Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 | 1x | /**
* fMP4 remuxer
*/
import AAC from '../helper/aac';
import Event from '../events';
import { logger } from '../utils/logger';
import MP4 from '../remux/mp4-generator';
import { ErrorTypes, ErrorDetails } from '../errors';
// 10 seconds
const MAX_SILENT_FRAME_DURATION = 10 * 1000;
class MP4Remuxer {
constructor (observer, config, typeSupported, vendor) {
this.observer = observer;
this.config = config;
this.typeSupported = typeSupported;
const userAgent = navigator.userAgent;
this.isSafari = vendor && vendor.indexOf('Apple') > -1 && userAgent && !userAgent.match('CriOS');
this.ISGenerated = false;
}
destroy () {
}
resetTimeStamp (defaultTimeStamp) {
this._initPTS = this._initDTS = defaultTimeStamp;
}
resetInitSegment () {
this.ISGenerated = false;
}
remux (audioTrack, videoTrack, id3Track, textTrack, timeOffset, contiguous, accurateTimeOffset) {
// generate Init Segment if needed
if (!this.ISGenerated)
this.generateIS(audioTrack, videoTrack, timeOffset);
if (this.ISGenerated) {
const nbAudioSamples = audioTrack.samples.length;
const nbVideoSamples = videoTrack.samples.length;
let audioTimeOffset = timeOffset;
let videoTimeOffset = timeOffset;
if (nbAudioSamples && nbVideoSamples) {
// timeOffset is expected to be the offset of the first timestamp of this fragment (first DTS)
// if first audio DTS is not aligned with first video DTS then we need to take that into account
// when providing timeOffset to remuxAudio / remuxVideo. if we don't do that, there might be a permanent / small
// drift between audio and video streams
let audiovideoDeltaDts = (audioTrack.samples[0].dts - videoTrack.samples[0].dts) / videoTrack.inputTimeScale;
audioTimeOffset += Math.max(0, audiovideoDeltaDts);
videoTimeOffset += Math.max(0, -audiovideoDeltaDts);
}
// Purposefully remuxing audio before video, so that remuxVideo can use nextAudioPts, which is
// calculated in remuxAudio.
// logger.log('nb AAC samples:' + audioTrack.samples.length);
if (nbAudioSamples) {
// if initSegment was generated without video samples, regenerate it again
if (!audioTrack.timescale) {
logger.warn('regenerate InitSegment as audio detected');
this.generateIS(audioTrack, videoTrack, timeOffset);
}
let audioData = this.remuxAudio(audioTrack, audioTimeOffset, contiguous, accurateTimeOffset);
// logger.log('nb AVC samples:' + videoTrack.samples.length);
if (nbVideoSamples) {
let audioTrackLength;
if (audioData)
audioTrackLength = audioData.endPTS - audioData.startPTS;
// if initSegment was generated without video samples, regenerate it again
if (!videoTrack.timescale) {
logger.warn('regenerate InitSegment as video detected');
this.generateIS(audioTrack, videoTrack, timeOffset);
}
this.remuxVideo(videoTrack, videoTimeOffset, contiguous, audioTrackLength, accurateTimeOffset);
}
} else {
// logger.log('nb AVC samples:' + videoTrack.samples.length);
if (nbVideoSamples) {
let videoData = this.remuxVideo(videoTrack, videoTimeOffset, contiguous, 0, accurateTimeOffset);
if (videoData && audioTrack.codec)
this.remuxEmptyAudio(audioTrack, audioTimeOffset, contiguous, videoData);
}
}
}
// logger.log('nb ID3 samples:' + audioTrack.samples.length);
if (id3Track.samples.length)
this.remuxID3(id3Track, timeOffset);
// logger.log('nb ID3 samples:' + audioTrack.samples.length);
if (textTrack.samples.length)
this.remuxText(textTrack, timeOffset);
// notify end of parsing
this.observer.trigger(Event.FRAG_PARSED);
}
generateIS (audioTrack, videoTrack, timeOffset) {
let observer = this.observer,
audioSamples = audioTrack.samples,
videoSamples = videoTrack.samples,
typeSupported = this.typeSupported,
container = 'audio/mp4',
tracks = {},
data = { tracks: tracks },
computePTSDTS = (this._initPTS === undefined),
initPTS, initDTS;
if (computePTSDTS)
initPTS = initDTS = Infinity;
if (audioTrack.config && audioSamples.length) {
// let's use audio sampling rate as MP4 time scale.
