RTMP: Support enhanced RTMP specification for HEVC. v6.0.42 (#3495)

* RTMP: Support enhanced RTMP specification for HEVC,  v6.0.42.
* Player: Upgrade mpegts.js to support it.

Enhanced RTMP specification: https://github.com/veovera/enhanced-rtmp

First, start SRS `v6.0.42+` with HTTP-TS support:

```bash
./objs/srs -c conf/http.ts.live.conf
```

Then, you can use [OBS 29.1+](https://github.com/obsproject/obs-studio/releases) to push HEVC via RTMP.
Start OBS with the following settings in the `Settings > Stream` tab:

* Server: `rtmp://localhost/live`
* Stream Key: `livestream`
* Encoder: Please select the HEVC hardware encoder.

Finally, open the player http://localhost:8080/players/srs_player.html?stream=livestream.ts

Or use VLS or ffplay to play `http://localhost:8080/live/livestream.ts`

---------

Co-authored-by: chundonglinlin <chundonglinlin@163.com>

trunk/doc/CHANGELOG.md

@@ -8,8 +8,9 @@ The changelog for SRS.
 
 ## SRS 6.0 Changelog
 
-* v5.0, 2023-04-01, Merge [#3392](https://github.com/ossrs/srs/pull/3392): Support composited bridges for 1:N protocols converting. v6.0.41 (#3392)
+* v6.0, 2023-04-08, Merge [#3495](https://github.com/ossrs/srs/pull/3495): RTMP: Support enhanced RTMP specification for HEVC. v6.0.42 (#3495)
-* v5.0, 2023-04-01, Merge [#3458](https://github.com/ossrs/srs/pull/3450): API: Support HTTP basic authentication for API. v6.0.40 (#3458)
+* v6.0, 2023-04-01, Merge [#3392](https://github.com/ossrs/srs/pull/3392): Support composited bridges for 1:N protocols converting. v6.0.41 (#3392)
+* v6.0, 2023-04-01, Merge [#3458](https://github.com/ossrs/srs/pull/3450): API: Support HTTP basic authentication for API. v6.0.40 (#3458)
 * v6.0, 2023-03-27, Merge [#3450](https://github.com/ossrs/srs/pull/3450): WebRTC: Error message carries the SDP when failed. v6.0.39 (#3450)
 * v6.0, 2023-03-25, Merge [#3477](https://github.com/ossrs/srs/pull/3477): Remove unnecessary NULL check in srs_freep. v6.0.38 (#3477)
 * v6.0, 2023-03-25, Merge [#3455](https://github.com/ossrs/srs/pull/3455): RTC: Call on_play before create session, for it might be freed for timeout. v6.0.37 (#3455)

trunk/research/players/srs_player.html

@@ -102,7 +102,7 @@
 </body>
 <script type="text/javascript" src="js/jquery-1.10.2.min.js"></script>
 <script type="text/javascript" src="js/bootstrap.min.js"></script>
-<script type="text/javascript" src="js/mpegts-1.7.2.min.js"></script>
+<script type="text/javascript" src="js/mpegts-1.7.3.min.js"></script>
 <script type="text/javascript" src="js/hls-0.14.17.min.js"></script>
 <script type="text/javascript" src="js/dash-v4.5.1.all.min.js"></script>
 <script type="text/javascript" src="js/json2.js"></script>

trunk/src/core/srs_core_version6.hpp

@@ -9,6 +9,6 @@
 
 #define VERSION_MAJOR       6
 #define VERSION_MINOR       0
-#define VERSION_REVISION    41
+#define VERSION_REVISION    42
 
 #endif

trunk/src/kernel/srs_kernel_codec.cpp

@@ -154,7 +154,9 @@ bool SrsFlvVideo::keyframe(char* data, int size)
         return false;
     }
 
