.videoCanvas.getContext("2d");
|
||||
ctx!.drawImage(frame, 0, 0, this.videoCanvas.width, this.videoCanvas.height) // TODO aspect ratio
|
||||
|
||||
this.last = frame.timestamp;
|
||||
frame.close()
|
||||
}
|
||||
}
|
||||
|
||||
function isAudioData(frame: AudioData | VideoFrame): frame is AudioData {
|
||||
return frame instanceof AudioData
|
||||
}
|
||||
|
||||
function isVideoFrame(frame: AudioData | VideoFrame): frame is VideoFrame {
|
||||
return frame instanceof VideoFrame
|
||||
}
|
||||
|
|
@ -13,10 +13,10 @@ self.addEventListener('message', async (e: MessageEvent) => {
|
|||
decoder = new Decoder(renderer)
|
||||
} else if (e.data.init) {
|
||||
const init = e.data.init as Message.Init
|
||||
await decoder.init(init)
|
||||
await decoder.receiveInit(init)
|
||||
} else if (e.data.segment) {
|
||||
const segment = e.data.segment as Message.Segment
|
||||
await decoder.decode(segment)
|
||||
await decoder.receiveSegment(segment)
|
||||
}
|
||||
})
|
||||
|
||||
|
|
@ -25,7 +25,7 @@ class Renderer extends AudioWorkletProcessor {
|
|||
}
|
||||
|
||||
config(config: Message.Config) {
|
||||
this.ring = new Ring(config.ring)
|
||||
this.ring = new Ring(config.audio.ring)
|
||||
}
|
||||
|
||||
// Inputs and outputs in groups of 128 samples.
|
||||
|
|
@ -1,11 +1,12 @@
|
|||
// Rename some stuff so it's on brand.
|
||||
export {
|
||||
createFile as New,
|
||||
MP4File as File,
|
||||
MP4ArrayBuffer as ArrayBuffer,
|
||||
MP4Info as Info,
|
||||
DataStream as Stream,
|
||||
Sample,
|
||||
} from "mp4box"
|
||||
import * as MP4 from "./rename"
|
||||
export * from "./rename"
|
||||
|
||||
export { Init, InitParser } from "./init"
|
||||
export { Init, InitParser } from "./init"
|
||||
|
||||
export function isAudioTrack(track: MP4.Track): track is MP4.AudioTrack {
|
||||
return (track as MP4.AudioTrack).audio !== undefined;
|
||||
}
|
||||
|
||||
export function isVideoTrack(track: MP4.Track): track is MP4.VideoTrack {
|
||||
return (track as MP4.VideoTrack).video !== undefined;
|
||||
}
|
||||
|
|
@ -20,19 +20,7 @@ export class InitParser {
|
|||
// Create a promise that gets resolved once the init segment has been parsed.
|
||||
this.info = new Promise((resolve, reject) => {
|
||||
this.mp4box.onError = reject
|
||||
|
||||
// https://github.com/gpac/mp4box.js#onreadyinfo
|
||||
this.mp4box.onReady = (info: MP4.Info) => {
|
||||
if (!info.isFragmented) {
|
||||
reject("expected a fragmented mp4")
|
||||
}
|
||||
|
||||
if (info.tracks.length != 1) {
|
||||
reject("expected a single track")
|
||||
}
|
||||
|
||||
resolve(info)
|
||||
}
|
||||
this.mp4box.onReady = resolve
|
||||
})
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -1,7 +1,7 @@
|
|||
// https://github.com/gpac/mp4box.js/issues/233
|
||||
|
||||
declare module "mp4box" {
|
||||
interface MP4MediaTrack {
|
||||
export interface MP4MediaTrack {
|
||||
id: number;
|
||||
created: Date;
|
||||
modified: Date;
|
||||
|
|
@ -19,26 +19,26 @@ declare module "mp4box" {
|
|||
nb_samples: number;
|
||||
}
|
||||
|
||||
interface MP4VideoData {
|
||||
export interface MP4VideoData {
|
||||
width: number;
|
||||
height: number;
|
||||
}
|
||||
|
||||
interface MP4VideoTrack extends MP4MediaTrack {
|
||||
export interface MP4VideoTrack extends MP4MediaTrack {
|
||||
video: MP4VideoData;
|
||||
}
|
||||
|
||||
interface MP4AudioData {
|
||||
export interface MP4AudioData {
|
||||
sample_rate: number;
|
||||
channel_count: number;
|
||||
sample_size: number;
|
||||
}
|
||||
|
||||
interface MP4AudioTrack extends MP4MediaTrack {
|
||||
export interface MP4AudioTrack extends MP4MediaTrack {
|
||||
audio: MP4AudioData;
|
||||
}
|
||||
|
||||
type MP4Track = MP4VideoTrack | MP4AudioTrack;
|
||||
export type MP4Track = MP4VideoTrack | MP4AudioTrack;
|
||||
|
||||
export interface MP4Info {
|
||||
duration: number;
|
||||
|
|
|
|||
|
|
@ -0,0 +1,12 @@
|
|||
// Rename some stuff so it's on brand.