// rationale is that there is a integer nb of audio frames per audio sample (1024 for AAC)
// using audio sampling rate here helps having an integer MP4 frame duration
// this avoids potential rounding issue and AV sync issue
audioTrack.timescale = audioTrack.samplerate;
logger.log(`audio sampling rate : ${audioTrack.samplerate}`);
if (!audioTrack.isAAC) {
if (typeSupported.mpeg) { // Chrome and Safari
container = 'audio/mpeg';
audioTrack.codec = '';
} else if (typeSupported.mp3) { // Firefox
audioTrack.codec = 'mp3';
}
}
tracks.audio = {
container: container,
codec: audioTrack.codec,
initSegment: !audioTrack.isAAC && typeSupported.mpeg ? new Uint8Array() : MP4.initSegment([audioTrack]),
metadata: {
channelCount: audioTrack.channelCount
}
};
if (computePTSDTS) {
// remember first PTS of this demuxing context. for audio, PTS = DTS
initPTS = initDTS = audioSamples[0].pts - audioTrack.inputTimeScale * timeOffset;
}
}
if (videoTrack.sps && videoTrack.pps && videoSamples.length) {
// let's use input time scale as MP4 video timescale
// we use input time scale straight away to avoid rounding issues on frame duration / cts computation
const inputTimeScale = videoTrack.inputTimeScale;
videoTrack.timescale = inputTimeScale;
tracks.video = {
container: 'video/mp4',
codec: videoTrack.codec,
initSegment: MP4.initSegment([videoTrack]),
metadata: {
width: videoTrack.width,
height: videoTrack.height
}
};
if (computePTSDTS) {
initPTS = Math.min(initPTS, videoSamples[0].pts - inputTimeScale * timeOffset);
initDTS = Math.min(initDTS, videoSamples[0].dts - inputTimeScale * timeOffset);
this.observer.trigger(Event.INIT_PTS_FOUND, { initPTS: initPTS });
}
}
if (Object.keys(tracks).length) {
observer.trigger(Event.FRAG_PARSING_INIT_SEGMENT, data);
this.ISGenerated = true;
if (computePTSDTS) {
this._initPTS = initPTS;
this._initDTS = initDTS;
}
} else {
observer.trigger(Event.ERROR, { type: ErrorTypes.MEDIA_ERROR, details: ErrorDetails.FRAG_PARSING_ERROR, fatal: false, reason: 'no audio/video samples found' });
}
}
remuxVideo (track, timeOffset, contiguous, audioTrackLength, accurateTimeOffset) {
let offset = 8,
timeScale = track.timescale,
mp4SampleDuration,
mdat, moof,
firstPTS, firstDTS,
nextDTS,
lastPTS, lastDTS,
inputSamples = track.samples,
outputSamples = [],
nbSamples = inputSamples.length,
ptsNormalize = this._PTSNormalize,
initDTS = this._initDTS;
// for (let i = 0; i < track.samples.length; i++) {
// let avcSample = track.samples[i];
// let units = avcSample.units;
// let unitsString = '';
// for (let j = 0; j < units.length ; j++) {
// unitsString += units[j].type + ',';
// if (units[j].data.length < 500) {
// unitsString += Hex.hexDump(units[j].data);
// }
// }
// logger.log(avcSample.pts + '/' + avcSample.dts + ',' + unitsString + avcSample.units.length);
// }
// if parsed fragment is contiguous with last one, let's use last DTS value as reference
let nextAvcDts = this.nextAvcDts;
const isSafari = this.isSafari;
if (nbSamples === 0)
return;
// Safari does not like overlapping DTS on consecutive fragments. let's use nextAvcDts to overcome this if fragments are consecutive
