feat: Waku v2 bridge

Issue #12610

vendor/github.com/pion/rtp/codecs/codecs.go

@@ -0,0 +1,2 @@
+// Package codecs implements codec specific RTP payloader/depayloaders
+package codecs

vendor/github.com/pion/rtp/codecs/common.go

@@ -0,0 +1,26 @@
+package codecs
+
+func min(a, b int) int {
+	if a < b {
+		return a
+	}
+	return b
+}
+
+// audioDepacketizer is a mixin for audio codec depacketizers
+type audioDepacketizer struct{}
+
+func (d *audioDepacketizer) IsPartitionTail(marker bool, payload []byte) bool {
+	return true
+}
+
+func (d *audioDepacketizer) IsPartitionHead(payload []byte) bool {
+	return true
+}
+
+// videoDepacketizer is a mixin for video codec depacketizers
+type videoDepacketizer struct{}
+
+func (d *videoDepacketizer) IsPartitionTail(marker bool, payload []byte) bool {
+	return marker
+}

vendor/github.com/pion/rtp/codecs/error.go

@@ -0,0 +1,11 @@
+package codecs
+
+import "errors"
+
+var (
+	errShortPacket          = errors.New("packet is not large enough")
+	errNilPacket            = errors.New("invalid nil packet")
+	errTooManyPDiff         = errors.New("too many PDiff")
+	errTooManySpatialLayers = errors.New("too many spatial layers")
+	errUnhandledNALUType    = errors.New("NALU Type is unhandled")
+)

vendor/github.com/pion/rtp/codecs/g711_packet.go

@@ -0,0 +1,22 @@
+package codecs
+
+// G711Payloader payloads G711 packets
+type G711Payloader struct{}
+
+// Payload fragments an G711 packet across one or more byte arrays
+func (p *G711Payloader) Payload(mtu uint16, payload []byte) [][]byte {
+	var out [][]byte
+	if payload == nil || mtu <= 0 {
+		return out
+	}
+
+	for len(payload) > int(mtu) {
+		o := make([]byte, mtu)
+		copy(o, payload[:mtu])
+		payload = payload[mtu:]
+		out = append(out, o)
+	}
+	o := make([]byte, len(payload))
+	copy(o, payload)
+	return append(out, o)
+}

vendor/github.com/pion/rtp/codecs/g722_packet.go

@@ -0,0 +1,22 @@
+package codecs
+
+// G722Payloader payloads G722 packets
+type G722Payloader struct{}
+
+// Payload fragments an G722 packet across one or more byte arrays
+func (p *G722Payloader) Payload(mtu uint16, payload []byte) [][]byte {
+	var out [][]byte
+	if payload == nil || mtu == 0 {
+		return out
+	}
+
+	for len(payload) > int(mtu) {
+		o := make([]byte, mtu)
+		copy(o, payload[:mtu])
+		payload = payload[mtu:]
+		out = append(out, o)
+	}
+	o := make([]byte, len(payload))
+	copy(o, payload)
+	return append(out, o)
+}