|
||||
export {
|
||||
createFile as New,
|
||||
MP4File as File,
|
||||
MP4ArrayBuffer as ArrayBuffer,
|
||||
MP4Info as Info,
|
||||
MP4Track as Track,
|
||||
MP4AudioTrack as AudioTrack,
|
||||
MP4VideoTrack as VideoTrack,
|
||||
DataStream as Stream,
|
||||
Sample,
|
||||
} from "mp4box"
|
||||
|
|
@ -1,6 +1,5 @@
|
|||
import Audio from "../audio"
|
||||
import Transport from "../transport"
|
||||
import Video from "../video"
|
||||
import Media from "../media"
|
||||
|
||||
export interface PlayerInit {
|
||||
url: string;
|
||||
|
|
@ -9,22 +8,18 @@ export interface PlayerInit {
|
|||
}
|
||||
|
||||
export default class Player {
|
||||
audio: Audio;
|
||||
video: Video;
|
||||
media: Media;
|
||||
transport: Transport;
|
||||
|
||||
constructor(props: PlayerInit) {
|
||||
this.audio = new Audio()
|
||||
this.video = new Video({
|
||||
this.media = new Media({
|
||||
canvas: props.canvas.transferControlToOffscreen(),
|
||||
})
|
||||
|
||||
this.transport = new Transport({
|
||||
url: props.url,
|
||||
fingerprint: props.fingerprint,
|
||||
|
||||
audio: this.audio,
|
||||
video: this.video,
|
||||
media: this.media,
|
||||
})
|
||||
}
|
||||
|
||||
|
|
@ -33,13 +28,6 @@ export default class Player {
|
|||
}
|
||||
|
||||
play() {
|
||||
this.audio.play({})
|
||||
//this.video.play()
|
||||
//this.media.play()
|
||||
}
|
||||
|
||||
onMessage(msg: any) {
|
||||
if (msg.sync) {
|
||||
msg.sync
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -2,30 +2,22 @@ import * as Message from "./message"
|
|||
import * as Stream from "../stream"
|
||||
import * as MP4 from "../mp4"
|
||||
|
||||
import Audio from "../audio"
|
||||
import Video from "../video"
|
||||
import Media from "../media"
|
||||
|
||||
export interface TransportInit {
|
||||
url: string;
|
||||
fingerprint?: WebTransportHash; // the certificate fingerprint, temporarily needed for local development
|
||||
|
||||
audio: Audio;
|
||||
video: Video;
|
||||
media: Media;
|
||||
}
|
||||
|
||||
export default class Transport {
|
||||
quic: Promise<WebTransport>;
|
||||
api: Promise<WritableStream>;
|
||||
tracks: Map<string, MP4.InitParser>
|
||||
|
||||
audio: Audio;
|
||||
video: Video;
|
||||
media: Media;
|
||||
|
||||
constructor(props: TransportInit) {
|
||||
this.tracks = new Map();
|
||||
|
||||
this.audio = props.audio;
|
||||
this.video = props.video;
|
||||
this.media = props.media;
|
||||
|
||||
this.quic = this.connect(props)
|
||||
|
||||
|
|
@ -94,82 +86,18 @@ export default class Transport {
|
|||
const msg = JSON.parse(payload)
|
||||
|
||||
if (msg.init) {
|
||||
return this.handleInit(r, msg.init as Message.Init)
|
||||
return this.media.init({
|
||||
buffer: r.buffer,
|
||||
reader: r.reader,
|
||||
})
|
||||
} else if (msg.segment) {
|
||||
return this.handleSegment(r, msg.segment as Message.Segment)
|
||||
return this.media.segment({
|
||||
buffer: r.buffer,
|
||||
reader: r.reader,
|
||||
})
|
||||
} else {
|
||||
console.warn("unknown message", msg);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
async handleInit(stream: Stream.Reader, msg: Message.Init) {
|
||||
console.log("handle init", msg);
|
||||
|
||||
let track = this.tracks.get(msg.id);
|
||||
if (!track) {
|
||||
track = new MP4.InitParser()
|
||||
this.tracks.set(msg.id, track)
|
||||
}
|
||||
|
||||
while (1) {
|
||||
const data = await stream.read()
|
||||
if (!data) break
|
||||
|
||||
track.push(data)
|
||||
}
|
||||
|
||||
const info = await track.info
|
||||
|
||||
console.log(info);
|
||||
|
||||
if (info.audioTracks.length + info.videoTracks.length != 1) {
|
||||
throw new Error("expected a single track")
|
||||
}
|
||||
|
||||
if (info.audioTracks.length) {
|
||||
this.audio.init({
|
||||
track: msg.id,
|
||||
info: info,
|
||||
raw: track.raw,
|
||||
})
|
||||
} else if (info.videoTracks.length) {
|
||||
this.video.init({
|
||||
track: msg.id,
|
||||
info: info,
|
||||
raw: track.raw,
|