if (isSafari) {
// also consider consecutive fragments as being contiguous (even if a level switch occurs),
// for sake of clarity:
// consecutive fragments are frags with
// - less than 100ms gaps between new time offset (if accurate) and next expected PTS OR
// - less than 200 ms PTS gaps (timeScale/5)
contiguous |= (inputSamples.length && nextAvcDts &&
((accurateTimeOffset && Math.abs(timeOffset - nextAvcDts / timeScale) < 0.1) ||
Math.abs((inputSamples[0].pts - nextAvcDts - initDTS)) < timeScale / 5)
);
}
if (!contiguous) {
// if not contiguous, let's use target timeOffset
nextAvcDts = timeOffset * timeScale;
}
// PTS is coded on 33bits, and can loop from -2^32 to 2^32
// ptsNormalize will make PTS/DTS value monotonic, we use last known DTS value as reference value
inputSamples.forEach(function (sample) {
sample.pts = ptsNormalize(sample.pts - initDTS, nextAvcDts);
sample.dts = ptsNormalize(sample.dts - initDTS, nextAvcDts);
});
// sort video samples by DTS then PTS then demux id order
inputSamples.sort(function (a, b) {
const deltadts = a.dts - b.dts;
const deltapts = a.pts - b.pts;
return deltadts || (deltapts || (a.id - b.id));
});
// handle broken streams with PTS < DTS, tolerance up 200ms (18000 in 90kHz timescale)
let PTSDTSshift = inputSamples.reduce((prev, curr) => Math.max(Math.min(prev, curr.pts - curr.dts), -18000), 0);
if (PTSDTSshift < 0) {
logger.warn(`PTS < DTS detected in video samples, shifting DTS by ${Math.round(PTSDTSshift / 90)} ms to overcome this issue`);
for (let i = 0; i < inputSamples.length; i++)
inputSamples[i].dts += PTSDTSshift;
}
// compute first DTS and last DTS, normalize them against reference value
let sample = inputSamples[0];
firstDTS = Math.max(sample.dts, 0);
firstPTS = Math.max(sample.pts, 0);
// check timestamp continuity accross consecutive fragments (this is to remove inter-fragment gap/hole)
let delta = Math.round((firstDTS - nextAvcDts) / 90);
// if fragment are contiguous, detect hole/overlapping between fragments
if (contiguous) {
if (delta) {
if (delta > 1)
logger.log(`AVC:${delta} ms hole between fragments detected,filling it`);
else if (delta < -1)
logger.log(`AVC:${(-delta)} ms overlapping between fragments detected`);
// remove hole/gap : set DTS to next expected DTS
firstDTS = nextAvcDts;
inputSamples[0].dts = firstDTS;
// offset PTS as well, ensure that PTS is smaller or equal than new DTS
firstPTS = Math.max(firstPTS - delta, nextAvcDts);
inputSamples[0].pts = firstPTS;
logger.log(`Video/PTS/DTS adjusted: ${Math.round(firstPTS / 90)}/${Math.round(firstDTS / 90)},delta:${delta} ms`);
}
}
nextDTS = firstDTS;
// compute lastPTS/lastDTS
sample = inputSamples[inputSamples.length - 1];
lastDTS = Math.max(sample.dts, 0);
lastPTS = Math.max(sample.pts, 0, lastDTS);
// on Safari let's signal the same sample duration for all samples
// sample duration (as expected by trun MP4 boxes), should be the delta between sample DTS
// set this constant duration as being the avg delta between consecutive DTS.