-    char frame_type = data[0];
+    // See rtmp_specification_1.0.pdf
+    // See https://github.com/veovera/enhanced-rtmp
+    uint8_t frame_type = data[0] & 0x7f;
     frame_type = (frame_type >> 4) & 0x0F;
     
     return frame_type == SrsVideoAvcFrameTypeKeyFrame;
@@ -174,11 +176,20 @@ bool SrsFlvVideo::sh(char* data, int size)
         return false;
     }
 
-    char frame_type = data[0];
+    uint8_t frame_type = data[0];
+    bool is_ext_header = frame_type & 0x80;
+    SrsVideoAvcFrameTrait avc_packet_type = SrsVideoAvcFrameTraitForbidden;
+    if (!is_ext_header) {
+        // See rtmp_specification_1.0.pdf
         frame_type = (frame_type >> 4) & 0x0F;
+        avc_packet_type = (SrsVideoAvcFrameTrait)data[1];
+    } else {
+        // See https://github.com/veovera/enhanced-rtmp
+        avc_packet_type = (SrsVideoAvcFrameTrait)(frame_type & 0x0f);
+        frame_type = (frame_type >> 4) & 0x07;
+    }
 
-    char avc_packet_type = data[1];
+    // Note that SrsVideoHEVCFrameTraitPacketTypeSequenceStart is equal to SrsVideoAvcFrameTraitSequenceHeader
-    
     return frame_type == SrsVideoAvcFrameTypeKeyFrame
         && avc_packet_type == SrsVideoAvcFrameTraitSequenceHeader;
 }
@@ -204,8 +215,24 @@ bool SrsFlvVideo::hevc(char* data, int size)
         return false;
     }
 
-    char codec_id = data[0];
+    uint8_t frame_type = data[0];
-    codec_id = codec_id & 0x0F;
+    bool is_ext_header = frame_type & 0x80;
+    SrsVideoCodecId codec_id = SrsVideoCodecIdForbidden;
+    if (!is_ext_header) {
+        // See rtmp_specification_1.0.pdf
+        codec_id = (SrsVideoCodecId)(frame_type & 0x0F);
+    } else {
+        // See https://github.com/veovera/enhanced-rtmp
+        if (size < 5) {
+            return false;
+        }
+
+        // Video FourCC
+        if (data[1] != 'h' || data[2] != 'v' || data[3] != 'c' || data[4] != '1') {
+            return false;
+        }
+        codec_id = SrsVideoCodecIdHEVC;
+    }
 
     return codec_id == SrsVideoCodecIdHEVC;
 }
@@ -218,13 +245,35 @@ bool SrsFlvVideo::acceptable(char* data, int size)
         return false;
     }
     
-    char frame_type = data[0];
+    uint8_t frame_type = data[0];
-    SrsVideoCodecId codec_id = (SrsVideoCodecId)(uint8_t)(frame_type & 0x0f);
+    bool is_ext_header = frame_type & 0x80;
+    SrsVideoCodecId codec_id = SrsVideoCodecIdForbidden;
+    if (!is_ext_header) {
+        // See rtmp_specification_1.0.pdf
+        codec_id = (SrsVideoCodecId)(frame_type & 0x0f);
         frame_type = (frame_type >> 4) & 0x0f;
 
         if (frame_type < 1 || frame_type > 5) {
             return false;
         }
+    } else {
+        // See https://github.com/veovera/enhanced-rtmp
+        uint8_t packet_type = frame_type & 0x0f;
+        frame_type = (frame_type >> 4) & 0x07;
+
+        if (packet_type > SrsVideoHEVCFrameTraitPacketTypeMPEG2TSSequenceStart || frame_type > SrsVideoAvcFrameTypeVideoInfoFrame) {
+            return false;
+        }
+
+        if (size < 5) {
+            return false;
+        }
+
+        if (data[1] != 'h' || data[2] != 'v' || data[3] != 'c' || data[4] != '1') {
+            return false;
+        }
+        codec_id = SrsVideoCodecIdHEVC;
+    }
     
     if (codec_id != SrsVideoCodecIdAVC && codec_id != SrsVideoCodecIdAV1 && codec_id != SrsVideoCodecIdHEVC) {
         return false;
@@ -776,32 +825,6 @@ srs_error_t SrsFormat::on_video(int64_t timestamp, char* data, int size)
     SrsBuffer* buffer = new SrsBuffer(data, size);
     SrsAutoFree(SrsBuffer, buffer);
 