||||
})
|
||||
} else {
|
||||
throw new Error("init is neither audio nor video")
|
||||
}
|
||||
}
|
||||
|
||||
async handleSegment(stream: Stream.Reader, msg: Message.Segment) {
|
||||
console.log("handle segment", msg);
|
||||
|
||||
let track = this.tracks.get(msg.init);
|
||||
if (!track) {
|
||||
track = new MP4.InitParser()
|
||||
this.tracks.set(msg.init, track)
|
||||
}
|
||||
|
||||
// Wait until we learn if this is an audio or video track
|
||||
const info = await track.info
|
||||
|
||||
if (info.audioTracks.length) {
|
||||
this.audio.segment({
|
||||
track: msg.init,
|
||||
buffer: stream.buffer,
|
||||
reader: stream.reader,
|
||||
})
|
||||
} else if (info.videoTracks.length) {
|
||||
this.video.segment({
|
||||
track: msg.init,
|
||||
buffer: stream.buffer,
|
||||
reader: stream.reader,
|
||||
})
|
||||
} else {
|
||||
throw new Error("segment is neither audio nor video")
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -1,127 +0,0 @@
|
|||
import * as Message from "./message";
|
||||
import * as MP4 from "../mp4"
|
||||
import * as Stream from "../stream"
|
||||
import * as Util from "../util"
|
||||
|
||||
import Renderer from "./renderer"
|
||||
|
||||
export default class Decoder {
|
||||
// Store the init message for each track
|
||||
tracks: Map<string, Util.Deferred<Message.Init>>
|
||||
renderer: Renderer;
|
||||
|
||||
constructor(renderer: Renderer) {
|
||||
this.tracks = new Map();
|
||||
this.renderer = renderer;
|
||||
}
|
||||
|
||||
async init(msg: Message.Init) {
|
||||
let track = this.tracks.get(msg.track);
|
||||
if (!track) {
|
||||
track = new Util.Deferred()
|
||||
this.tracks.set(msg.track, track)
|
||||
}
|
||||
|
||||
if (msg.info.videoTracks.length != 1 || msg.info.audioTracks.length != 0) {
|
||||
throw new Error("Expected a single video track")
|
||||
}
|
||||
|
||||
track.resolve(msg)
|
||||
}
|
||||
|
||||
async decode(msg: Message.Segment) {
|
||||
let track = this.tracks.get(msg.track);
|
||||
if (!track) {
|
||||
track = new Util.Deferred()
|
||||
this.tracks.set(msg.track, track)
|
||||
}
|
||||
|
||||
// Wait for the init segment to be fully received and parsed
|
||||
const init = await track.promise;
|
||||
const info = init.info;
|
||||
const video = info.videoTracks[0]
|
||||
|
||||
const decoder = new VideoDecoder({
|
||||
output: (frame: VideoFrame) => {
|
||||
this.renderer.emit(frame)
|
||||
},
|
||||
error: (err: Error) => {
|
||||
console.warn(err)
|
||||
}
|
||||
});
|
||||
|
||||
const input = MP4.New();
|
||||
|
||||
input.onSamples = (id: number, user: any, samples: MP4.Sample[]) => {
|
||||
for (let sample of samples) {
|
||||
const timestamp = 1000 * sample.dts / sample.timescale // milliseconds
|
||||
|
||||
if (sample.is_sync) {
|
||||
// Configure the decoder using the AVC box for H.264
|
||||
const avcc = sample.description.avcC;
|
||||
const description = new MP4.Stream(new Uint8Array(avcc.size), 0, false)
|
||||
avcc.write(description)
|
||||
|
||||
decoder.configure({
|
||||
codec: video.codec,
|
||||
codedHeight: video.track_height,
|
||||
codedWidth: video.track_width,
|
||||
description: description.buffer?.slice(8),
|
||||
// optimizeForLatency: true
|
||||
})
|
||||
}
|
||||
|
||||
decoder.decode(new EncodedVideoChunk({
|
||||
data: sample.data,
|
||||
duration: sample.duration,
|
||||
timestamp: timestamp,
|
||||
type: sample.is_sync ? "key" : "delta",
|
||||
}))
|
||||
}
|
||||
}
|
||||
|
||||
input.onReady = (info: any) => {
|
||||
input.setExtractionOptions(info.tracks[0].id, {}, { nbSamples: 1 });
|
||||
input.start();
|
||||
}
|
||||
|
||||
// MP4box requires us to reparse the init segment unfortunately
|
||||
let offset = 0;
|
||||
|
||||
for (let raw of init.raw) {
|
||||
raw.fileStart = offset
|
||||
offset = input.appendBuffer(raw)
|
||||
}
|
||||
|
||||
const stream = new Stream.Reader(msg.reader, msg.buffer)
|
||||
|
||||