if (isSafari)
mp4SampleDuration = Math.round((lastDTS - firstDTS) / (inputSamples.length - 1));
let nbNalu = 0, naluLen = 0;
for (let i = 0; i < nbSamples; i++) {
// compute total/avc sample length and nb of NAL units
let sample = inputSamples[i], units = sample.units, nbUnits = units.length, sampleLen = 0;
for (let j = 0; j < nbUnits; j++)
sampleLen += units[j].data.length;
naluLen += sampleLen;
nbNalu += nbUnits;
sample.length = sampleLen;
// normalize PTS/DTS
if (isSafari) {
// sample DTS is computed using a constant decoding offset (mp4SampleDuration) between samples
sample.dts = firstDTS + i * mp4SampleDuration;
} else {
// ensure sample monotonic DTS
sample.dts = Math.max(sample.dts, firstDTS);
}
// ensure that computed value is greater or equal than sample DTS
sample.pts = Math.max(sample.pts, sample.dts);
}
/* concatenate the video data and construct the mdat in place
(need 8 more bytes to fill length and mpdat type) */
let mdatSize = naluLen + (4 * nbNalu) + 8;
try {
mdat = new Uint8Array(mdatSize);
} catch (err) {
this.observer.trigger(Event.ERROR, { type: ErrorTypes.MUX_ERROR, details: ErrorDetails.REMUX_ALLOC_ERROR, fatal: false, bytes: mdatSize, reason: `fail allocating video mdat ${mdatSize}` });
return;
}
let view = new DataView(mdat.buffer);
view.setUint32(0, mdatSize);
mdat.set(MP4.types.mdat, 4);
for (let i = 0; i < nbSamples; i++) {
let avcSample = inputSamples[i],
avcSampleUnits = avcSample.units,
mp4SampleLength = 0,
compositionTimeOffset;
// convert NALU bitstream to MP4 format (prepend NALU with size field)
for (let j = 0, nbUnits = avcSampleUnits.length; j < nbUnits; j++) {
let unit = avcSampleUnits[j],
unitData = unit.data,
unitDataLen = unit.data.byteLength;
view.setUint32(offset, unitDataLen);
offset += 4;
mdat.set(unitData, offset);
offset += unitDataLen;
mp4SampleLength += 4 + unitDataLen;
}
if (!isSafari) {
// expected sample duration is the Decoding Timestamp diff of consecutive samples
if (i < nbSamples - 1) {
mp4SampleDuration = inputSamples[i + 1].dts - avcSample.dts;
} else {
let config = this.config,
lastFrameDuration = avcSample.dts - inputSamples[i > 0 ? i - 1 : i].dts;
if (config.stretchShortVideoTrack) {
// In some cases, a segment's audio track duration may exceed the video track duration.
// Since we've already remuxed audio, and we know how long the audio track is, we look to
// see if the delta to the next segment is longer than maxBufferHole.
// If so, playback would potentially get stuck, so we artificially inflate
// the duration of the last frame to minimize any potential gap between segments.
let maxBufferHole = config.maxBufferHole,
gapTolerance = Math.floor(maxBufferHole * timeScale),
deltaToFrameEnd = (audioTrackLength ? firstPTS + audioTrackLength * timeScale : this.nextAudioPts) - avcSample.pts;
if (deltaToFrameEnd > gapTolerance) {
// We subtract lastFrameDuration from deltaToFrameEnd to try to prevent any video
// frame overlap. maxBufferHole should be >> lastFrameDuration anyway.