-    // We already checked the size is positive and data is not NULL.
-    srs_assert(buffer->require(1));
-    
-    // @see: E.4.3 Video Tags, video_file_format_spec_v10_1.pdf, page 78
-    int8_t frame_type = buffer->read_1bytes();
-    SrsVideoCodecId codec_id = (SrsVideoCodecId)(frame_type & 0x0f);
-
-    // Check codec for H.264 and H.265.
-    bool codec_ok = (codec_id == SrsVideoCodecIdAVC);
-#ifdef SRS_H265
-    codec_ok = codec_ok ? true : (codec_id == SrsVideoCodecIdHEVC);
-#endif
-    if (!codec_ok) return err;
-    
-    if (!vcodec) {
-        vcodec = new SrsVideoCodecConfig();
-    }
-    if (!video) {
-        video = new SrsVideoFrame();
-    }
-    
-    if ((err = video->initialize(vcodec)) != srs_success) {
-        return srs_error_wrap(err, "init video");
-    }
-    
-    buffer->skip(-1 * buffer->pos());
     return video_avc_demux(buffer, timestamp);
 }
 
@@ -848,17 +871,54 @@ srs_error_t SrsFormat::video_avc_demux(SrsBuffer* stream, int64_t timestamp)
 {
     srs_error_t err = srs_success;
 
+    if (!stream->require(1)) {
+        return srs_error_new(ERROR_HLS_DECODE_ERROR, "video avc demux shall atleast 1bytes");
+    }
+
+    // Parse the frame type and the first bit indicates the ext header.
+    uint8_t frame_type = stream->read_1bytes();
+    bool is_ext_header = frame_type & 0x80;
+
     // @see: E.4.3 Video Tags, video_file_format_spec_v10_1.pdf, page 78
-    int8_t frame_type = stream->read_1bytes();
+    SrsVideoCodecId codec_id = SrsVideoCodecIdForbidden;
-    SrsVideoCodecId codec_id = (SrsVideoCodecId)(frame_type & 0x0f);
+    SrsVideoAvcFrameTrait packet_type = SrsVideoAvcFrameTraitForbidden;
+    if (!is_ext_header) {
+        // See rtmp_specification_1.0.pdf
+        codec_id = (SrsVideoCodecId)(frame_type & 0x0f);
         frame_type = (frame_type >> 4) & 0x0f;
+    } else {
+        // See https://github.com/veovera/enhanced-rtmp
+        packet_type = (SrsVideoAvcFrameTrait)(frame_type & 0x0f);
+        frame_type = (frame_type >> 4) & 0x07;
+
+        if (!stream->require(4)) {
+            return srs_error_new(ERROR_HLS_DECODE_ERROR, "fourCC requires 4bytes, only %dbytes", stream->left());
+        }
+
+        uint32_t four_cc = stream->read_4bytes();
+        if (four_cc == 0x68766331) { // 'hvc1'=0x68766331
+            codec_id = SrsVideoCodecIdHEVC;
+        }
+    }
+
+    if (!vcodec) {
+        vcodec = new SrsVideoCodecConfig();
+    }
+
+    if (!video) {
+        video = new SrsVideoFrame();
+    }
+
+    if ((err = video->initialize(vcodec)) != srs_success) {
+        return srs_error_wrap(err, "init video");
+    }
 
     video->frame_type = (SrsVideoAvcFrameType)frame_type;
     
     // ignore info frame without error,
     // @see https://github.com/ossrs/srs/issues/288#issuecomment-69863909
     if (video->frame_type == SrsVideoAvcFrameTypeVideoInfoFrame) {
-        srs_warn("avc igone the info frame");
+        srs_warn("avc ignore the info frame");
         return err;
     }
 