/* TODO I'm not actually sure why this code doesn't work; something trips up the MP4 parser
|
||||
while (1) {
|
||||
const data = await stream.read()
|
||||
if (!data) break
|
||||
|
||||
input.appendBuffer(data)
|
||||
input.flush()
|
||||
}
|
||||
*/
|
||||
|
||||
// One day I'll figure it out; until then read one top-level atom at a time
|
||||
while (!await stream.done()) {
|
||||
const raw = await stream.peek(4)
|
||||
const size = new DataView(raw.buffer, raw.byteOffset, raw.byteLength).getUint32(0)
|
||||
const atom = await stream.bytes(size)
|
||||
|
||||
// Make a copy of the atom because mp4box only accepts an ArrayBuffer unfortunately
|
||||
let box = new Uint8Array(atom.byteLength);
|
||||
box.set(atom)
|
||||
|
||||
// and for some reason we need to modify the underlying ArrayBuffer with offset
|
||||
let buffer = box.buffer as MP4.ArrayBuffer
|
||||
buffer.fileStart = offset
|
||||
|
||||
// Parse the data
|
||||
offset = input.appendBuffer(buffer)
|
||||
input.flush()
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -1,27 +0,0 @@
|
|||
import * as Message from "./message"
|
||||
|
||||
// Wrapper around the WebWorker API
|
||||
export default class Video {
|
||||
worker: Worker;
|
||||
|
||||
constructor(config: Message.Config) {
|
||||
const url = new URL('worker.ts', import.meta.url)
|
||||
this.worker = new Worker(url, {
|
||||
type: "module",
|
||||
name: "video",
|
||||
})
|
||||
this.worker.postMessage({ config }, [ config.canvas ])
|
||||
}
|
||||
|
||||
init(init: Message.Init) {
|
||||
this.worker.postMessage({ init }) // note: we copy the raw init bytes each time
|
||||
}
|
||||
|
||||
segment(segment: Message.Segment) {
|
||||
this.worker.postMessage({ segment }, [ segment.buffer.buffer, segment.reader ])
|
||||
}
|
||||
|
||||
play() {
|
||||
// TODO
|
||||
}
|
||||
}
|
||||
|
|
@ -1,17 +0,0 @@
|
|||
import * as MP4 from "../mp4"
|
||||
|
||||
export interface Config {
|
||||
canvas: OffscreenCanvas;
|
||||
}
|
||||
|
||||
export interface Init {
|
||||
track: string;
|
||||
info: MP4.Info;
|
||||
raw: MP4.ArrayBuffer[];
|
||||
}
|
||||
|
||||
export interface Segment {
|
||||
track: string;
|
||||
buffer: Uint8Array; // unread buffered data
|
||||
reader: ReadableStream; // unread unbuffered data
|
||||
}
|
||||
|
|
@ -1,91 +0,0 @@
|
|||
import * as Message from "./message";
|
||||
|
||||
export default class Renderer {
|
||||
canvas: OffscreenCanvas;
|
||||
queue: Array<VideoFrame>;
|
||||
render: number; // non-zero if requestAnimationFrame has been called
|
||||
sync?: DOMHighResTimeStamp; // the wall clock value for timestamp 0
|
||||
last?: number; // the timestamp of the last rendered frame
|
||||
|
||||
constructor(config: Message.Config) {
|
||||
this.canvas = config.canvas;
|
||||
this.queue = [];
|
||||
this.render = 0;
|
||||
}
|
||||
|
||||
emit(frame: VideoFrame) {
|
||||
if (!this.sync) {
|
||||
// Save the frame as the sync point
|
||||
this.sync = performance.now() - frame.timestamp
|
||||
}
|
||||
|
||||
// Drop any old frames
|
||||
if (this.last && frame.timestamp <= this.last) {
|
||||
frame.close()
|
||||
return
|
||||
}
|
||||
|
||||
// Insert the frame into the queue sorted by timestamp.
|
||||
if (this.queue.length > 0 && this.queue[this.queue.length-1].timestamp <= frame.timestamp) {
|
||||
// Fast path because we normally append to the end.
|
||||
this.queue.push(frame)
|
||||
} else {
|
||||
// Do a full binary search
|
||||
let low = 0
|
||||
let high = this.queue.length;
|
||||
|
||||
while (low < high) {
|
||||
var mid = (low + high) >>> 1;
|
||||
if (this.queue[mid].timestamp < frame.timestamp) low = mid + 1;
|
||||
else high = mid;
|
||||
}
|
||||
|
||||
this.queue.splice(low, 0, frame)
|
||||
}
|
||||
|
||||
// Queue up to render the next frame.
|
||||
if (!this.render) {
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
}
|
||||
}
|
||||
|
||||
draw(now: DOMHighResTimeStamp) {
|
||||
// Determine the target timestamp.