mp4SampleDuration = deltaToFrameEnd - lastFrameDuration;
if (mp4SampleDuration < 0)
mp4SampleDuration = lastFrameDuration;
logger.log(`It is approximately ${deltaToFrameEnd / 90} ms to the next segment; using duration ${mp4SampleDuration / 90} ms for the last video frame.`);
} else {
mp4SampleDuration = lastFrameDuration;
}
} else {
mp4SampleDuration = lastFrameDuration;
}
}
compositionTimeOffset = Math.round(avcSample.pts - avcSample.dts);
} else {
compositionTimeOffset = Math.max(0, mp4SampleDuration * Math.round((avcSample.pts - avcSample.dts) / mp4SampleDuration));
}
// console.log('PTS/DTS/initDTS/normPTS/normDTS/relative PTS : ${avcSample.pts}/${avcSample.dts}/${initDTS}/${ptsnorm}/${dtsnorm}/${(avcSample.pts/4294967296).toFixed(3)}');
outputSamples.push({
size: mp4SampleLength,
// constant duration
duration: mp4SampleDuration,
cts: compositionTimeOffset,
flags: {
isLeading: 0,
isDependedOn: 0,
hasRedundancy: 0,
degradPrio: 0,
dependsOn: avcSample.key ? 2 : 1,
isNonSync: avcSample.key ? 0 : 1
}
});
}
// next AVC sample DTS should be equal to last sample DTS + last sample duration (in PES timescale)
this.nextAvcDts = lastDTS + mp4SampleDuration;
let dropped = track.dropped;
track.len = 0;
track.nbNalu = 0;
track.dropped = 0;
if (outputSamples.length && navigator.userAgent.toLowerCase().indexOf('chrome') > -1) {
let flags = outputSamples[0].flags;
// chrome workaround, mark first sample as being a Random Access Point to avoid sourcebuffer append issue
// https://code.google.com/p/chromium/issues/detail?id=229412
flags.dependsOn = 2;
flags.isNonSync = 0;
}
track.samples = outputSamples;
moof = MP4.moof(track.sequenceNumber++, firstDTS, track);
track.samples = [];
let data = {
data1: moof,
data2: mdat,
startPTS: firstPTS / timeScale,
endPTS: (lastPTS + mp4SampleDuration) / timeScale,
startDTS: firstDTS / timeScale,
endDTS: this.nextAvcDts / timeScale,
type: 'video',
hasAudio: false,
hasVideo: true,
nb: outputSamples.length,
dropped: dropped
};
this.observer.trigger(Event.FRAG_PARSING_DATA, data);
return data;
}
remuxAudio (track, timeOffset, contiguous, accurateTimeOffset) {
const inputTimeScale = track.inputTimeScale,
mp4timeScale = track.timescale,
scaleFactor = inputTimeScale / mp4timeScale,
mp4SampleDuration = track.isAAC ? 1024 : 1152,
inputSampleDuration = mp4SampleDuration * scaleFactor,
ptsNormalize = this._PTSNormalize,
initDTS = this._initDTS,
rawMPEG = !track.isAAC && this.typeSupported.mpeg;
let offset,
mp4Sample,
fillFrame,
mdat, moof,
firstPTS, lastPTS,
inputSamples = track.samples,
outputSamples = [],
nextAudioPts = this.nextAudioPts;
// for audio samples, also consider consecutive fragments as being contiguous (even if a level switch occurs),
// for sake of clarity:
// consecutive fragments are frags with
// - less than 100ms gaps between new time offset (if accurate) and next expected PTS OR
// - less than 20 audio frames distance
// contiguous fragments are consecutive fragments from same quality level (same level, new SN = old SN + 1)
// this helps ensuring audio continuity
// and this also avoids audio glitches/cut when switching quality, or reporting wrong duration on first audio frame
contiguous |= (inputSamples.length && nextAudioPts &&
((accurateTimeOffset && Math.abs(timeOffset - nextAudioPts / inputTimeScale) < 0.1) ||
Math.abs((inputSamples[0].pts - nextAudioPts - initDTS)) < 20 * inputSampleDuration)
);
// compute normalized PTS
inputSamples.forEach(function (sample) {
sample.pts = sample.dts = ptsNormalize(sample.pts - initDTS, timeOffset * inputTimeScale);
});
// filter out sample with negative PTS that are not playable anyway
// if we don't remove these negative samples, they will shift all audio samples forward.