vendor/github.com/pion/rtp/codecs/h264_packet.go

@@ -0,0 +1,285 @@
+package codecs
+
+import (
+	"encoding/binary"
+	"fmt"
+)
+
+// H264Payloader payloads H264 packets
+type H264Payloader struct {
+	spsNalu, ppsNalu []byte
+}
+
+const (
+	stapaNALUType  = 24
+	fuaNALUType    = 28
+	fubNALUType    = 29
+	spsNALUType    = 7
+	ppsNALUType    = 8
+	audNALUType    = 9
+	fillerNALUType = 12
+
+	fuaHeaderSize       = 2
+	stapaHeaderSize     = 1
+	stapaNALULengthSize = 2
+
+	naluTypeBitmask   = 0x1F
+	naluRefIdcBitmask = 0x60
+	fuStartBitmask    = 0x80
+	fuEndBitmask      = 0x40
+
+	outputStapAHeader = 0x78
+)
+
+func annexbNALUStartCode() []byte { return []byte{0x00, 0x00, 0x00, 0x01} }
+
+func emitNalus(nals []byte, emit func([]byte)) {
+	nextInd := func(nalu []byte, start int) (indStart int, indLen int) {
+		zeroCount := 0
+
+		for i, b := range nalu[start:] {
+			if b == 0 {
+				zeroCount++
+				continue
+			} else if b == 1 {
+				if zeroCount >= 2 {
+					return start + i - zeroCount, zeroCount + 1
+				}
+			}
+			zeroCount = 0
+		}
+		return -1, -1
+	}
+
+	nextIndStart, nextIndLen := nextInd(nals, 0)
+	if nextIndStart == -1 {
+		emit(nals)
+	} else {
+		for nextIndStart != -1 {
+			prevStart := nextIndStart + nextIndLen
+			nextIndStart, nextIndLen = nextInd(nals, prevStart)
+			if nextIndStart != -1 {
+				emit(nals[prevStart:nextIndStart])
+			} else {
+				// Emit until end of stream, no end indicator found
+				emit(nals[prevStart:])
+			}
+		}
+	}
+}
+
+// Payload fragments a H264 packet across one or more byte arrays
+func (p *H264Payloader) Payload(mtu uint16, payload []byte) [][]byte {
+	var payloads [][]byte
+	if len(payload) == 0 {
+		return payloads
+	}
+
+	emitNalus(payload, func(nalu []byte) {
+		if len(nalu) == 0 {
+			return
+		}
+
+		naluType := nalu[0] & naluTypeBitmask
+		naluRefIdc := nalu[0] & naluRefIdcBitmask
+
+		switch {
+		case naluType == audNALUType || naluType == fillerNALUType:
+			return
+		case naluType == spsNALUType:
+			p.spsNalu = nalu
+			return
+		case naluType == ppsNALUType:
+			p.ppsNalu = nalu
+			return
+		case p.spsNalu != nil && p.ppsNalu != nil:
+			// Pack current NALU with SPS and PPS as STAP-A
+			spsLen := make([]byte, 2)
+			binary.BigEndian.PutUint16(spsLen, uint16(len(p.spsNalu)))
+
+			ppsLen := make([]byte, 2)
+			binary.BigEndian.PutUint16(ppsLen, uint16(len(p.ppsNalu)))
+
+			stapANalu := []byte{outputStapAHeader}
+			stapANalu = append(stapANalu, spsLen...)
+			stapANalu = append(stapANalu, p.spsNalu...)
+			stapANalu = append(stapANalu, ppsLen...)
+			stapANalu = append(stapANalu, p.ppsNalu...)
+			if len(stapANalu) <= int(mtu) {
+				out := make([]byte, len(stapANalu))
+				copy(out, stapANalu)
+				payloads = append(payloads, out)
+			}
+
+			p.spsNalu = nil
+			p.ppsNalu = nil
+		}
+
+		// Single NALU
+		if len(nalu) <= int(mtu) {
+			out := make([]byte, len(nalu))
+			copy(out, nalu)
+			payloads = append(payloads, out)
+			return
+		}
+
+		// FU-A
+		maxFragmentSize := int(mtu) - fuaHeaderSize
+
+		// The FU payload consists of fragments of the payload of the fragmented
+		// NAL unit so that if the fragmentation unit payloads of consecutive
+		// FUs are sequentially concatenated, the payload of the fragmented NAL
+		// unit can be reconstructed.  The NAL unit type octet of the fragmented
+		// NAL unit is not included as such in the fragmentation unit payload,
+		// 	but rather the information of the NAL unit type octet of the
+		// fragmented NAL unit is conveyed in the F and NRI fields of the FU
+		// indicator octet of the fragmentation unit and in the type field of
+		// the FU header.  An FU payload MAY have any number of octets and MAY
+		// be empty.
+
+		naluData := nalu
+		// According to the RFC, the first octet is skipped due to redundant information
+		naluDataIndex := 1
+		naluDataLength := len(nalu) - naluDataIndex
+		naluDataRemaining := naluDataLength
+
+		if min(maxFragmentSize, naluDataRemaining) <= 0 {
+			return
+		}
+
+		for naluDataRemaining > 0 {
+			currentFragmentSize := min(maxFragmentSize, naluDataRemaining)
+			out := make([]byte, fuaHeaderSize+currentFragmentSize)
+
+			// +---------------+
+			// |0|1|2|3|4|5|6|7|
+			// +-+-+-+-+-+-+-+-+
+			// |F|NRI|  Type   |
+			// +---------------+
+			out[0] = fuaNALUType
+			out[0] |= naluRefIdc
+
+			// +---------------+
+			// |0|1|2|3|4|5|6|7|
+			// +-+-+-+-+-+-+-+-+
+			// |S|E|R|  Type   |
+			// +---------------+
+
+			out[1] = naluType
+			if naluDataRemaining == naluDataLength {
+				// Set start bit
+				out[1] |= 1 << 7
+			} else if naluDataRemaining-currentFragmentSize == 0 {
+				// Set end bit
+				out[1] |= 1 << 6
+			}
+
+			copy(out[fuaHeaderSize:], naluData[naluDataIndex:naluDataIndex+currentFragmentSize])
+			payloads = append(payloads, out)
+
+			naluDataRemaining -= currentFragmentSize
+			naluDataIndex += currentFragmentSize
+		}
+	})
+
+	return payloads
+}
+
+// H264Packet represents the H264 header that is stored in the payload of an RTP Packet
+type H264Packet struct {
+	IsAVC     bool
+	fuaBuffer []byte
+
+	videoDepacketizer
+}
+
+func (p *H264Packet) doPackaging(nalu []byte) []byte {
+	if p.IsAVC {
+		naluLength := make([]byte, 4)
+		binary.BigEndian.PutUint32(naluLength, uint32(len(nalu)))
+		return append(naluLength, nalu...)
+	}
+
+	return append(annexbNALUStartCode(), nalu...)
+}
+
+// IsDetectedFinalPacketInSequence returns true of the packet passed in has the
+// marker bit set indicated the end of a packet sequence
+func (p *H264Packet) IsDetectedFinalPacketInSequence(rtpPacketMarketBit bool) bool {
+	return rtpPacketMarketBit
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H264Packet this method is called upon
+func (p *H264Packet) Unmarshal(payload []byte) ([]byte, error) {
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= 2 {
+		return nil, fmt.Errorf("%w: %d <= 2", errShortPacket, len(payload))
+	}
+
+	// NALU Types
+	// https://tools.ietf.org/html/rfc6184#section-5.4
+	naluType := payload[0] & naluTypeBitmask
+	switch {
+	case naluType > 0 && naluType < 24:
+		return p.doPackaging(payload), nil
+
+	case naluType == stapaNALUType:
+		currOffset := int(stapaHeaderSize)
+		result := []byte{}
+		for currOffset < len(payload) {
+			naluSize := int(binary.BigEndian.Uint16(payload[currOffset:]))
+			currOffset += stapaNALULengthSize
+
+			if len(payload) < currOffset+naluSize {
+				return nil, fmt.Errorf("%w STAP-A declared size(%d) is larger than buffer(%d)", errShortPacket, naluSize, len(payload)-currOffset)
+			}
+
+			result = append(result, p.doPackaging(payload[currOffset:currOffset+naluSize])...)
+			currOffset += naluSize
+		}
+		return result, nil
+
+	case naluType == fuaNALUType:
+		if len(payload) < fuaHeaderSize {
+			return nil, errShortPacket
+		}
+
+		if p.fuaBuffer == nil {
+			p.fuaBuffer = []byte{}
+		}
+
+		p.fuaBuffer = append(p.fuaBuffer, payload[fuaHeaderSize:]...)
+
+		if payload[1]&fuEndBitmask != 0 {
+			naluRefIdc := payload[0] & naluRefIdcBitmask
+			fragmentedNaluType := payload[1] & naluTypeBitmask
+
+			nalu := append([]byte{}, naluRefIdc|fragmentedNaluType)
+			nalu = append(nalu, p.fuaBuffer...)
+			p.fuaBuffer = nil
+			return p.doPackaging(nalu), nil
+		}
+
+		return []byte{}, nil
+	}
+
+	return nil, fmt.Errorf("%w: %d", errUnhandledNALUType, naluType)
+}
+
+// H264PartitionHeadChecker is obsolete
+type H264PartitionHeadChecker struct{}
+
+// IsPartitionHead checks if this is the head of a packetized nalu stream.
+func (*H264Packet) IsPartitionHead(payload []byte) bool {
+	if len(payload) < 2 {
+		return false
+	}
+
+	if payload[0]&naluTypeBitmask == fuaNALUType ||
+		payload[0]&naluTypeBitmask == fubNALUType {
+		return payload[1]&fuStartBitmask != 0
+	}
+
+	return true
+}