|
||||
const target = now - this.sync!
|
||||
|
||||
let frame = this.queue[0]
|
||||
if (frame.timestamp >= target) {
|
||||
// nothing to render yet, wait for the next animation frame
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
return
|
||||
}
|
||||
|
||||
this.queue.shift()
|
||||
|
||||
// Check if we should skip some frames
|
||||
while (this.queue.length) {
|
||||
const next = this.queue[0]
|
||||
if (next.timestamp > target) {
|
||||
break
|
||||
}
|
||||
|
||||
frame.close()
|
||||
|
||||
this.queue.shift()
|
||||
frame = next
|
||||
}
|
||||
|
||||
const ctx = this.canvas.getContext("2d");
|
||||
ctx!.drawImage(frame, 0, 0, this.canvas.width, this.canvas.height) // TODO aspect ratio
|
||||
|
||||
this.last = frame.timestamp;
|
||||
frame.close()
|
||||
|
||||
if (this.queue.length > 0) {
|
||||
this.render = self.requestAnimationFrame(this.draw.bind(this))
|
||||
} else {
|
||||
// Break the loop for now
|
||||
this.render = 0
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -329,7 +329,6 @@ dependencies = [
|
|||
[[package]]
|
||||
name = "mp4"
|
||||
version = "0.13.0"
|
||||
source = "git+https://github.com/kixelated/mp4-rust.git?branch=trexs#efefcc47353f477518bff01493785ae0daa8efd4"
|
||||
dependencies = [
|
||||
"byteorder",
|
||||
"bytes",
|
||||
|
|
|
|||
|
|
@ -13,6 +13,6 @@ mio = { version = "0.8", features = ["net", "os-poll"] }
|
|||
env_logger = "0.9.3"
|
||||
ring = "0.16"
|
||||
anyhow = "1.0.70"
|
||||
mp4 = { git = "https://github.com/kixelated/mp4-rust.git", branch = "trexs" }
|
||||
mp4 = { path = "../../mp4-rust" } # { git = "https://github.com/kixelated/mp4-rust.git", branch = "trexs" }
|
||||
serde = "1.0.160"
|
||||
serde_json = "1.0"
|
||||
|
|
@ -1,3 +1,3 @@
|
|||
mod source;
|
||||
|
||||
pub use source::{Fragment, Source};
|
||||
pub use source::{Fragment, Source};
|
||||
|
|
@ -1,37 +1,32 @@
|
|||
use io::Read;
|
||||
use std::collections::VecDeque;
|
||||
use std::{fs, io, time};
|
||||
|
||||
use std::io::Write;
|
||||
use std::collections::{HashMap,VecDeque};
|
||||
use std::io::Read;
|
||||
|
||||
use anyhow;
|
||||
use mp4;
|
||||
|
||||
use mp4::{ReadBox, WriteBox};
|
||||
use mp4;
|
||||
use mp4::ReadBox;
|
||||
|
||||
pub struct Source {
|
||||
// We read the file once, in order, and don't seek backwards.
|
||||
reader: io::BufReader<fs::File>,
|
||||
|
||||
// Any fragments parsed and ready to be returned by next().
|
||||
fragments: VecDeque<Fragment>,
|
||||
|
||||
// The timestamp when the broadcast "started", so we can sleep to simulate a live stream.
|
||||
start: time::Instant,
|
||||
|
||||
// The raw ftyp box, which we need duplicate for each track, but we don't know how many tracks exist yet.
|
||||
ftyp: Vec<u8>,
|
||||
// The initialization payload; ftyp + moov boxes.
|
||||
pub init: Vec<u8>,
|
||||
|
||||
// The parsed moov box, so we can look up track information later.
|
||||
moov: Option<mp4::MoovBox>,
|
||||
// The timescale used for each track.
|
||||
timescale: HashMap<u32, u32>,
|
||||
|
||||
// Any fragments parsed and ready to be returned by next().
|
||||
fragments: VecDeque<Fragment>,
|
||||
}
|
||||
|
||||
pub struct Fragment {
|
||||
// The track ID for the fragment.
|
||||
pub track: u32,
|
||||
|
||||
// The type of the fragment.
|
||||
pub typ: mp4::BoxType,
|
||||
pub track_id: u32,
|
||||
|
||||
// The data of the fragment.
|
||||
pub data: Vec<u8>,
|
||||
|
|
@ -44,21 +39,42 @@ pub struct Fragment {
|
|||
}
|
||||
|
||||
impl Source {
|
||||
pub fn new(path: &str) -> io::Result<Self> {
|
||||
pub fn new(path: &str) -> anyhow::Result<Self> {
|
||||
let f = fs::File::open(path)?;
|
||||
let reader = io::BufReader::new(f);
|
||||
let mut reader = io::BufReader::new(f);
|
||||
let start = time::Instant::now();
|
||||
|
||||
Ok(Self {
|
||||
let ftyp = read_atom(&mut reader)?;
|
||||
anyhow::ensure!(&ftyp[4..8] == b"ftyp", "expected ftyp atom");
|
||||
|
||||
let moov = read_atom(&mut reader)?;
|
||||
anyhow::ensure!(&moov[4..8] == b"moov", "expected moov atom");
|
||||
|
||||
let mut init = ftyp;
|
||||
init.extend(&moov);
|
||||
|
||||
// We're going to parse the moov box.