// leading to audio overlap between current / next fragment
inputSamples = inputSamples.filter(function (sample) {
return sample.pts >= 0;
});
// in case all samples have negative PTS, and have been filtered out, return now
if (inputSamples.length === 0)
return;
if (!contiguous) {
if (!accurateTimeOffset) {
// if frag are mot contiguous and if we cant trust time offset, let's use first sample PTS as next audio PTS
nextAudioPts = inputSamples[0].pts;
} else {
// if timeOffset is accurate, let's use it as predicted next audio PTS
nextAudioPts = timeOffset * inputTimeScale;
}
}
// If the audio track is missing samples, the frames seem to get "left-shifted" within the
// resulting mp4 segment, causing sync issues and leaving gaps at the end of the audio segment.
// In an effort to prevent this from happening, we inject frames here where there are gaps.
// When possible, we inject a silent frame; when that's not possible, we duplicate the last
// frame.
if (track.isAAC) {
const maxAudioFramesDrift = this.config.maxAudioFramesDrift;
for (let i = 0, nextPts = nextAudioPts; i < inputSamples.length;) {
// First, let's see how far off this frame is from where we expect it to be
var sample = inputSamples[i], delta;
let pts = sample.pts;
delta = pts - nextPts;
const duration = Math.abs(1000 * delta / inputTimeScale);
// If we're overlapping by more than a duration, drop this sample
if (delta <= -maxAudioFramesDrift * inputSampleDuration) {
logger.warn(`Dropping 1 audio frame @ ${(nextPts / inputTimeScale).toFixed(3)}s due to ${Math.round(duration)} ms overlap.`);
inputSamples.splice(i, 1);
track.len -= sample.unit.length;
// Don't touch nextPtsNorm or i
} // eslint-disable-line brace-style
// Insert missing frames if:
// 1: We're more than maxAudioFramesDrift frame away
// 2: Not more than MAX_SILENT_FRAME_DURATION away
// 3: currentTime (aka nextPtsNorm) is not 0
else if (delta >= maxAudioFramesDrift * inputSampleDuration && duration < MAX_SILENT_FRAME_DURATION && nextPts) {
let missing = Math.round(delta / inputSampleDuration);
logger.warn(`Injecting ${missing} audio frame @ ${(nextPts / inputTimeScale).toFixed(3)}s due to ${Math.round(1000 * delta / inputTimeScale)} ms gap.`);
for (let j = 0; j < missing; j++) {
let newStamp = Math.max(nextPts, 0);
fillFrame = AAC.getSilentFrame(track.manifestCodec || track.codec, track.channelCount);
if (!fillFrame) {
logger.log('Unable to get silent frame for given audio codec; duplicating last frame instead.');
fillFrame = sample.unit.subarray();
}
inputSamples.splice(i, 0, { unit: fillFrame, pts: newStamp, dts: newStamp });
track.len += fillFrame.length;
nextPts += inputSampleDuration;
i++;
}
// Adjust sample to next expected pts
sample.pts = sample.dts = nextPts;
nextPts += inputSampleDuration;
i++;
} else {
// Otherwise, just adjust pts
if (Math.abs(delta) > (0.1 * inputSampleDuration)) {
// logger.log(`Invalid frame delta ${Math.round(delta + inputSampleDuration)} at PTS ${Math.round(pts / 90)} (should be ${Math.round(inputSampleDuration)}).`);
}
sample.pts = sample.dts = nextPts;
nextPts += inputSampleDuration;
i++;
}
}
}
for (let j = 0, nbSamples = inputSamples.length; j < nbSamples; j++) {
let audioSample = inputSamples[j];
let unit = audioSample.unit;
let pts = audioSample.pts;
// logger.log(`Audio/PTS:${Math.round(pts/90)}`);
// if not first sample
if (lastPTS !== undefined) {
mp4Sample.duration = Math.round((pts - lastPTS) / scaleFactor);
} else {