vendor/github.com/pion/rtp/codecs/h265_packet.go

@@ -0,0 +1,803 @@
+package codecs
+
+import (
+	"errors"
+	"fmt"
+)
+
+//
+// Errors
+//
+
+var (
+	errH265CorruptedPacket   = errors.New("corrupted h265 packet")
+	errInvalidH265PacketType = errors.New("invalid h265 packet type")
+)
+
+//
+// Network Abstraction Unit Header implementation
+//
+
+const (
+	// sizeof(uint16)
+	h265NaluHeaderSize = 2
+	// https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.2
+	h265NaluAggregationPacketType = 48
+	// https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.3
+	h265NaluFragmentationUnitType = 49
+	// https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.4
+	h265NaluPACIPacketType = 50
+)
+
+// H265NALUHeader is a H265 NAL Unit Header
+// https://datatracker.ietf.org/doc/html/rfc7798#section-1.1.4
+// +---------------+---------------+
+//  |0|1|2|3|4|5|6|7|0|1|2|3|4|5|6|7|
+//  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//  |F|   Type    |  LayerID  | TID |
+//  +-------------+-----------------+
+type H265NALUHeader uint16
+
+func newH265NALUHeader(highByte, lowByte uint8) H265NALUHeader {
+	return H265NALUHeader((uint16(highByte) << 8) | uint16(lowByte))
+}
+
+// F is the forbidden bit, should always be 0.
+func (h H265NALUHeader) F() bool {
+	return (uint16(h) >> 15) != 0
+}
+
+// Type of NAL Unit.
+func (h H265NALUHeader) Type() uint8 {
+	// 01111110 00000000
+	const mask = 0b01111110 << 8
+	return uint8((uint16(h) & mask) >> (8 + 1))
+}
+
+// IsTypeVCLUnit returns whether or not the NAL Unit type is a VCL NAL unit.
+func (h H265NALUHeader) IsTypeVCLUnit() bool {
+	// Type is coded on 6 bits
+	const msbMask = 0b00100000
+	return (h.Type() & msbMask) == 0
+}
+
+// LayerID should always be 0 in non-3D HEVC context.
+func (h H265NALUHeader) LayerID() uint8 {
+	// 00000001 11111000
+	const mask = (0b00000001 << 8) | 0b11111000
+	return uint8((uint16(h) & mask) >> 3)
+}
+
+// TID is the temporal identifier of the NAL unit +1.
+func (h H265NALUHeader) TID() uint8 {
+	const mask = 0b00000111
+	return uint8(uint16(h) & mask)
+}
+
+// IsAggregationPacket returns whether or not the packet is an Aggregation packet.
+func (h H265NALUHeader) IsAggregationPacket() bool {
+	return h.Type() == h265NaluAggregationPacketType
+}
+
+// IsFragmentationUnit returns whether or not the packet is a Fragmentation Unit packet.
+func (h H265NALUHeader) IsFragmentationUnit() bool {
+	return h.Type() == h265NaluFragmentationUnitType
+}
+
+// IsPACIPacket returns whether or not the packet is a PACI packet.
+func (h H265NALUHeader) IsPACIPacket() bool {
+	return h.Type() == h265NaluPACIPacketType
+}
+
+//
+// Single NAL Unit Packet implementation
+//
+
+// H265SingleNALUnitPacket represents a NALU packet, containing exactly one NAL unit.
+//     0                   1                   2                   3
+//    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |           PayloadHdr          |      DONL (conditional)       |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                                                               |
+//   |                  NAL unit payload data                        |
+//   |                                                               |
+//   |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                               :...OPTIONAL RTP padding        |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.1
+type H265SingleNALUnitPacket struct {
+	// payloadHeader is the header of the H265 packet.
+	payloadHeader H265NALUHeader
+	// donl is a 16-bit field, that may or may not be present.
+	donl *uint16
+	// payload of the fragmentation unit.
+	payload []byte
+
+	mightNeedDONL bool
+}
+
+// WithDONL can be called to specify whether or not DONL might be parsed.
+// DONL may need to be parsed if `sprop-max-don-diff` is greater than 0 on the RTP stream.
+func (p *H265SingleNALUnitPacket) WithDONL(value bool) {
+	p.mightNeedDONL = value
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H265SingleNALUnitPacket this method is called upon.
+func (p *H265SingleNALUnitPacket) Unmarshal(payload []byte) ([]byte, error) {
+	// sizeof(headers)
+	const totalHeaderSize = h265NaluHeaderSize
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= totalHeaderSize {
+		return nil, fmt.Errorf("%w: %d <= %v", errShortPacket, len(payload), totalHeaderSize)
+	}
+
+	payloadHeader := newH265NALUHeader(payload[0], payload[1])
+	if payloadHeader.F() {
+		return nil, errH265CorruptedPacket
+	}
+	if payloadHeader.IsFragmentationUnit() || payloadHeader.IsPACIPacket() || payloadHeader.IsAggregationPacket() {
+		return nil, errInvalidH265PacketType
+	}
+
+	payload = payload[2:]
+
+	if p.mightNeedDONL {
+		// sizeof(uint16)
+		if len(payload) <= 2 {
+			return nil, errShortPacket
+		}
+
+		donl := (uint16(payload[0]) << 8) | uint16(payload[1])
+		p.donl = &donl
+		payload = payload[2:]
+	}
+
+	p.payloadHeader = payloadHeader
+	p.payload = payload
+
+	return nil, nil
+}
+
+// PayloadHeader returns the NALU header of the packet.
+func (p *H265SingleNALUnitPacket) PayloadHeader() H265NALUHeader {
+	return p.payloadHeader
+}
+
+// DONL returns the DONL of the packet.
+func (p *H265SingleNALUnitPacket) DONL() *uint16 {
+	return p.donl
+}
+
+// Payload returns the Fragmentation Unit packet payload.
+func (p *H265SingleNALUnitPacket) Payload() []byte {
+	return p.payload
+}
+
+func (p *H265SingleNALUnitPacket) isH265Packet() {}
+
+//
+// Aggregation Packets implementation
+//
+
+// H265AggregationUnitFirst represent the First Aggregation Unit in an AP.
+//
+//    0                   1                   2                   3
+//    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//                   :       DONL (conditional)      |   NALU size   |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |   NALU size   |                                               |
+//   +-+-+-+-+-+-+-+-+         NAL unit                              |
+//   |                                                               |
+//   |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                               :
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.2
+type H265AggregationUnitFirst struct {
+	donl        *uint16
+	nalUnitSize uint16
+	nalUnit     []byte
+}
+
+// DONL field, when present, specifies the value of the 16 least
+// significant bits of the decoding order number of the aggregated NAL
+// unit.
+func (u H265AggregationUnitFirst) DONL() *uint16 {
+	return u.donl
+}
+
+// NALUSize represents the size, in bytes, of the NalUnit.
+func (u H265AggregationUnitFirst) NALUSize() uint16 {
+	return u.nalUnitSize
+}
+
+// NalUnit payload.
+func (u H265AggregationUnitFirst) NalUnit() []byte {
+	return u.nalUnit
+}
+
+// H265AggregationUnit represent the an Aggregation Unit in an AP, which is not the first one.
+//
+//    0                   1                   2                   3
+//    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//                   : DOND (cond)   |          NALU size            |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                                                               |
+//   |                       NAL unit                                |
+//   |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                               :
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.2
+type H265AggregationUnit struct {
+	dond        *uint8
+	nalUnitSize uint16
+	nalUnit     []byte
+}
+
+// DOND field plus 1 specifies the difference between
+// the decoding order number values of the current aggregated NAL unit
+// and the preceding aggregated NAL unit in the same AP.
+func (u H265AggregationUnit) DOND() *uint8 {
+	return u.dond
+}
+
+// NALUSize represents the size, in bytes, of the NalUnit.
+func (u H265AggregationUnit) NALUSize() uint16 {