@@ -872,16 +932,28 @@ srs_error_t SrsFormat::video_avc_demux(SrsBuffer* stream, int64_t timestamp)
     }
     vcodec->id = codec_id;
 
+    int32_t composition_time = 0;
+    if (!is_ext_header) {
+        // See rtmp_specification_1.0.pdf
         if (!stream->require(4)) {
-        return srs_error_new(ERROR_HLS_DECODE_ERROR, "avc decode avc_packet_type");
+            return srs_error_new(ERROR_HLS_DECODE_ERROR, "requires 4bytes, only %dbytes", stream->left());
+        }
+        packet_type = (SrsVideoAvcFrameTrait)stream->read_1bytes();
+        composition_time = stream->read_3bytes();
+    } else {
+        // See https://github.com/veovera/enhanced-rtmp
+        if (packet_type == SrsVideoHEVCFrameTraitPacketTypeCodedFrames) {
+            if (!stream->require(3)) {
+                return srs_error_new(ERROR_HLS_DECODE_ERROR, "requires 3 bytes, only %dbytes", stream->left());
+            }
+            composition_time = stream->read_3bytes();
+        }
     }
-    int8_t avc_packet_type = stream->read_1bytes();
-    int32_t composition_time = stream->read_3bytes();
 
     // pts = dts + cts.
     video->dts = timestamp;
     video->cts = composition_time;
-    video->avc_packet_type = (SrsVideoAvcFrameTrait)avc_packet_type;
+    video->avc_packet_type = packet_type;
     
     // Update the RAW AVC data.
     raw = stream->data() + stream->pos();
@@ -890,12 +962,12 @@ srs_error_t SrsFormat::video_avc_demux(SrsBuffer* stream, int64_t timestamp)
     // Parse sequence header for H.265/HEVC.
     if (codec_id == SrsVideoCodecIdHEVC) {
 #ifdef SRS_H265
-        if (avc_packet_type == SrsVideoAvcFrameTraitSequenceHeader) {
+        if (packet_type == SrsVideoAvcFrameTraitSequenceHeader) {
             // TODO: demux vps/sps/pps for hevc
             if ((err = hevc_demux_hvcc(stream)) != srs_success) {
                 return srs_error_wrap(err, "demux hevc VPS/SPS/PPS");
             }
-        } else if (avc_packet_type == SrsVideoAvcFrameTraitNALU) {
+        } else if (packet_type == SrsVideoAvcFrameTraitNALU || packet_type == SrsVideoHEVCFrameTraitPacketTypeCodedFramesX) {
             // TODO: demux nalu for hevc
             if ((err = video_nalu_demux(stream)) != srs_success) {
                 return srs_error_wrap(err, "demux hevc NALU");
@@ -908,12 +980,12 @@ srs_error_t SrsFormat::video_avc_demux(SrsBuffer* stream, int64_t timestamp)
     }
 
     // Parse sequence header for H.264/AVC.
-    if (avc_packet_type == SrsVideoAvcFrameTraitSequenceHeader) {
+    if (packet_type == SrsVideoAvcFrameTraitSequenceHeader) {
         // TODO: FIXME: Maybe we should ignore any error for parsing sps/pps.
         if ((err = avc_demux_sps_pps(stream)) != srs_success) {
             return srs_error_wrap(err, "demux SPS/PPS");
         }
-    } else if (avc_packet_type == SrsVideoAvcFrameTraitNALU){
+    } else if (packet_type == SrsVideoAvcFrameTraitNALU){
         if ((err = video_nalu_demux(stream)) != srs_success) {
             return srs_error_wrap(err, "demux NALU");
         }

trunk/src/kernel/srs_kernel_codec.hpp

@@ -64,12 +64,32 @@ std::string srs_video_codec_id2str(SrsVideoCodecId codec);
 enum SrsVideoAvcFrameTrait
 {
     // set to the max value to reserved, for array map.
-    SrsVideoAvcFrameTraitReserved = 3,
+    SrsVideoAvcFrameTraitReserved = 6,
-    SrsVideoAvcFrameTraitForbidden = 3,
+    SrsVideoAvcFrameTraitForbidden = 6,
     