let delta = Math.round(1000 * (pts - nextAudioPts) / inputTimeScale),
numMissingFrames = 0;
// if fragment are contiguous, detect hole/overlapping between fragments
// contiguous fragments are consecutive fragments from same quality level (same level, new SN = old SN + 1)
if (contiguous && track.isAAC) {
// log delta
if (delta) {
if (delta > 0 && delta < MAX_SILENT_FRAME_DURATION) {
numMissingFrames = Math.round((pts - nextAudioPts) / inputSampleDuration);
logger.log(`${delta} ms hole between AAC samples detected,filling it`);
if (numMissingFrames > 0) {
fillFrame = AAC.getSilentFrame(track.manifestCodec || track.codec, track.channelCount);
if (!fillFrame)
fillFrame = unit.subarray();
track.len += numMissingFrames * fillFrame.length;
}
// if we have frame overlap, overlapping for more than half a frame duraion
} else if (delta < -12) {
// drop overlapping audio frames... browser will deal with it
logger.log(`drop overlapping AAC sample, expected/parsed/delta:${(nextAudioPts / inputTimeScale).toFixed(3)}s/${(pts / inputTimeScale).toFixed(3)}s/${(-delta)}ms`);
track.len -= unit.byteLength;
continue;
}
// set PTS/DTS to expected PTS/DTS
pts = nextAudioPts;
}
}
// remember first PTS of our audioSamples
firstPTS = pts;
if (track.len > 0) {
/* concatenate the audio data and construct the mdat in place
(need 8 more bytes to fill length and mdat type) */
let mdatSize = rawMPEG ? track.len : track.len + 8;
offset = rawMPEG ? 0 : 8;
try {
mdat = new Uint8Array(mdatSize);
} catch (err) {
this.observer.trigger(Event.ERROR, { type: ErrorTypes.MUX_ERROR, details: ErrorDetails.REMUX_ALLOC_ERROR, fatal: false, bytes: mdatSize, reason: `fail allocating audio mdat ${mdatSize}` });
return;
}
if (!rawMPEG) {
const view = new DataView(mdat.buffer);
view.setUint32(0, mdatSize);
mdat.set(MP4.types.mdat, 4);
}
} else {
// no audio samples
return;
}
for (let i = 0; i < numMissingFrames; i++) {
fillFrame = AAC.getSilentFrame(track.manifestCodec || track.codec, track.channelCount);
if (!fillFrame) {
logger.log('Unable to get silent frame for given audio codec; duplicating this frame instead.');
fillFrame = unit.subarray();
}
mdat.set(fillFrame, offset);
offset += fillFrame.byteLength;
mp4Sample = {
size: fillFrame.byteLength,
cts: 0,
duration: 1024,
flags: {
isLeading: 0,
isDependedOn: 0,
hasRedundancy: 0,
degradPrio: 0,
dependsOn: 1
}
};
outputSamples.push(mp4Sample);
}
}
mdat.set(unit, offset);
let unitLen = unit.byteLength;
offset += unitLen;
// console.log('PTS/DTS/initDTS/normPTS/normDTS/relative PTS : ${audioSample.pts}/${audioSample.dts}/${initDTS}/${ptsnorm}/${dtsnorm}/${(audioSample.pts/4294967296).toFixed(3)}');
mp4Sample = {
size: unitLen,
cts: 0,
duration: 0,
flags: {
isLeading: 0,
isDependedOn: 0,
hasRedundancy: 0,
degradPrio: 0,
dependsOn: 1
}
};
outputSamples.push(mp4Sample);
lastPTS = pts;
}
let lastSampleDuration = 0;
let nbSamples = outputSamples.length;
// set last sample duration as being identical to previous sample
if (nbSamples >= 2) {
lastSampleDuration = outputSamples[nbSamples - 2].duration;
mp4Sample.duration = lastSampleDuration;
}
if (nbSamples) {
// next audio sample PTS should be equal to last sample PTS + duration
this.nextAudioPts = nextAudioPts = lastPTS + scaleFactor * lastSampleDuration;
// logger.log('Audio/PTS/PTSend:' + audioSample.pts.toFixed(0) + '/' + this.nextAacDts.toFixed(0));
track.len = 0;
track.samples = outputSamples;
if (rawMPEG)
moof = new Uint8Array();
