Skip to content

Feature/fix h265 rtp #2112

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
pedroSG94 merged 4 commits into from
Jun 19, 2026
Merged

Feature/fix h265 rtp #2112

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
5348b74
fixing h265 rtp
pedroSG94 Jun 18, 2026
373b1af
remove unnecessary h265 rtp checks
pedroSG94 Jun 18, 2026
234d7cb
remove h264 stapA
pedroSG94 Jun 18, 2026
c123248
fix extract nals
pedroSG94 Jun 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion common/src/main/java/com/pedro/common/nal/NalReader.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,8 @@ object NalReader {
val duplicate = buffer.duplicate()
duplicate.position(payloadStart - offset)
duplicate.limit(limit - offset)
units.add(duplicate.slice())
val nal = duplicate.slice()
if (shouldKeepNal(nal, codec, shouldDiscardVideoInfo)) units.add(nal)
}
if (units.isEmpty()) units.add(buffer)
return units
Expand Down
2 changes: 1 addition & 1 deletion rtmp/src/main/java/com/pedro/rtmp/flv/video/packet/H265Packet.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,7 +99,7 @@ class H265Packet: BasePacket() {
val size = nals.sumOf { it.capacity() }
buffer = ByteArray(header.size + size + naluSize * nals.size)

val type: Int = nals[0].get(0).toInt().shr(1 and 0x3f)
val type: Int = nals[0].get(0).toInt().shr(1) and 0x3F
var nalType = VideoDataType.INTER_FRAME.value
if (type == VideoNalType.IDR_N_LP.value || type == VideoNalType.IDR_W_DLP.value || mediaFrame.info.isKeyFrame) {
nalType = VideoDataType.KEYFRAME.value
Expand Down
138 changes: 39 additions & 99 deletions rtsp/src/main/java/com/pedro/rtsp/rtp/packets/H264Packet.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,17 +16,14 @@

package com.pedro.rtsp.rtp.packets

import android.util.Log
import com.pedro.common.VideoCodec
import com.pedro.common.frame.MediaFrame
import com.pedro.common.nal.NalReader
import com.pedro.common.removeInfo
import com.pedro.common.toByteArray
import com.pedro.rtsp.rtsp.RtpFrame
import com.pedro.rtsp.utils.RtpConstants
import com.pedro.rtsp.utils.getVideoStartCodeSize
import java.nio.ByteBuffer
import kotlin.experimental.and
import kotlin.experimental.or

/**
* Created by pedro on 27/11/18.
Expand All @@ -37,29 +34,15 @@ class H264Packet: BasePacket(RtpConstants.clockVideoFrequency,
RtpConstants.payloadType + RtpConstants.trackVideo
) {

private var stapA: ByteArray? = null
private var sendKeyFrame = false
private var videoInfo: Set<ByteBuffer>? = null
private val header = ByteArray(2)

init {
channelIdentifier = RtpConstants.trackVideo
}

fun sendVideoInfo(sps: ByteBuffer, pps: ByteBuffer) {
videoInfo = setOf(sps, pps)
setSpsPps(sps.toByteArray(), pps.toByteArray())
}

override suspend fun createAndSendPacket(
mediaFrame: MediaFrame,
callback: suspend (List<RtpFrame>) -> Unit
) {
val videoInfo = this.videoInfo
if (videoInfo == null) {
Log.e(TAG, "waiting for a valid sps and pps")
return
}
val fixedBuffer = mediaFrame.data.removeInfo(mediaFrame.info)
// We read a NAL units from ByteBuffer and we send them
// NAL units are preceded with 0x00000001
Expand All @@ -69,98 +52,55 @@ class H264Packet: BasePacket(RtpConstants.clockVideoFrequency,