+	return u.nalUnitSize
+}
+
+// NalUnit payload.
+func (u H265AggregationUnit) NalUnit() []byte {
+	return u.nalUnit
+}
+
+// H265AggregationPacket represents an Aggregation packet.
+//   0                   1                   2                   3
+//    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |    PayloadHdr (Type=48)       |                               |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+                               |
+//   |                                                               |
+//   |             two or more aggregation units                     |
+//   |                                                               |
+//   |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//   |                               :...OPTIONAL RTP padding        |
+//   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.2
+type H265AggregationPacket struct {
+	firstUnit  *H265AggregationUnitFirst
+	otherUnits []H265AggregationUnit
+
+	mightNeedDONL bool
+}
+
+// WithDONL can be called to specify whether or not DONL might be parsed.
+// DONL may need to be parsed if `sprop-max-don-diff` is greater than 0 on the RTP stream.
+func (p *H265AggregationPacket) WithDONL(value bool) {
+	p.mightNeedDONL = value
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H265AggregationPacket this method is called upon.
+func (p *H265AggregationPacket) Unmarshal(payload []byte) ([]byte, error) {
+	// sizeof(headers)
+	const totalHeaderSize = h265NaluHeaderSize
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= totalHeaderSize {
+		return nil, fmt.Errorf("%w: %d <= %v", errShortPacket, len(payload), totalHeaderSize)
+	}
+
+	payloadHeader := newH265NALUHeader(payload[0], payload[1])
+	if payloadHeader.F() {
+		return nil, errH265CorruptedPacket
+	}
+	if !payloadHeader.IsAggregationPacket() {
+		return nil, errInvalidH265PacketType
+	}
+
+	// First parse the first aggregation unit
+	payload = payload[2:]
+	firstUnit := &H265AggregationUnitFirst{}
+
+	if p.mightNeedDONL {
+		if len(payload) < 2 {
+			return nil, errShortPacket
+		}
+
+		donl := (uint16(payload[0]) << 8) | uint16(payload[1])
+		firstUnit.donl = &donl
+
+		payload = payload[2:]
+	}
+	if len(payload) < 2 {
+		return nil, errShortPacket
+	}
+	firstUnit.nalUnitSize = (uint16(payload[0]) << 8) | uint16(payload[1])
+	payload = payload[2:]
+
+	if len(payload) < int(firstUnit.nalUnitSize) {
+		return nil, errShortPacket
+	}
+
+	firstUnit.nalUnit = payload[:firstUnit.nalUnitSize]
+	payload = payload[firstUnit.nalUnitSize:]
+
+	// Parse remaining Aggregation Units
+	var units []H265AggregationUnit
+	for {
+		unit := H265AggregationUnit{}
+
+		if p.mightNeedDONL {
+			if len(payload) < 1 {
+				break
+			}
+
+			dond := payload[0]
+			unit.dond = &dond
+
+			payload = payload[1:]
+		}
+
+		if len(payload) < 2 {
+			break
+		}
+		unit.nalUnitSize = (uint16(payload[0]) << 8) | uint16(payload[1])
+		payload = payload[2:]
+
+		if len(payload) < int(unit.nalUnitSize) {
+			break
+		}
+
+		unit.nalUnit = payload[:unit.nalUnitSize]
+		payload = payload[unit.nalUnitSize:]
+
+		units = append(units, unit)
+	}
+
+	// There need to be **at least** two Aggregation Units (first + another one)
+	if len(units) == 0 {
+		return nil, errShortPacket
+	}
+
+	p.firstUnit = firstUnit
+	p.otherUnits = units
+
+	return nil, nil
+}
+
+// FirstUnit returns the first Aggregated Unit of the packet.
+func (p *H265AggregationPacket) FirstUnit() *H265AggregationUnitFirst {
+	return p.firstUnit
+}
+
+// OtherUnits returns the all the other Aggregated Unit of the packet (excluding the first one).
+func (p *H265AggregationPacket) OtherUnits() []H265AggregationUnit {
+	return p.otherUnits
+}
+
+func (p *H265AggregationPacket) isH265Packet() {}
+
+//
+// Fragmentation Unit implementation
+//
+
+const (
+	// sizeof(uint8)
+	h265FragmentationUnitHeaderSize = 1
+)
+
+// H265FragmentationUnitHeader is a H265 FU Header
+// +---------------+
+// |0|1|2|3|4|5|6|7|
+// +-+-+-+-+-+-+-+-+
+// |S|E|  FuType   |
+// +---------------+
+type H265FragmentationUnitHeader uint8
+
+// S represents the start of a fragmented NAL unit.
+func (h H265FragmentationUnitHeader) S() bool {
+	const mask = 0b10000000
+	return ((h & mask) >> 7) != 0
+}
+
+// E represents the end of a fragmented NAL unit.
+func (h H265FragmentationUnitHeader) E() bool {
+	const mask = 0b01000000
+	return ((h & mask) >> 6) != 0
+}
+
+// FuType MUST be equal to the field Type of the fragmented NAL unit.
+func (h H265FragmentationUnitHeader) FuType() uint8 {
+	const mask = 0b00111111
+	return uint8(h) & mask
+}
+
+// H265FragmentationUnitPacket represents a single Fragmentation Unit packet.
+//
+//  0                   1                   2                   3
+// 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+// |    PayloadHdr (Type=49)       |   FU header   | DONL (cond)   |
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-|
+// | DONL (cond)   |                                               |
+// |-+-+-+-+-+-+-+-+                                               |
+// |                         FU payload                            |
+// |                                                               |
+// |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+// |                               :...OPTIONAL RTP padding        |
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.3
+type H265FragmentationUnitPacket struct {
+	// payloadHeader is the header of the H265 packet.
+	payloadHeader H265NALUHeader
+	// fuHeader is the header of the fragmentation unit
+	fuHeader H265FragmentationUnitHeader
+	// donl is a 16-bit field, that may or may not be present.
+	donl *uint16
+	// payload of the fragmentation unit.
+	payload []byte
+
+	mightNeedDONL bool
+}
+
+// WithDONL can be called to specify whether or not DONL might be parsed.
+// DONL may need to be parsed if `sprop-max-don-diff` is greater than 0 on the RTP stream.
+func (p *H265FragmentationUnitPacket) WithDONL(value bool) {
+	p.mightNeedDONL = value
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H265FragmentationUnitPacket this method is called upon.
+func (p *H265FragmentationUnitPacket) Unmarshal(payload []byte) ([]byte, error) {
+	// sizeof(headers)
+	const totalHeaderSize = h265NaluHeaderSize + h265FragmentationUnitHeaderSize
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= totalHeaderSize {
+		return nil, fmt.Errorf("%w: %d <= %v", errShortPacket, len(payload), totalHeaderSize)
+	}
+
+	payloadHeader := newH265NALUHeader(payload[0], payload[1])
+	if payloadHeader.F() {
+		return nil, errH265CorruptedPacket
+	}
+	if !payloadHeader.IsFragmentationUnit() {
+		return nil, errInvalidH265PacketType
+	}
+
+	fuHeader := H265FragmentationUnitHeader(payload[2])
+	payload = payload[3:]
+
+	if fuHeader.S() && p.mightNeedDONL {
+		// sizeof(uint16)
+		if len(payload) <= 2 {
+			return nil, errShortPacket
+		}
+
+		donl := (uint16(payload[0]) << 8) | uint16(payload[1])
+		p.donl = &donl
+		payload = payload[2:]
+	}
+
+	p.payloadHeader = payloadHeader
+	p.fuHeader = fuHeader
+	p.payload = payload
+
+	return nil, nil
+}
+
+// PayloadHeader returns the NALU header of the packet.
+func (p *H265FragmentationUnitPacket) PayloadHeader() H265NALUHeader {
+	return p.payloadHeader
+}
+
+// FuHeader returns the Fragmentation Unit Header of the packet.
+func (p *H265FragmentationUnitPacket) FuHeader() H265FragmentationUnitHeader {
+	return p.fuHeader
+}
+
+// DONL returns the DONL of the packet.
+func (p *H265FragmentationUnitPacket) DONL() *uint16 {
+	return p.donl
+}
+
+// Payload returns the Fragmentation Unit packet payload.
+func (p *H265FragmentationUnitPacket) Payload() []byte {
+	return p.payload
+}
+
+func (p *H265FragmentationUnitPacket) isH265Packet() {}
+
+//
+// PACI implementation
+//
+
+// H265PACIPacket represents a single H265 PACI packet.