|
||||
// We have to read the moov box header to correctly advance the cursor for the mp4 crate.
|
||||
let mut moov_reader = io::Cursor::new(&moov);
|
||||
let moov_header = mp4::BoxHeader::read(&mut moov_reader)?;
|
||||
|
||||
// Parse the moov box so we can detect the timescales for each track.
|
||||
let moov = mp4::MoovBox::read_box(&mut moov_reader, moov_header.size)?;
|
||||
let timescale = moov.traks
|
||||
.iter()
|
||||
.map(|trak| (trak.tkhd.track_id, trak.mdia.mdhd.timescale))
|
||||
.collect();
|
||||
|
||||
Ok(Self{
|
||||
reader,
|
||||
start,
|
||||
init,
|
||||
timescale,
|
||||
fragments: VecDeque::new(),
|
||||
ftyp: Vec::new(),
|
||||
moov: None,
|
||||
})
|
||||
}
|
||||
|
||||
pub fn get(&mut self) -> anyhow::Result<Option<Fragment>> {
|
||||
pub fn fragment(&mut self) -> anyhow::Result<Option<Fragment>> {
|
||||
if self.fragments.is_empty() {
|
||||
self.parse()?;
|
||||
};
|
||||
|
|
@ -72,63 +88,13 @@ impl Source {
|
|||
|
||||
fn parse(&mut self) -> anyhow::Result<()> {
|
||||
loop {
|
||||
// Read the next full atom.
|
||||
let atom = read_box(&mut self.reader)?;
|
||||
let atom = read_atom(&mut self.reader)?;
|
||||
|
||||
// Before we return it, let's do some simple parsing.
|
||||
let mut reader = io::Cursor::new(&atom);
|
||||
let header = mp4::BoxHeader::read(&mut reader)?;
|
||||
|
||||
match header.name {
|
||||
mp4::BoxType::FtypBox => {
|
||||
// Don't return anything until we know the total number of tracks.
|
||||
// To be honest, I didn't expect the borrow checker to allow this, but it does!
|
||||
self.ftyp = atom;
|
||||
}
|
||||
mp4::BoxType::MoovBox => {
|
||||
// We need to split the moov based on the tracks.
|
||||
let moov = mp4::MoovBox::read_box(&mut reader, header.size)?;
|
||||
|
||||
for trak in &moov.traks {
|
||||
let track_id = trak.tkhd.track_id;
|
||||
|
||||
// Push the styp atom for each track.
|
||||
self.fragments.push_back(Fragment {
|
||||
track: track_id,
|
||||
typ: mp4::BoxType::FtypBox,
|
||||
data: self.ftyp.clone(),
|
||||
keyframe: false,
|
||||
timestamp: None,
|
||||
});
|
||||
|
||||
// Unfortunately, we need to create a brand new moov atom for each track.
|
||||
// We remove every box for other track IDs.
|
||||
let mut toov = moov.clone();
|
||||
toov.traks.retain(|t| t.tkhd.track_id == track_id);
|
||||
toov.mvex
|
||||
.as_mut()
|
||||
.expect("missing mvex")
|
||||
.trexs
|
||||
.retain(|f| f.track_id == track_id);
|
||||
|
||||
// Marshal the box.
|
||||
let mut toov_data = Vec::new();
|
||||
toov.write_box(&mut toov_data)?;
|
||||
|
||||
let mut file = std::fs::File::create(format!("track{}.mp4", track_id))?;
|
||||
file.write_all(toov_data.as_slice())?;
|
||||
|
||||
self.fragments.push_back(Fragment {
|
||||
track: track_id,
|
||||
typ: mp4::BoxType::MoovBox,
|
||||
data: toov_data,
|
||||
keyframe: false,
|
||||
timestamp: None,
|
||||
});
|
||||
}
|
||||
|
||||
self.moov = Some(moov);
|
||||
}
|
||||
mp4::BoxType::FtypBox | mp4::BoxType::MoovBox => anyhow::bail!("must call init first"),
|
||||
mp4::BoxType::MoofBox => {
|
||||
let moof = mp4::MoofBox::read_box(&mut reader, header.size)?;
|
||||
|
||||
|
|
@ -138,8 +104,7 @@ impl Source {
|
|||
}
|
||||
|
||||
self.fragments.push_back(Fragment {
|
||||
track: moof.trafs[0].tfhd.track_id,
|
||||
typ: mp4::BoxType::MoofBox,
|
||||
track_id: moof.trafs[0].tfhd.track_id,
|
||||
data: atom,
|
||||
keyframe: has_keyframe(&moof),
|
||||
timestamp: first_timestamp(&moof),
|
||||
|
|
@ -147,11 +112,9 @@ impl Source {
|
|||
}
|
||||
mp4::BoxType::MdatBox => {
|
||||
let moof = self.fragments.back().expect("no atom before mdat");
|
||||
assert!(moof.typ == mp4::BoxType::MoofBox, "no moof before mdat");
|
||||
|
||||
self.fragments.push_back(Fragment {
|
||||
track: moof.track,
|
||||
typ: mp4::BoxType::MoofBox,
|
||||
track_id: moof.track_id,
|
||||
data: atom,
|
||||
keyframe: false,
|
||||
timestamp: None,
|
||||
|
|
@ -160,7 +123,9 @@ impl Source {
|
|||
// We have some media data, return so we can start sending it.