val ts = mediaFrame.info.timestamp * 1000L
val frames = mutableListOf<RtpFrame>()
val nalType = nals[0].get(0)
val type: Int = (nalType and 0x1F).toInt()
if (type == RtpConstants.IDR || mediaFrame.info.isKeyFrame) {
stapA?.let {
val buffer = getBuffer(it.size + RtpConstants.RTP_HEADER_LENGTH)
nals.forEachIndexed { index, data ->
val nalType = data.get()
val nalSize = data.remaining()
// Small NAL unit => Single NAL unit
if (nalSize <= maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 1) {
val buffer = getBuffer(nalSize + RtpConstants.RTP_HEADER_LENGTH + 1)
buffer[RtpConstants.RTP_HEADER_LENGTH] = nalType
data.get(buffer, RtpConstants.RTP_HEADER_LENGTH + 1, nalSize)
val rtpTs = updateTimeStamp(buffer, ts)
markPacket(buffer) //mark end frame
System.arraycopy(it, 0, buffer, RtpConstants.RTP_HEADER_LENGTH, it.size)
if (index == nals.size - 1) markPacket(buffer) //mark end frame
updateSeq(buffer)
val rtpFrame = RtpFrame(buffer, rtpTs, it.size + RtpConstants.RTP_HEADER_LENGTH, channelIdentifier)
val rtpFrame = RtpFrame(buffer, rtpTs, buffer.size, channelIdentifier)
frames.add(rtpFrame)
sendKeyFrame = true
} ?: run {
Log.i(TAG, "can't create key frame because setSpsPps was not called")
}
}
if (sendKeyFrame) {
nals.forEachIndexed { index, data ->
val nalType = data.get()
val nalSize = data.remaining()
// Small NAL unit => Single NAL unit
if (nalSize <= maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 1) {
val buffer = getBuffer(nalSize + RtpConstants.RTP_HEADER_LENGTH + 1)
buffer[RtpConstants.RTP_HEADER_LENGTH] = nalType
data.get(buffer, RtpConstants.RTP_HEADER_LENGTH + 1, nalSize)
val rtpTs = updateTimeStamp(buffer, ts)
if (index == nals.size - 1) markPacket(buffer) //mark end frame
updateSeq(buffer)
val rtpFrame = RtpFrame(buffer, rtpTs, buffer.size, channelIdentifier)
frames.add(rtpFrame)
} else {
// Set FU-A header
header[1] = nalType and 0x1F // FU header type
header[1] = header[1].plus(0x80).toByte() // set start bit to 1
// Set FU-A indicator
header[0] = nalType and 0x60 and 0xFF.toByte() // FU indicator NRI
header[0] = header[0].plus(28).toByte()
} else {
// Set FU-A header
val fuHeader = nalType and 0x1F // FU header type
// Set FU-A indicator
val fuIndicator = (nalType and 0x60).plus(28).toByte() // FU indicator NRI

var sum = 0
while (sum < nalSize) {
val length = if (nalSize - sum > maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 2) {
maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 2
} else {
data.remaining()
}
val buffer = getBuffer(length + RtpConstants.RTP_HEADER_LENGTH + 2)
buffer[RtpConstants.RTP_HEADER_LENGTH] = header[0]
// Switch start bit
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = if (sum > 0) header[1] and 0x7F else header[1]
val rtpTs = updateTimeStamp(buffer, ts)
data.get(buffer, RtpConstants.RTP_HEADER_LENGTH + 2, length)
sum += length
// Last packet before next NAL
if (sum >= nalSize) {
// End bit on
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = buffer[RtpConstants.RTP_HEADER_LENGTH + 1].plus(0x40).toByte()
if (index == nals.size - 1) markPacket(buffer) //mark end frame
}
updateSeq(buffer)
val rtpFrame = RtpFrame(buffer, rtpTs, buffer.size, channelIdentifier)
frames.add(rtpFrame)
}
var sum = 0
while (sum < nalSize) {
val length = if (nalSize - sum > maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 2) {
maxPacketSize - RtpConstants.RTP_HEADER_LENGTH - 2
} else {
data.remaining()
}
val buffer = getBuffer(length + RtpConstants.RTP_HEADER_LENGTH + 2)
buffer[RtpConstants.RTP_HEADER_LENGTH] = fuIndicator
// Switch start bit
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = if (sum > 0) fuHeader else (fuHeader or 0x80.toByte())
val rtpTs = updateTimeStamp(buffer, ts)
data.get(buffer, RtpConstants.RTP_HEADER_LENGTH + 2, length)
sum += length
// Last packet before next NAL
if (sum >= nalSize) {
// End bit on
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = buffer[RtpConstants.RTP_HEADER_LENGTH + 1].plus(0x40).toByte()
if (index == nals.size - 1) markPacket(buffer) //mark end frame
}
updateSeq(buffer)
val rtpFrame = RtpFrame(buffer, rtpTs, buffer.size, channelIdentifier)
frames.add(rtpFrame)
}
}
} else {
Log.i(TAG, "waiting for keyframe")
}
if (frames.isNotEmpty()) callback(frames)
}