     SrsVideoAvcFrameTraitSequenceHeader = 0,
     SrsVideoAvcFrameTraitNALU = 1,
     SrsVideoAvcFrameTraitSequenceHeaderEOF = 2,
+
+    SrsVideoHEVCFrameTraitPacketTypeSequenceStart = 0,
+    SrsVideoHEVCFrameTraitPacketTypeCodedFrames = 1,
+    SrsVideoHEVCFrameTraitPacketTypeSequenceEnd = 2,
+    // CompositionTime Offset is implied to equal zero. This is
+    // an optimization to save putting SI24 composition time value of zero on
+    // the wire. See pseudo code below in the VideoTagBody section
+    SrsVideoHEVCFrameTraitPacketTypeCodedFramesX = 3,
+    // VideoTagBody does not contain video data. VideoTagBody
+    // instead contains an AMF encoded metadata. See Metadata Frame
+    // section for an illustration of its usage. As an example, the metadata
+    // can be HDR information. This is a good way to signal HDR
+    // information. This also opens up future ways to express additional
+    // metadata that is meant for the next video sequence.
+    //
+    // note: presence of PacketTypeMetadata means that FrameType
+    // flags at the top of this table should be ignored
+    SrsVideoHEVCFrameTraitPacketTypeMetadata = 4,
+    // Carriage of bitstream in MPEG-2 TS format
+    SrsVideoHEVCFrameTraitPacketTypeMPEG2TSSequenceStart = 5,
 };
 
 /**

trunk/src/utest/srs_utest_kernel.cpp

@@ -3897,7 +3897,8 @@ VOID TEST(KernelCodecTest, VideoFormat)
         
         HELPER_EXPECT_SUCCESS(f.on_video(0, NULL, 0));
         HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x00", 0));
-        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x00", 1));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x00", 1));
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x57", 1));
     }
     
     if (true) {
@@ -4001,7 +4002,16 @@ VOID TEST(KernelCodecTest, HevcVideoFormat)
 
         HELPER_EXPECT_SUCCESS(f.on_video(0, NULL, 0));
         HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x00", 0));
-        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x00", 1));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x00", 1));
+
+        // enhanced rtmp/flv
+        HELPER_EXPECT_SUCCESS(f.on_video(0, NULL, 0));
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x80", 0));
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x90", 0));
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\xd0\x68\x76\x63\x31", 5));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x80", 1));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x90", 1));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x90\x68\x76\x63\x31", 5));
     }
 
     if (true) {
@@ -4010,11 +4020,22 @@ VOID TEST(KernelCodecTest, HevcVideoFormat)
 
         //HEVC: 0x5c
         HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)"\x5c", 1));
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*)"\x1c", 1));
 
         // CodecId: 0x00
         SrsBuffer b((char*)"\x00", 1);
         srs_error_t err = f.video_avc_demux(&b, 0);
         HELPER_EXPECT_FAILED(err);
+
+        // enhanced rtmp/flv
+        SrsBuffer b1((char*)"\x80", 1);
+        HELPER_EXPECT_FAILED(f.video_avc_demux(&b1, 0));
+        SrsBuffer b2((char*)"\x90", 1);
+        HELPER_EXPECT_FAILED(f.video_avc_demux(&b2, 0));
+        SrsBuffer b3((char*)"\x90\x68\x76\x63\x31", 5);
+        HELPER_EXPECT_FAILED(f.video_avc_demux(&b3, 0));
+        SrsBuffer b4((char*)"\xd0\x68\x76\x63\x31", 5);
+        HELPER_EXPECT_SUCCESS(f.video_avc_demux(&b4, 0));
     }
 