|
||||
return Ok(());
|
||||
}
|
||||
_ => anyhow::bail!("unknown top-level atom: {:?}", header.name),
|
||||
_ => {
|
||||
// Skip unknown atoms
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -171,15 +136,9 @@ impl Source {
|
|||
let timestamp = next.timestamp?;
|
||||
|
||||
// Find the timescale for the track.
|
||||
let track = self
|
||||
.moov
|
||||
.as_ref()?
|
||||
.traks
|
||||
.iter()
|
||||
.find(|t| t.tkhd.track_id == next.track)?;
|
||||
let timescale = track.mdia.mdhd.timescale as u64;
|
||||
let timescale = self.timescale.get(&next.track_id).unwrap();
|
||||
|
||||
let delay = time::Duration::from_millis(1000 * timestamp / timescale);
|
||||
let delay = time::Duration::from_millis(1000 * timestamp / *timescale as u64);
|
||||
let elapsed = self.start.elapsed();
|
||||
|
||||
delay.checked_sub(elapsed)
|
||||
|
|
@ -187,14 +146,16 @@ impl Source {
|
|||
}
|
||||
|
||||
// Read a full MP4 atom into a vector.
|
||||
fn read_box<R: io::Read>(reader: &mut R) -> anyhow::Result<Vec<u8>> {
|
||||
pub fn read_atom<R: Read>(reader: &mut R) -> anyhow::Result<Vec<u8>> {
|
||||
// Read the 8 bytes for the size + type
|
||||
let mut buf = [0u8; 8];
|
||||
reader.read_exact(&mut buf)?;
|
||||
|
||||
// Convert the first 4 bytes into the size.
|
||||
let size = u32::from_be_bytes(buf[0..4].try_into()?) as u64;
|
||||
let mut out = buf.to_vec();
|
||||
//let typ = &buf[4..8].try_into().ok().unwrap();
|
||||
|
||||
let mut raw = buf.to_vec();
|
||||
|
||||
let mut limit = match size {
|
||||
// Runs until the end of the file.
|
||||
|
|
@ -222,9 +183,9 @@ fn read_box<R: io::Read>(reader: &mut R) -> anyhow::Result<Vec<u8>> {
|
|||
};
|
||||
|
||||
// Append to the vector and return it.
|
||||
limit.read_to_end(&mut out)?;
|
||||
limit.read_to_end(&mut raw)?;
|
||||
|
||||
Ok(out)
|
||||
Ok(raw)
|
||||
}
|
||||
|
||||
fn has_keyframe(moof: &mp4::MoofBox) -> bool {
|
||||
|
|
@ -261,4 +222,4 @@ fn has_keyframe(moof: &mp4::MoofBox) -> bool {
|
|||
|
||||
fn first_timestamp(moof: &mp4::MoofBox) -> Option<u64> {
|
||||
Some(moof.trafs.first()?.tfdt.as_ref()?.base_media_decode_time)
|
||||
}
|
||||
}
|
||||
|
|
@ -7,13 +7,11 @@ pub struct Message {
|
|||
}
|
||||
|
||||
#[derive(Serialize, Deserialize)]
|
||||
pub struct Init {
|
||||
pub id: String,
|
||||
}
|
||||
pub struct Init {}
|
||||
|
||||
#[derive(Serialize, Deserialize)]
|
||||
pub struct Segment {
|
||||
pub init: String,
|
||||
pub track_id: u32,
|
||||
}
|
||||
|
||||
impl Message {
|
||||
|
|
|
|||
|
|
@ -1,6 +1,7 @@
|
|||
mod message;
|
||||
|
||||
use std::time;
|
||||
use std::collections::hash_map as hmap;
|
||||
|
||||
use quiche;
|
||||
use quiche::h3::webtransport;
|
||||
|
|
@ -10,9 +11,8 @@ use crate::{media, transport};
|
|||
#[derive(Default)]
|
||||
pub struct Session {
|
||||
media: Option<media::Source>,
|
||||
stream_id: Option<u64>, // stream ID of the current segment
|
||||
|
||||
streams: transport::Streams, // An easy way of buffering stream data.