private fun setSpsPps(sps: ByteArray, pps: ByteArray) {
stapA = ByteArray(sps.size + pps.size + 5)
stapA?.let {
// STAP-A NAL header is 24
it[0] = 24

// Write NALU 1 size into the array (NALU 1 is the SPS).
it[1] = (sps.size shr 8).toByte()
it[2] = (sps.size and 0xFF).toByte()

// Write NALU 2 size into the array (NALU 2 is the PPS).
it[sps.size + 3] = (pps.size shr 8).toByte()
it[sps.size + 4] = (pps.size and 0xFF).toByte()

// Write NALU 1 into the array, then write NALU 2 into the array.
System.arraycopy(sps, 0, it, 3, sps.size)
System.arraycopy(pps, 0, it, 5 + sps.size, pps.size)
}
}

override fun reset() {
super.reset()
sendKeyFrame = false
}
}
33 changes: 8 additions & 25 deletions rtsp/src/main/java/com/pedro/rtsp/rtp/packets/H265Packet.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,15 +16,13 @@

package com.pedro.rtsp.rtp.packets

import android.util.Log
import com.pedro.common.VideoCodec
import com.pedro.common.frame.MediaFrame
import com.pedro.common.nal.NalReader
import com.pedro.common.removeInfo
import com.pedro.rtsp.rtsp.RtpFrame
import com.pedro.rtsp.utils.RtpConstants
import java.nio.ByteBuffer
import kotlin.experimental.and
import kotlin.experimental.or

/**
* Created by pedro on 28/11/18.
Expand All @@ -36,30 +34,18 @@ class H265Packet : BasePacket(
RtpConstants.payloadType + RtpConstants.trackVideo
) {

private var videoInfo: Set<ByteBuffer>? = null
private val header = ByteArray(3)

init {
channelIdentifier = RtpConstants.trackVideo
}

fun sendVideoInfo(sps: ByteBuffer, pps: ByteBuffer, vps: ByteBuffer) {
videoInfo = setOf(sps, pps, vps)
}

override suspend fun createAndSendPacket(
mediaFrame: MediaFrame,
callback: suspend (List<RtpFrame>) -> Unit
) {
val videoInfo = this.videoInfo
if (videoInfo == null) {
Log.e(TAG, "waiting for a valid sps, pps and vps")
return
}
val fixedBuffer = mediaFrame.data.removeInfo(mediaFrame.info)
// We read a NAL units from ByteBuffer and we send them
// NAL units are preceded with 0x00000001
val nals = NalReader.extractNals(fixedBuffer, VideoCodec.H265, true)
val nals = NalReader.extractNals(fixedBuffer, VideoCodec.H265, false)
if (nals.isEmpty()) return

val ts = mediaFrame.info.timestamp * 1000L
Expand All @@ -82,18 +68,14 @@ class H265Packet : BasePacket(
val rtpFrame = RtpFrame(buffer, rtpTs, buffer.size, channelIdentifier)
frames.add(rtpFrame)
} else {
val type: Int = nalType.toInt().shr(1 and 0x3f)
//Set PayloadHdr (16bit type=49)
header[0] = (49 shl 1).toByte()
header[1] = 1
// Set FU header
// +---------------+
// |0|1|2|3|4|5|6|7|
// +-+-+-+-+-+-+-+-+
// |S|E| FuType |
// +---------------+
header[2] = type.toByte() // FU header type
header[2] = header[2].plus(0x80).toByte() // Start bit
val type: Int = nalType.toInt().shr(1) and 0x3F
val fuHeader = type.toByte() // FU header type