     uint8_t vps_sps_pps[] = {
@@ -4075,6 +4096,93 @@ VOID TEST(KernelCodecTest, HevcVideoFormat)
         HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)rawIBMF, sizeof(rawIBMF)));
         EXPECT_EQ(1, f.video->nb_samples);
     }
+
+    // enhanced rtmp
+    uint8_t ext_vps_sps_pps[] = {
+        // IsExHeader | FrameType: UB[4]
+        // PacketType: UB[4]
+        0x90,
+        // Video FourCC
+        0x68, 0x76, 0x63, 0x31,
+        // SrsHevcDecoderConfigurationRecord
+        0x01, 0x01, 0x60, 0x00, 0x00, 0x00, 0x90, 0x00, 0x00, 0x00, 0x00, 0x00, 0x5d, 0xf0, 0x00, 0xfc, 0xfd, 0xf8, 0xf8, 0x00, 0x00, 0x0f, 0x03,
+        // Nalus
+        // data_byte(1B)+num_nalus(2B)+nal_unit_length(2B)
+        0x20, 0x00, 0x01, 0x00, 0x18,
+        // VPS
+        0x40, 0x01, 0x0c, 0x01, 0xff, 0xff, 0x01, 0x60, 0x00, 0x00, 0x03, 0x00, 0x90, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x5d, 0x95, 0x98, 0x09,
+        // data_byte(1B)+num_nalus(2B)+nal_unit_length(2B)
+        0x21, 0x00, 0x01, 0x00, 0x28,
+        // SPS
+        0x42, 0x01, 0x01, 0x01, 0x60, 0x00, 0x00, 0x03, 0x00, 0x90, 0x00, 0x00, 0x03, 0x00, 0x00, 0x03, 0x00, 0x5d, 0xa0, 0x02, 0x80, 0x80, 0x2d, 0x16,
+        0x59, 0x59, 0xa4, 0x93, 0x2b, 0xc0, 0x40, 0x40, 0x00, 0x00, 0xfa, 0x40, 0x00, 0x17, 0x70, 0x02,
+        // data_byte(1B)+num_nalus(2B)+nal_unit_length(2B)
+        0x22, 0x00, 0x01, 0x00, 0x07,
+        // PPS
+        0x44, 0x01, 0xc1, 0x72, 0xb4, 0x62, 0x40
+    };
+
+    uint8_t ext_rawIBMF[] = {
+        // IsExHeader | FrameType: UB[4]
+        // PacketType: UB[4]
+        0x93,
+        // Video FourCC
+        0x68, 0x76, 0x63, 0x31,
+        // HEVC NALU
+        0x00, 0x00, 0x00, 0x0b,
+        0x28, 0x1, 0xaf, 0x1d, 0x18, 0x38, 0xd4, 0x38, 0x32, 0xda, 0x23
+    };
+
+    uint8_t ext_rawIBMF1[] = {
+        // IsExHeader | FrameType: UB[4]
+        // PacketType: UB[4]
+        0x91,
+        // Video FourCC
+        0x68, 0x76, 0x63, 0x31,
+        // CompositionTime Offset
+        0x00, 0x00, 0x7d,
+        // HEVC NALU
+        0x00, 0x00, 0x00, 0x0b,
+        0x28, 0x1, 0xaf, 0x1d, 0x18, 0x38, 0xd4, 0x38, 0x32, 0xda, 0x23
+    };
+
+    if (true) {
+        SrsFormat f;
+        HELPER_EXPECT_SUCCESS(f.initialize());
+
+        // firstly demux sequence header
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)ext_vps_sps_pps, sizeof(ext_vps_sps_pps)));
+        EXPECT_EQ(1, f.video->frame_type);
+        EXPECT_EQ(0, f.video->avc_packet_type);
+        EXPECT_EQ(3, f.vcodec->hevc_dec_conf_record_.nalu_vec.size());
+        EXPECT_EQ(1280, f.vcodec->width);
+        EXPECT_EQ(720, f.vcodec->height);
+
+        // secondly demux sequence header
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)ext_vps_sps_pps, sizeof(ext_vps_sps_pps)));
+        EXPECT_EQ(1, f.video->frame_type);
+        EXPECT_EQ(0, f.video->avc_packet_type);
+        EXPECT_EQ(3, f.vcodec->hevc_dec_conf_record_.nalu_vec.size());
+        EXPECT_EQ(1280, f.vcodec->width);
+        EXPECT_EQ(720, f.vcodec->height);
+
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)ext_rawIBMF, sizeof(ext_rawIBMF)));
+        EXPECT_EQ(1, f.video->frame_type);
+        EXPECT_EQ(3, f.video->avc_packet_type);
+        EXPECT_EQ(1, f.video->nb_samples);
+
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)ext_rawIBMF, sizeof(ext_rawIBMF)));
+        EXPECT_EQ(1, f.video->frame_type);
+        EXPECT_EQ(3, f.video->avc_packet_type);
+        EXPECT_EQ(1, f.video->nb_samples);
+
+        // check cts
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*)ext_rawIBMF1, sizeof(ext_rawIBMF1)));
+        EXPECT_EQ(1, f.video->frame_type);
+        EXPECT_EQ(125, f.video->cts);
+        EXPECT_EQ(1, f.video->avc_packet_type);
+        EXPECT_EQ(1, f.video->nb_samples);
+    }
 }
 #endif
 