+//
+//  0                   1                   2                   3
+// 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+// |    PayloadHdr (Type=50)       |A|   cType   | PHSsize |F0..2|Y|
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+// |        Payload Header Extension Structure (PHES)              |
+// |=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=|
+// |                                                               |
+// |                  PACI payload: NAL unit                       |
+// |                   . . .                                       |
+// |                                                               |
+// |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+// |                               :...OPTIONAL RTP padding        |
+// +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+//
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.4.4
+type H265PACIPacket struct {
+	// payloadHeader is the header of the H265 packet.
+	payloadHeader H265NALUHeader
+
+	// Field which holds value for `A`, `cType`, `PHSsize`, `F0`, `F1`, `F2` and `Y` fields.
+	paciHeaderFields uint16
+
+	// phes is a header extension, of byte length `PHSsize`
+	phes []byte
+
+	// Payload contains NAL units & optional padding
+	payload []byte
+}
+
+// PayloadHeader returns the NAL Unit Header.
+func (p *H265PACIPacket) PayloadHeader() H265NALUHeader {
+	return p.payloadHeader
+}
+
+// A copies the F bit of the PACI payload NALU.
+func (p *H265PACIPacket) A() bool {
+	const mask = 0b10000000 << 8
+	return (p.paciHeaderFields & mask) != 0
+}
+
+// CType copies the Type field of the PACI payload NALU.
+func (p *H265PACIPacket) CType() uint8 {
+	const mask = 0b01111110 << 8
+	return uint8((p.paciHeaderFields & mask) >> (8 + 1))
+}
+
+// PHSsize indicates the size of the PHES field.
+func (p *H265PACIPacket) PHSsize() uint8 {
+	const mask = (0b00000001 << 8) | 0b11110000
+	return uint8((p.paciHeaderFields & mask) >> 4)
+}
+
+// F0 indicates the presence of a Temporal Scalability support extension in the PHES.
+func (p *H265PACIPacket) F0() bool {
+	const mask = 0b00001000
+	return (p.paciHeaderFields & mask) != 0
+}
+
+// F1 must be zero, reserved for future extensions.
+func (p *H265PACIPacket) F1() bool {
+	const mask = 0b00000100
+	return (p.paciHeaderFields & mask) != 0
+}
+
+// F2 must be zero, reserved for future extensions.
+func (p *H265PACIPacket) F2() bool {
+	const mask = 0b00000010
+	return (p.paciHeaderFields & mask) != 0
+}
+
+// Y must be zero, reserved for future extensions.
+func (p *H265PACIPacket) Y() bool {
+	const mask = 0b00000001
+	return (p.paciHeaderFields & mask) != 0
+}
+
+// PHES contains header extensions. Its size is indicated by PHSsize.
+func (p *H265PACIPacket) PHES() []byte {
+	return p.phes
+}
+
+// Payload is a single NALU or NALU-like struct, not including the first two octets (header).
+func (p *H265PACIPacket) Payload() []byte {
+	return p.payload
+}
+
+// TSCI returns the Temporal Scalability Control Information extension, if present.
+func (p *H265PACIPacket) TSCI() *H265TSCI {
+	if !p.F0() || p.PHSsize() < 3 {
+		return nil
+	}
+
+	tsci := H265TSCI((uint32(p.phes[0]) << 16) | (uint32(p.phes[1]) << 8) | uint32(p.phes[0]))
+	return &tsci
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H265PACIPacket this method is called upon.
+func (p *H265PACIPacket) Unmarshal(payload []byte) ([]byte, error) {
+	// sizeof(headers)
+	const totalHeaderSize = h265NaluHeaderSize + 2
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= totalHeaderSize {
+		return nil, fmt.Errorf("%w: %d <= %v", errShortPacket, len(payload), totalHeaderSize)
+	}
+
+	payloadHeader := newH265NALUHeader(payload[0], payload[1])
+	if payloadHeader.F() {
+		return nil, errH265CorruptedPacket
+	}
+	if !payloadHeader.IsPACIPacket() {
+		return nil, errInvalidH265PacketType
+	}
+
+	paciHeaderFields := (uint16(payload[2]) << 8) | uint16(payload[3])
+	payload = payload[4:]
+
+	p.paciHeaderFields = paciHeaderFields
+	headerExtensionSize := p.PHSsize()
+
+	if len(payload) < int(headerExtensionSize)+1 {
+		p.paciHeaderFields = 0
+		return nil, errShortPacket
+	}
+
+	p.payloadHeader = payloadHeader
+
+	if headerExtensionSize > 0 {
+		p.phes = payload[:headerExtensionSize]
+	}
+
+	payload = payload[headerExtensionSize:]
+	p.payload = payload
+
+	return nil, nil
+}
+
+func (p *H265PACIPacket) isH265Packet() {}
+
+//
+// Temporal Scalability Control Information
+//
+
+// H265TSCI is a Temporal Scalability Control Information header extension.
+// Reference: https://datatracker.ietf.org/doc/html/rfc7798#section-4.5
+type H265TSCI uint32
+
+// TL0PICIDX see RFC7798 for more details.
+func (h H265TSCI) TL0PICIDX() uint8 {
+	const m1 = 0xFFFF0000
+	const m2 = 0xFF00
+	return uint8((((h & m1) >> 16) & m2) >> 8)
+}
+
+// IrapPicID see RFC7798 for more details.
+func (h H265TSCI) IrapPicID() uint8 {
+	const m1 = 0xFFFF0000
+	const m2 = 0x00FF
+	return uint8(((h & m1) >> 16) & m2)
+}
+
+// S see RFC7798 for more details.
+func (h H265TSCI) S() bool {
+	const m1 = 0xFF00
+	const m2 = 0b10000000
+	return (uint8((h&m1)>>8) & m2) != 0
+}
+
+// E see RFC7798 for more details.
+func (h H265TSCI) E() bool {
+	const m1 = 0xFF00
+	const m2 = 0b01000000
+	return (uint8((h&m1)>>8) & m2) != 0
+}
+
+// RES see RFC7798 for more details.
+func (h H265TSCI) RES() uint8 {
+	const m1 = 0xFF00
+	const m2 = 0b00111111
+	return uint8((h&m1)>>8) & m2
+}
+
+//
+// H265 Packet interface
+//
+
+type isH265Packet interface {
+	isH265Packet()
+}
+
+var (
+	_ isH265Packet = (*H265FragmentationUnitPacket)(nil)
+	_ isH265Packet = (*H265PACIPacket)(nil)
+	_ isH265Packet = (*H265SingleNALUnitPacket)(nil)
+	_ isH265Packet = (*H265AggregationPacket)(nil)
+)
+
+//
+// Packet implementation
+//
+
+// H265Packet represents a H265 packet, stored in the payload of an RTP packet.
+type H265Packet struct {
+	packet        isH265Packet
+	mightNeedDONL bool
+}
+
+// WithDONL can be called to specify whether or not DONL might be parsed.
+// DONL may need to be parsed if `sprop-max-don-diff` is greater than 0 on the RTP stream.
+func (p *H265Packet) WithDONL(value bool) {
+	p.mightNeedDONL = value
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the H265Packet this method is called upon
+func (p *H265Packet) Unmarshal(payload []byte) ([]byte, error) {
+	if payload == nil {
+		return nil, errNilPacket
+	} else if len(payload) <= h265NaluHeaderSize {
+		return nil, fmt.Errorf("%w: %d <= %v", errShortPacket, len(payload), h265NaluHeaderSize)
+	}
+
+	payloadHeader := newH265NALUHeader(payload[0], payload[1])
+	if payloadHeader.F() {
+		return nil, errH265CorruptedPacket
+	}
+
+	switch {
+	case payloadHeader.IsPACIPacket():
+		decoded := &H265PACIPacket{}
+		if _, err := decoded.Unmarshal(payload); err != nil {
+			return nil, err
+		}
+
+		p.packet = decoded
+
+	case payloadHeader.IsFragmentationUnit():
+		decoded := &H265FragmentationUnitPacket{}
+		decoded.WithDONL(p.mightNeedDONL)
+
+		if _, err := decoded.Unmarshal(payload); err != nil {
+			return nil, err
+		}
+
+		p.packet = decoded
+
+	case payloadHeader.IsAggregationPacket():
+		decoded := &H265AggregationPacket{}
+		decoded.WithDONL(p.mightNeedDONL)
+
+		if _, err := decoded.Unmarshal(payload); err != nil {
+			return nil, err
+		}
+
+		p.packet = decoded
+
+	default:
+		decoded := &H265SingleNALUnitPacket{}
+		decoded.WithDONL(p.mightNeedDONL)
+
+		if _, err := decoded.Unmarshal(payload); err != nil {
+			return nil, err
+		}
+
+		p.packet = decoded
+	}
+
+	return nil, nil
+}
+
+// Packet returns the populated packet.
+// Must be casted to one of:
+// - *H265SingleNALUnitPacket
+// - *H265FragmentationUnitPacket
+// - *H265AggregationPacket
+// - *H265PACIPacket
+// nolint:golint
+func (p *H265Packet) Packet() isH265Packet {
+	return p.packet
+}