else
moof = MP4.moof(track.sequenceNumber++, firstPTS / scaleFactor, track);
track.samples = [];
const start = firstPTS / inputTimeScale;
const end = nextAudioPts / inputTimeScale;
const audioData = {
data1: moof,
data2: mdat,
startPTS: start,
endPTS: end,
startDTS: start,
endDTS: end,
type: 'audio',
hasAudio: true,
hasVideo: false,
nb: nbSamples
};
this.observer.trigger(Event.FRAG_PARSING_DATA, audioData);
return audioData;
}
return null;
}
remuxEmptyAudio (track, timeOffset, contiguous, videoData) {
let inputTimeScale = track.inputTimeScale,
mp4timeScale = track.samplerate ? track.samplerate : inputTimeScale,
scaleFactor = inputTimeScale / mp4timeScale,
nextAudioPts = this.nextAudioPts,
// sync with video's timestamp
startDTS = (nextAudioPts !== undefined ? nextAudioPts : videoData.startDTS * inputTimeScale) + this._initDTS,
endDTS = videoData.endDTS * inputTimeScale + this._initDTS,
// one sample's duration value
sampleDuration = 1024,
frameDuration = scaleFactor * sampleDuration,
// samples count of this segment's duration
nbSamples = Math.ceil((endDTS - startDTS) / frameDuration),
// silent frame
silentFrame = AAC.getSilentFrame(track.manifestCodec || track.codec, track.channelCount);
logger.warn('remux empty Audio');
// Can't remux if we can't generate a silent frame...
if (!silentFrame) {
logger.trace('Unable to remuxEmptyAudio since we were unable to get a silent frame for given audio codec!');
return;
}
let samples = [];
for (let i = 0; i < nbSamples; i++) {
let stamp = startDTS + i * frameDuration;
samples.push({ unit: silentFrame, pts: stamp, dts: stamp });
track.len += silentFrame.length;
}
track.samples = samples;
this.remuxAudio(track, timeOffset, contiguous);
}
remuxID3 (track, timeOffset) {
let length = track.samples.length, sample;
const inputTimeScale = track.inputTimeScale;
const initPTS = this._initPTS;
const initDTS = this._initDTS;
// consume samples
if (length) {
for (let index = 0; index < length; index++) {
sample = track.samples[index];
// setting id3 pts, dts to relative time
// using this._initPTS and this._initDTS to calculate relative time
sample.pts = ((sample.pts - initPTS) / inputTimeScale);
sample.dts = ((sample.dts - initDTS) / inputTimeScale);
}
this.observer.trigger(Event.FRAG_PARSING_METADATA, {
samples: track.samples
});
}
track.samples = [];
timeOffset = timeOffset;
}
remuxText (track, timeOffset) {
track.samples.sort(function (a, b) {
return (a.pts - b.pts);
});
let length = track.samples.length, sample;
const inputTimeScale = track.inputTimeScale;
const initPTS = this._initPTS;
// consume samples
if (length) {
for (let index = 0; index < length; index++) {
sample = track.samples[index];
// setting text pts, dts to relative time
// using this._initPTS and this._initDTS to calculate relative time
sample.pts = ((sample.pts - initPTS) / inputTimeScale);
}
this.observer.trigger(Event.FRAG_PARSING_USERDATA, {
samples: track.samples
});
}
track.samples = [];
timeOffset = timeOffset;
}
_PTSNormalize (value, reference) {
let offset;
if (reference === undefined)
return value;
if (reference < value) {
// - 2^33
offset = -8589934592;
} else {
// + 2^33
offset = 8589934592;
}
/* PTS is 33bit (from 0 to 2^33 -1)
if diff between value and reference is bigger than half of the amplitude (2^32) then it means that
PTS looping occured. fill the gap */
while (Math.abs(value - reference) > 4294967296)
value += offset;
return value;
}
}
export default MP4Remuxer;
|