trunk/src/utest/srs_utest_kernel2.cpp

@@ -411,3 +411,75 @@ VOID TEST(KernelFileWriterTest, RealfileTest)
     EXPECT_STREQ("HelloWorld", str.substr(20).c_str());
 }
 
+VOID TEST(KernelRTMPExtTest, ExtRTMPTest)
+{
+    srs_error_t err;
+
+    // For legacy RTMP specification, without ext tag header.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*) "\x17\x01\x00\x00\x12", 5));
+
+        // Verify the frame type, codec id, avc packet type and composition time.
+        EXPECT_EQ(SrsVideoAvcFrameTypeKeyFrame, f.video->frame_type);
+        EXPECT_EQ(SrsVideoCodecIdAVC, f.vcodec->id);
+        EXPECT_EQ(SrsVideoAvcFrameTraitNALU, f.video->avc_packet_type);
+        EXPECT_EQ(0x12, f.video->cts);
+    }
+
+    // For new RTMP enhanced specification, with ext tag header.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*) "\x91hvc1\x00\x00\x12", 8));
+
+        // Verify the frame type, codec id, avc packet type and composition time.
+        EXPECT_EQ(SrsVideoAvcFrameTypeKeyFrame, f.video->frame_type);
+        EXPECT_EQ(SrsVideoCodecIdHEVC, f.vcodec->id);
+        EXPECT_EQ(SrsVideoHEVCFrameTraitPacketTypeCodedFrames, f.video->avc_packet_type);
+        EXPECT_EQ(0x12, f.video->cts);
+    }
+
+    // If packet type is 3, which is coded frame X, the composition time is 0.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_SUCCESS(f.on_video(0, (char*) "\x93hvc1", 5));
+
+        // Verify the frame type, codec id, avc packet type and composition time.
+        EXPECT_EQ(SrsVideoAvcFrameTypeKeyFrame, f.video->frame_type);
+        EXPECT_EQ(SrsVideoCodecIdHEVC, f.vcodec->id);
+        EXPECT_EQ(SrsVideoHEVCFrameTraitPacketTypeCodedFramesX, f.video->avc_packet_type);
+        EXPECT_EQ(0, f.video->cts);
+    }
+
+    // Should fail if only 1 byte for ext tag header, should be more bytes for fourcc.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*) "\x91", 1));
+    }
+
+    // Should fail if only 5 bytes for ext tag header, should be more bytes for fourcc.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*) "\x91hvc1", 5));
+    }
+
+    // Should fail if codec id is hvc2 for ext tag header, should be hvc1.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*) "\x93hvc2", 5));
+    }
+
+    // Should fail if codec id is mvc1 for ext tag header, should be hvc1.
+    if (true) {
+        SrsFormat f;
+        HELPER_ASSERT_SUCCESS(f.initialize());
+        HELPER_EXPECT_FAILED(f.on_video(0, (char*) "\x93mvc1", 5));
+    }
+}
+