var sum = 0
while (sum < nalSize) {
Expand All @@ -103,10 +85,11 @@ class H265Packet : BasePacket(
data.remaining()
}
val buffer = getBuffer(length + RtpConstants.RTP_HEADER_LENGTH + 3)
buffer[RtpConstants.RTP_HEADER_LENGTH] = header[0]
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = header[1]
//Set PayloadHdr (16bit type=49)
buffer[RtpConstants.RTP_HEADER_LENGTH] = (49 shl 1).toByte()
buffer[RtpConstants.RTP_HEADER_LENGTH + 1] = 1
// Switch start bit
buffer[RtpConstants.RTP_HEADER_LENGTH + 2] = if (sum > 0) header[2] and 0x7F else header[2]
buffer[RtpConstants.RTP_HEADER_LENGTH + 2] = if (sum > 0) fuHeader else (fuHeader or 0x80.toByte())
val rtpTs = updateTimeStamp(buffer, ts)
data.get(buffer, RtpConstants.RTP_HEADER_LENGTH + 3, length)
sum += length
Expand Down
4 changes: 2 additions & 2 deletions rtsp/src/main/java/com/pedro/rtsp/rtsp/RtspSender.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,11 +78,11 @@ class RtspSender(
videoPacket = when (commandsManager.videoCodec) {
VideoCodec.H264 -> {
if (pps == null) throw IllegalArgumentException("pps can't be null with h264")
H264Packet().apply { sendVideoInfo(sps, pps) }
H264Packet()
}
VideoCodec.H265 -> {
if (vps == null || pps == null) throw IllegalArgumentException("pps or vps can't be null with h265")
H265Packet().apply { sendVideoInfo(sps, pps, vps) }
H265Packet()
}
VideoCodec.AV1 -> Av1Packet()
}
Expand Down
30 changes: 10 additions & 20 deletions rtsp/src/test/java/com/pedro/rtsp/rtp/H264PacketTest.kt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,25 +39,20 @@ class H264PacketTest {
val fakeH264 = header.plus(ByteArray(300) { 0x00 })

val info = MediaFrame.Info(0, fakeH264.size, timestamp, true)
val fakeSps = byteArrayOf(0x00, 0x00, 0x00, 0x01, 0x02, 0x03, 0x04)
val fakePps = byteArrayOf(0x00, 0x00, 0x00, 0x01, 0x0A, 0x0B, 0x0C)
val mediaFrame = MediaFrame(ByteBuffer.wrap(fakeH264), info, MediaFrame.Type.VIDEO)
val h264Packet = H264Packet().apply { sendVideoInfo(ByteBuffer.wrap(fakeSps), ByteBuffer.wrap(fakePps)) }
val h264Packet = H264Packet()
h264Packet.setSSRC(123456789)
val frames = mutableListOf<RtpFrame>()
h264Packet.createAndSendPacket(mediaFrame) { frames.addAll(it) }

val expectedRtp = byteArrayOf(-128, -32, 0, 2, 0, -87, -118, -57, 7, 91, -51, 21, 5).plus(fakeH264.copyOfRange(header.size, fakeH264.size))
val expectedStapA = byteArrayOf(-128, -32, 0, 1, 0, -87, -118, -57, 7, 91, -51, 21, 24, 0, 7, 0, 0, 0, 1, 2, 3, 4, 0, 7, 0, 0, 0, 1, 10, 11, 12)
val expectedRtp = byteArrayOf(-128, -32, 0, 1, 0, -87, -118, -57, 7, 91, -51, 21, 5).plus(fakeH264.copyOfRange(header.size, fakeH264.size))
val expectedTimeStamp = 11111111L
val expectedSize = RtpConstants.RTP_HEADER_LENGTH + 1 + info.size - header.size
val expectedStapAResult = RtpFrame(expectedStapA, expectedTimeStamp, fakePps.size + fakePps.size + 5 + RtpConstants.RTP_HEADER_LENGTH, RtpConstants.trackVideo)
val expectedPacketResult = RtpFrame(expectedRtp, expectedTimeStamp, expectedSize, RtpConstants.trackVideo)