vendor/github.com/pion/rtp/codecs/opus_packet.go

@@ -0,0 +1,37 @@
+package codecs
+
+// OpusPayloader payloads Opus packets
+type OpusPayloader struct{}
+
+// Payload fragments an Opus packet across one or more byte arrays
+func (p *OpusPayloader) Payload(mtu uint16, payload []byte) [][]byte {
+	if payload == nil {
+		return [][]byte{}
+	}
+
+	out := make([]byte, len(payload))
+	copy(out, payload)
+	return [][]byte{out}
+}
+
+// OpusPacket represents the Opus header that is stored in the payload of an RTP Packet
+type OpusPacket struct {
+	Payload []byte
+
+	audioDepacketizer
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the OpusPacket this method is called upon
+func (p *OpusPacket) Unmarshal(packet []byte) ([]byte, error) {
+	if packet == nil {
+		return nil, errNilPacket
+	} else if len(packet) == 0 {
+		return nil, errShortPacket
+	}
+
+	p.Payload = packet
+	return packet, nil
+}
+
+// OpusPartitionHeadChecker is obsolete
+type OpusPartitionHeadChecker struct{}

vendor/github.com/pion/rtp/codecs/vp8_packet.go

@@ -0,0 +1,201 @@
+package codecs
+
+// VP8Payloader payloads VP8 packets
+type VP8Payloader struct {
+	EnablePictureID bool
+	pictureID       uint16
+}
+
+const (
+	vp8HeaderSize = 1
+)
+
+// Payload fragments a VP8 packet across one or more byte arrays
+func (p *VP8Payloader) Payload(mtu uint16, payload []byte) [][]byte {
+	/*
+	 * https://tools.ietf.org/html/rfc7741#section-4.2
+	 *
+	 *       0 1 2 3 4 5 6 7
+	 *      +-+-+-+-+-+-+-+-+
+	 *      |X|R|N|S|R| PID | (REQUIRED)
+	 *      +-+-+-+-+-+-+-+-+
+	 * X:   |I|L|T|K| RSV   | (OPTIONAL)
+	 *      +-+-+-+-+-+-+-+-+
+	 * I:   |M| PictureID   | (OPTIONAL)
+	 *      +-+-+-+-+-+-+-+-+
+	 * L:   |   TL0PICIDX   | (OPTIONAL)
+	 *      +-+-+-+-+-+-+-+-+
+	 * T/K: |TID|Y| KEYIDX  | (OPTIONAL)
+	 *      +-+-+-+-+-+-+-+-+
+	 *  S: Start of VP8 partition.  SHOULD be set to 1 when the first payload
+	 *     octet of the RTP packet is the beginning of a new VP8 partition,
+	 *     and MUST NOT be 1 otherwise.  The S bit MUST be set to 1 for the
+	 *     first packet of each encoded frame.
+	 */
+
+	usingHeaderSize := vp8HeaderSize
+	if p.EnablePictureID {
+		switch {
+		case p.pictureID == 0:
+		case p.pictureID < 128:
+			usingHeaderSize = vp8HeaderSize + 2
+		default:
+			usingHeaderSize = vp8HeaderSize + 3
+		}
+	}
+
+	maxFragmentSize := int(mtu) - usingHeaderSize
+
+	payloadData := payload
+	payloadDataRemaining := len(payload)
+
+	payloadDataIndex := 0
+	var payloads [][]byte
+
+	// Make sure the fragment/payload size is correct
+	if min(maxFragmentSize, payloadDataRemaining) <= 0 {
+		return payloads
+	}
+	first := true
+	for payloadDataRemaining > 0 {
+		currentFragmentSize := min(maxFragmentSize, payloadDataRemaining)
+		out := make([]byte, usingHeaderSize+currentFragmentSize)
+
+		if first {
+			out[0] = 0x10
+			first = false
+		}
+		if p.EnablePictureID {
+			switch usingHeaderSize {
+			case vp8HeaderSize:
+			case vp8HeaderSize + 2:
+				out[0] |= 0x80
+				out[1] |= 0x80
+				out[2] |= uint8(p.pictureID & 0x7F)
+			case vp8HeaderSize + 3:
+				out[0] |= 0x80
+				out[1] |= 0x80
+				out[2] |= 0x80 | uint8((p.pictureID>>8)&0x7F)
+				out[3] |= uint8(p.pictureID & 0xFF)
+			}
+		}
+
+		copy(out[usingHeaderSize:], payloadData[payloadDataIndex:payloadDataIndex+currentFragmentSize])
+		payloads = append(payloads, out)
+
+		payloadDataRemaining -= currentFragmentSize
+		payloadDataIndex += currentFragmentSize
+	}
+
+	p.pictureID++
+	p.pictureID &= 0x7FFF
+
+	return payloads
+}
+
+// VP8Packet represents the VP8 header that is stored in the payload of an RTP Packet
+type VP8Packet struct {
+	// Required Header
+	X   uint8 /* extended control bits present */
+	N   uint8 /* when set to 1 this frame can be discarded */
+	S   uint8 /* start of VP8 partition */
+	PID uint8 /* partition index */
+
+	// Extended control bits
+	I uint8 /* 1 if PictureID is present */
+	L uint8 /* 1 if TL0PICIDX is present */
+	T uint8 /* 1 if TID is present */
+	K uint8 /* 1 if KEYIDX is present */
+
+	// Optional extension
+	PictureID uint16 /* 8 or 16 bits, picture ID */
+	TL0PICIDX uint8  /* 8 bits temporal level zero index */
+	TID       uint8  /* 2 bits temporal layer index */
+	Y         uint8  /* 1 bit layer sync bit */
+	KEYIDX    uint8  /* 5 bits temporal key frame index */
+
+	Payload []byte
+
+	videoDepacketizer
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the VP8Packet this method is called upon
+func (p *VP8Packet) Unmarshal(payload []byte) ([]byte, error) {
+	if payload == nil {
+		return nil, errNilPacket
+	}
+
+	payloadLen := len(payload)
+
+	if payloadLen < 4 {
+		return nil, errShortPacket
+	}
+
+	payloadIndex := 0
+
+	p.X = (payload[payloadIndex] & 0x80) >> 7
+	p.N = (payload[payloadIndex] & 0x20) >> 5
+	p.S = (payload[payloadIndex] & 0x10) >> 4
+	p.PID = payload[payloadIndex] & 0x07
+
+	payloadIndex++
+
+	if p.X == 1 {
+		p.I = (payload[payloadIndex] & 0x80) >> 7
+		p.L = (payload[payloadIndex] & 0x40) >> 6
+		p.T = (payload[payloadIndex] & 0x20) >> 5
+		p.K = (payload[payloadIndex] & 0x10) >> 4
+		payloadIndex++
+	}
+
+	if p.I == 1 { // PID present?
+		if payload[payloadIndex]&0x80 > 0 { // M == 1, PID is 16bit
+			p.PictureID = (uint16(payload[payloadIndex]&0x7F) << 8) | uint16(payload[payloadIndex+1])
+			payloadIndex += 2
+		} else {
+			p.PictureID = uint16(payload[payloadIndex])
+			payloadIndex++
+		}
+	}
+
+	if payloadIndex >= payloadLen {
+		return nil, errShortPacket
+	}
+
+	if p.L == 1 {
+		p.TL0PICIDX = payload[payloadIndex]
+		payloadIndex++
+	}
+
+	if payloadIndex >= payloadLen {
+		return nil, errShortPacket
+	}
+
+	if p.T == 1 || p.K == 1 {
+		if p.T == 1 {
+			p.TID = payload[payloadIndex] >> 6
+			p.Y = (payload[payloadIndex] >> 5) & 0x1
+		}
+		if p.K == 1 {
+			p.KEYIDX = payload[payloadIndex] & 0x1F
+		}
+		payloadIndex++
+	}
+
+	if payloadIndex >= payloadLen {
+		return nil, errShortPacket
+	}
+	p.Payload = payload[payloadIndex:]
+	return p.Payload, nil
+}
+
+// VP8PartitionHeadChecker is obsolete
+type VP8PartitionHeadChecker struct{}
+
+// IsPartitionHead checks whether if this is a head of the VP8 partition
+func (*VP8Packet) IsPartitionHead(payload []byte) bool {
+	if len(payload) < 1 {
+		return false
+	}
+	return (payload[0] & 0x10) != 0
+}