|
||||
tracks: hmap::HashMap<u32, u64>, // map from track_id to current stream_id
|
||||
}
|
||||
|
||||
impl transport::App for Session {
|
||||
|
|
@ -38,12 +38,23 @@ impl transport::App for Session {
|
|||
// req.authority()
|
||||
// req.path()
|
||||
// and you can validate this request with req.origin()
|
||||
session.accept_connect_request(conn, None).unwrap();
|
||||
|
||||
// TODO
|
||||
let media = media::Source::new("../media/fragmented.mp4")?;
|
||||
self.media = Some(media);
|
||||
let init = &media.init;
|
||||
|
||||
session.accept_connect_request(conn, None).unwrap();
|
||||
// Create a JSON header.
|
||||
let mut message = message::Message::new();
|
||||
message.init = Some(message::Init{});
|
||||
let data = message.serialize()?;
|
||||
|
||||
// Create a new stream and write the header.
|
||||
let stream_id = session.open_stream(conn, false)?;
|
||||
self.streams.send(conn, stream_id, data.as_slice(), false)?;
|
||||
self.streams.send(conn, stream_id, init.as_slice(), true)?;
|
||||
|
||||
self.media = Some(media);
|
||||
}
|
||||
webtransport::ServerEvent::StreamData(stream_id) => {
|
||||
let mut buf = vec![0; 10000];
|
||||
|
|
@ -84,62 +95,54 @@ impl Session {
|
|||
};
|
||||
|
||||
// Get the next media fragment.
|
||||
let fragment = match media.get()? {
|
||||
let fragment = match media.fragment()? {
|
||||
Some(f) => f,
|
||||
None => return Ok(()),
|
||||
};
|
||||
|
||||
// Check if we have already created a stream for this fragment.
|
||||
let stream_id = match self.stream_id {
|
||||
Some(old_stream_id) if fragment.keyframe => {
|
||||
// This is the start of a new segment.
|
||||
let stream_id = match self.tracks.get(&fragment.track_id) {
|
||||
// Close the old stream.
|
||||
Some(stream_id) if fragment.keyframe => {
|
||||
self.streams.send(conn, *stream_id, &[], true)?;
|
||||
None
|
||||
},
|
||||
|
||||
// Close the prior stream.
|
||||
self.streams.send(conn, old_stream_id, &[], true)?;
|
||||
// Use the existing stream
|
||||
Some(stream_id) => Some(*stream_id),
|
||||
|
||||
// Encode a JSON header indicating this is the video track.
|
||||
let mut message = message::Message::new();
|
||||
message.segment = Some(message::Segment {
|
||||
init: "video".to_string(),
|
||||
});
|
||||
// No existing stream.
|
||||
_ => None,
|
||||
};
|
||||
|
||||
// Open a new stream.
|
||||
let stream_id = match stream_id {
|
||||
// Use the existing stream,
|
||||
Some(stream_id) => stream_id,
|
||||
|
||||
// Open a new stream.
|
||||
None => {
|
||||
let stream_id = session.open_stream(conn, false)?;
|
||||
// TODO: conn.stream_priority(stream_id, urgency, incremental)
|
||||
|
||||
// Encode a JSON header indicating this is a new track.
|
||||
let mut message = message::Message::new();
|
||||
message.segment = Some(message::Segment {
|
||||
track_id: fragment.track_id,
|
||||
});
|
||||
|
||||
// Write the header.
|
||||
let data = message.serialize()?;
|
||||
self.streams.send(conn, stream_id, &data, false)?;
|
||||
|
||||
stream_id
|
||||
}
|
||||
None => {
|
||||
// This is the start of an init segment.
|
||||
|
||||
// Create a JSON header.
|
||||
let mut message = message::Message::new();
|
||||
message.init = Some(message::Init {
|
||||
id: "video".to_string(),
|
||||
});
|
||||
|
||||
let data = message.serialize()?;
|
||||
|
||||
// Create a new stream and write the header.
|
||||
let stream_id = session.open_stream(conn, false)?;
|
||||
self.streams.send(conn, stream_id, data.as_slice(), false)?;
|
||||
self.tracks.insert(fragment.track_id, stream_id);
|
||||
|
||||
stream_id
|
||||
}
|
||||
Some(stream_id) => stream_id, // Continuation of init or segment
|
||||
},
|
||||
};
|
||||
|
||||
// Write the current fragment.
|
||||
let data = fragment.data.as_slice();
|
||||
self.streams.send(conn, stream_id, data, false)?;
|
||||
|
||||
// Save the stream ID for the next fragment.
|
||||
self.stream_id = Some(stream_id);
|
||||
|
||||
Ok(())
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1 +0,0 @@
|
|||
|
||||
Loading…
Reference in New Issue