assertNotEquals(0, frames.size)
assertTrue(frames.size == 2)
assertEquals(expectedStapAResult, frames[0])
assertEquals(expectedPacketResult, frames[1])
assertTrue(frames.size == 1)
assertEquals(expectedPacketResult, frames[0])
}

@Test
Expand All @@ -67,33 +62,28 @@ class H264PacketTest {
val fakeH264 = header.plus(ByteArray(2500) { 0x00 })

val info = MediaFrame.Info(0, fakeH264.size, timestamp, true)
val fakeSps = byteArrayOf(0x00, 0x00, 0x00, 0x01, 0x02, 0x03, 0x04)
val fakePps = byteArrayOf(0x00, 0x00, 0x00, 0x01, 0x0A, 0x0B, 0x0C)
val mediaFrame = MediaFrame(ByteBuffer.wrap(fakeH264), info, MediaFrame.Type.VIDEO)
val h264Packet = H264Packet().apply { sendVideoInfo(ByteBuffer.wrap(fakeSps), ByteBuffer.wrap(fakePps)) }
val h264Packet = H264Packet()
h264Packet.setSSRC(123456789)
val frames = mutableListOf<RtpFrame>()
h264Packet.createAndSendPacket(mediaFrame) { frames.addAll(it) }

val packet1Size = RtpConstants.MTU - 28 - RtpConstants.RTP_HEADER_LENGTH - 2
val chunk1 = fakeH264.copyOfRange(header.size, header.size + packet1Size)
val chunk2 = fakeH264.copyOfRange(header.size + packet1Size, fakeH264.size)
val expectedRtp = byteArrayOf(-128, 96, 0, 2, 0, -87, -118, -57, 7, 91, -51, 21, 28, -123).plus(chunk1)
val expectedRtp2 = byteArrayOf(-128, -32, 0, 3, 0, -87, -118, -57, 7, 91, -51, 21, 28, 69).plus(chunk2)
val expectedRtp = byteArrayOf(-128, 96, 0, 1, 0, -87, -118, -57, 7, 91, -51, 21, 28, -123).plus(chunk1)
val expectedRtp2 = byteArrayOf(-128, -32, 0, 2, 0, -87, -118, -57, 7, 91, -51, 21, 28, 69).plus(chunk2)

val expectedStapA = byteArrayOf(-128, -32, 0, 1, 0, -87, -118, -57, 7, 91, -51, 21, 24, 0, 7, 0, 0, 0, 1, 2, 3, 4, 0, 7, 0, 0, 0, 1, 10, 11, 12)
val expectedTimeStamp = 11111111L
val expectedSize = chunk1.size + RtpConstants.RTP_HEADER_LENGTH + 2
val expectedSize2 = chunk2.size + RtpConstants.RTP_HEADER_LENGTH + 2
val expectedStapAResult = RtpFrame(expectedStapA, expectedTimeStamp, fakePps.size + fakePps.size + 5 + RtpConstants.RTP_HEADER_LENGTH, RtpConstants.trackVideo)

val expectedPacketResult = RtpFrame(expectedRtp, expectedTimeStamp, expectedSize, RtpConstants.trackVideo)
val expectedPacketResult2 = RtpFrame(expectedRtp2, expectedTimeStamp, expectedSize2, RtpConstants.trackVideo)

assertNotEquals(0, frames.size)
assertTrue(frames.size == 3)
assertEquals(expectedStapAResult, frames[0])
assertEquals(expectedPacketResult, frames[1])
assertEquals(expectedPacketResult2, frames[2])
assertTrue(frames.size == 2)
assertEquals(expectedPacketResult, frames[0])
assertEquals(expectedPacketResult2, frames[1])
}
}
Loading
Loading