vendor/github.com/pion/rtp/codecs/vp9_packet.go

@@ -0,0 +1,388 @@
+package codecs
+
+import (
+	"github.com/pion/randutil"
+)
+
+// Use global random generator to properly seed by crypto grade random.
+var globalMathRandomGenerator = randutil.NewMathRandomGenerator() // nolint:gochecknoglobals
+
+// VP9Payloader payloads VP9 packets
+type VP9Payloader struct {
+	pictureID   uint16
+	initialized bool
+
+	// InitialPictureIDFn is a function that returns random initial picture ID.
+	InitialPictureIDFn func() uint16
+}
+
+const (
+	vp9HeaderSize    = 3 // Flexible mode 15 bit picture ID
+	maxSpatialLayers = 5
+	maxVP9RefPics    = 3
+)
+
+// Payload fragments an VP9 packet across one or more byte arrays
+func (p *VP9Payloader) Payload(mtu uint16, payload []byte) [][]byte {
+	/*
+	 * https://www.ietf.org/id/draft-ietf-payload-vp9-13.txt
+	 *
+	 * Flexible mode (F=1)
+	 *        0 1 2 3 4 5 6 7
+	 *       +-+-+-+-+-+-+-+-+
+	 *       |I|P|L|F|B|E|V|Z| (REQUIRED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  I:   |M| PICTURE ID  | (REQUIRED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  M:   | EXTENDED PID  | (RECOMMENDED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  L:   | TID |U| SID |D| (CONDITIONALLY RECOMMENDED)
+	 *       +-+-+-+-+-+-+-+-+                             -\
+	 *  P,F: | P_DIFF      |N| (CONDITIONALLY REQUIRED)    - up to 3 times
+	 *       +-+-+-+-+-+-+-+-+                             -/
+	 *  V:   | SS            |
+	 *       | ..            |
+	 *       +-+-+-+-+-+-+-+-+
+	 *
+	 * Non-flexible mode (F=0)
+	 *        0 1 2 3 4 5 6 7
+	 *       +-+-+-+-+-+-+-+-+
+	 *       |I|P|L|F|B|E|V|Z| (REQUIRED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  I:   |M| PICTURE ID  | (RECOMMENDED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  M:   | EXTENDED PID  | (RECOMMENDED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  L:   | TID |U| SID |D| (CONDITIONALLY RECOMMENDED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *       |   TL0PICIDX   | (CONDITIONALLY REQUIRED)
+	 *       +-+-+-+-+-+-+-+-+
+	 *  V:   | SS            |
+	 *       | ..            |
+	 *       +-+-+-+-+-+-+-+-+
+	 */
+
+	if !p.initialized {
+		if p.InitialPictureIDFn == nil {
+			p.InitialPictureIDFn = func() uint16 {
+				return uint16(globalMathRandomGenerator.Intn(0x7FFF))
+			}
+		}
+		p.pictureID = p.InitialPictureIDFn() & 0x7FFF
+		p.initialized = true
+	}
+	if payload == nil {
+		return [][]byte{}
+	}
+
+	maxFragmentSize := int(mtu) - vp9HeaderSize
+	payloadDataRemaining := len(payload)
+	payloadDataIndex := 0
+
+	if min(maxFragmentSize, payloadDataRemaining) <= 0 {
+		return [][]byte{}
+	}
+
+	var payloads [][]byte
+	for payloadDataRemaining > 0 {
+		currentFragmentSize := min(maxFragmentSize, payloadDataRemaining)
+		out := make([]byte, vp9HeaderSize+currentFragmentSize)
+
+		out[0] = 0x90 // F=1 I=1
+		if payloadDataIndex == 0 {
+			out[0] |= 0x08 // B=1
+		}
+		if payloadDataRemaining == currentFragmentSize {
+			out[0] |= 0x04 // E=1
+		}
+		out[1] = byte(p.pictureID>>8) | 0x80
+		out[2] = byte(p.pictureID)
+		copy(out[vp9HeaderSize:], payload[payloadDataIndex:payloadDataIndex+currentFragmentSize])
+		payloads = append(payloads, out)
+
+		payloadDataRemaining -= currentFragmentSize
+		payloadDataIndex += currentFragmentSize
+	}
+	p.pictureID++
+	if p.pictureID >= 0x8000 {
+		p.pictureID = 0
+	}
+
+	return payloads
+}
+
+// VP9Packet represents the VP9 header that is stored in the payload of an RTP Packet
+type VP9Packet struct {
+	// Required header
+	I bool // PictureID is present
+	P bool // Inter-picture predicted frame
+	L bool // Layer indices is present
+	F bool // Flexible mode
+	B bool // Start of a frame
+	E bool // End of a frame
+	V bool // Scalability structure (SS) data present
+	Z bool // Not a reference frame for upper spatial layers
+
+	// Recommended headers
+	PictureID uint16 // 7 or 16 bits, picture ID
+
+	// Conditionally recommended headers
+	TID uint8 // Temporal layer ID
+	U   bool  // Switching up point
+	SID uint8 // Spatial layer ID
+	D   bool  // Inter-layer dependency used
+
+	// Conditionally required headers
+	PDiff     []uint8 // Reference index (F=1)
+	TL0PICIDX uint8   // Temporal layer zero index (F=0)
+
+	// Scalability structure headers
+	NS      uint8 // N_S + 1 indicates the number of spatial layers present in the VP9 stream
+	Y       bool  // Each spatial layer's frame resolution present
+	G       bool  // PG description present flag.
+	NG      uint8 // N_G indicates the number of pictures in a Picture Group (PG)
+	Width   []uint16
+	Height  []uint16
+	PGTID   []uint8   // Temporal layer ID of pictures in a Picture Group
+	PGU     []bool    // Switching up point of pictures in a Picture Group
+	PGPDiff [][]uint8 // Reference indecies of pictures in a Picture Group
+
+	Payload []byte
+
+	videoDepacketizer
+}
+
+// Unmarshal parses the passed byte slice and stores the result in the VP9Packet this method is called upon
+func (p *VP9Packet) Unmarshal(packet []byte) ([]byte, error) {
+	if packet == nil {
+		return nil, errNilPacket
+	}
+	if len(packet) < 1 {
+		return nil, errShortPacket
+	}
+
+	p.I = packet[0]&0x80 != 0
+	p.P = packet[0]&0x40 != 0
+	p.L = packet[0]&0x20 != 0
+	p.F = packet[0]&0x10 != 0
+	p.B = packet[0]&0x08 != 0
+	p.E = packet[0]&0x04 != 0
+	p.V = packet[0]&0x02 != 0
+	p.Z = packet[0]&0x01 != 0
+
+	pos := 1
+	var err error
+
+	if p.I {
+		pos, err = p.parsePictureID(packet, pos)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	if p.L {
+		pos, err = p.parseLayerInfo(packet, pos)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	if p.F && p.P {
+		pos, err = p.parseRefIndices(packet, pos)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	if p.V {
+		pos, err = p.parseSSData(packet, pos)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	p.Payload = packet[pos:]
+	return p.Payload, nil
+}
+
+// Picture ID:
+//
+//      +-+-+-+-+-+-+-+-+
+// I:   |M| PICTURE ID  |   M:0 => picture id is 7 bits.
+//      +-+-+-+-+-+-+-+-+   M:1 => picture id is 15 bits.
+// M:   | EXTENDED PID  |
+//      +-+-+-+-+-+-+-+-+
+//
+func (p *VP9Packet) parsePictureID(packet []byte, pos int) (int, error) {
+	if len(packet) <= pos {
+		return pos, errShortPacket
+	}
+
+	p.PictureID = uint16(packet[pos] & 0x7F)
+	if packet[pos]&0x80 != 0 {
+		pos++
+		if len(packet) <= pos {
+			return pos, errShortPacket
+		}
+		p.PictureID = p.PictureID<<8 | uint16(packet[pos])
+	}
+	pos++
+	return pos, nil
+}
+
+func (p *VP9Packet) parseLayerInfo(packet []byte, pos int) (int, error) {
+	pos, err := p.parseLayerInfoCommon(packet, pos)
+	if err != nil {
+		return pos, err
+	}
+
+	if p.F {
+		return pos, nil
+	}
+
+	return p.parseLayerInfoNonFlexibleMode(packet, pos)
+}
+
+// Layer indices (flexible mode):
+//
+//      +-+-+-+-+-+-+-+-+
+// L:   |  T  |U|  S  |D|
+//      +-+-+-+-+-+-+-+-+
+//
+func (p *VP9Packet) parseLayerInfoCommon(packet []byte, pos int) (int, error) {
+	if len(packet) <= pos {
+		return pos, errShortPacket
+	}
+
+	p.TID = packet[pos] >> 5
+	p.U = packet[pos]&0x10 != 0
+	p.SID = (packet[pos] >> 1) & 0x7
+	p.D = packet[pos]&0x01 != 0
+
+	if p.SID >= maxSpatialLayers {
+		return pos, errTooManySpatialLayers
+	}
+
+	pos++
+	return pos, nil
+}
+
+// Layer indices (non-flexible mode):
+//
+//      +-+-+-+-+-+-+-+-+
+// L:   |  T  |U|  S  |D|
+//      +-+-+-+-+-+-+-+-+
+//      |   TL0PICIDX   |
+//      +-+-+-+-+-+-+-+-+
+//
+func (p *VP9Packet) parseLayerInfoNonFlexibleMode(packet []byte, pos int) (int, error) {
+	if len(packet) <= pos {
+		return pos, errShortPacket
+	}
+
+	p.TL0PICIDX = packet[pos]
+	pos++
+	return pos, nil
+}
+
+// Reference indices:
+//
+//      +-+-+-+-+-+-+-+-+                P=1,F=1: At least one reference index
+// P,F: | P_DIFF      |N|  up to 3 times          has to be specified.
+//      +-+-+-+-+-+-+-+-+                    N=1: An additional P_DIFF follows
+//                                                current P_DIFF.
+//
+func (p *VP9Packet) parseRefIndices(packet []byte, pos int) (int, error) {
+	for {
+		if len(packet) <= pos {
+			return pos, errShortPacket
+		}
+		p.PDiff = append(p.PDiff, packet[pos]>>1)
+		if packet[pos]&0x01 == 0 {
+			break
+		}
+		if len(p.PDiff) >= maxVP9RefPics {
+			return pos, errTooManyPDiff
+		}
+		pos++
+	}
+	pos++
+
+	return pos, nil
+}
+
+// Scalability structure (SS):
+//
+//      +-+-+-+-+-+-+-+-+
+// V:   | N_S |Y|G|-|-|-|
+//      +-+-+-+-+-+-+-+-+              -|
+// Y:   |     WIDTH     | (OPTIONAL)    .
+//      +               +               .
+//      |               | (OPTIONAL)    .
+//      +-+-+-+-+-+-+-+-+               . N_S + 1 times
+//      |     HEIGHT    | (OPTIONAL)    .
+//      +               +               .
+//      |               | (OPTIONAL)    .
+//      +-+-+-+-+-+-+-+-+              -|
+// G:   |      N_G      | (OPTIONAL)
+//      +-+-+-+-+-+-+-+-+                           -|
+// N_G: |  T  |U| R |-|-| (OPTIONAL)                 .
+//      +-+-+-+-+-+-+-+-+              -|            . N_G times
+//      |    P_DIFF     | (OPTIONAL)    . R times    .
+//      +-+-+-+-+-+-+-+-+              -|           -|
+//
+func (p *VP9Packet) parseSSData(packet []byte, pos int) (int, error) {
+	if len(packet) <= pos {
+		return pos, errShortPacket
+	}
+
+	p.NS = packet[pos] >> 5
+	p.Y = packet[pos]&0x10 != 0
+	p.G = (packet[pos]>>1)&0x7 != 0
+	pos++
+
+	NS := p.NS + 1
+	p.NG = 0
+
+	if p.Y {
+		p.Width = make([]uint16, NS)
+		p.Height = make([]uint16, NS)
+		for i := 0; i < int(NS); i++ {
+			p.Width[i] = uint16(packet[pos])<<8 | uint16(packet[pos+1])
+			pos += 2
+			p.Height[i] = uint16(packet[pos])<<8 | uint16(packet[pos+1])
+			pos += 2
+		}
+	}
+
+	if p.G {
+		p.NG = packet[pos]
+		pos++
+	}
+
+	for i := 0; i < int(p.NG); i++ {
+		p.PGTID = append(p.PGTID, packet[pos]>>5)
+		p.PGU = append(p.PGU, packet[pos]&0x10 != 0)
+		R := (packet[pos] >> 2) & 0x3
+		pos++
+
+		p.PGPDiff = append(p.PGPDiff, []uint8{})
+		for j := 0; j < int(R); j++ {
+			p.PGPDiff[i] = append(p.PGPDiff[i], packet[pos])
+			pos++
+		}
+	}
+
+	return pos, nil
+}
+
+// VP9PartitionHeadChecker is obsolete
+type VP9PartitionHeadChecker struct{}
+
+// IsPartitionHead checks whether if this is a head of the VP9 partition
+func (*VP9Packet) IsPartitionHead(payload []byte) bool {
+	if len(payload) < 1 {
+		return false
+	}
+	return (payload[0] & 0x08) != 0
+}