| 1 | /********** |
| 2 | This library is free software; you can redistribute it and/or modify it under |
| 3 | the terms of the GNU Lesser General Public License as published by the |
| 4 | Free Software Foundation; either version 3 of the License, or (at your |
| 5 | option) any later version. (See <http://www.gnu.org/copyleft/lesser.html>.) |
| 6 | |
| 7 | This library is distributed in the hope that it will be useful, but WITHOUT |
| 8 | ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS |
| 9 | FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for |
| 10 | more details. |
| 11 | |
| 12 | You should have received a copy of the GNU Lesser General Public License |
| 13 | along with this library; if not, write to the Free Software Foundation, Inc., |
| 14 | 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 15 | **********/ |
| 16 | // "liveMedia" |
| 17 | // Copyright (c) 1996-2020 Live Networks, Inc. All rights reserved. |
| 18 | // A simplified version of "H264or5VideoStreamFramer" that takes only complete, |
| 19 | // discrete frames (rather than an arbitrary byte stream) as input. |
| 20 | // This avoids the parsing and data copying overhead of the full |
| 21 | // "H264or5VideoStreamFramer". |
| 22 | // Implementation |
| 23 | |
| 24 | #include "H264or5VideoStreamDiscreteFramer.hh" |
| 25 | |
| 26 | H264or5VideoStreamDiscreteFramer |
| 27 | ::H264or5VideoStreamDiscreteFramer(int hNumber, UsageEnvironment& env, FramedSource* inputSource, |
| 28 | Boolean includeStartCodeInOutput, |
| 29 | Boolean insertAccessUnitDelimiters) |
| 30 | : H264or5VideoStreamFramer(hNumber, env, inputSource, False/*don't create a parser*/, |
| 31 | includeStartCodeInOutput, insertAccessUnitDelimiters) { |
| 32 | } |
| 33 | |
| 34 | H264or5VideoStreamDiscreteFramer::~H264or5VideoStreamDiscreteFramer() { |
| 35 | } |
| 36 | |
| 37 | void H264or5VideoStreamDiscreteFramer::doGetNextFrame() { |
| 38 | if (fIncludeStartCodeInOutput) { |
| 39 | // Prepend a 4-byte 'start code' (0x00000001) to the output: |
| 40 | if (fMaxSize < 4) { // there's no space |
| 41 | fNumTruncatedBytes = 4 - fMaxSize; |
| 42 | handleClosure(); |
| 43 | return; |
| 44 | } |
| 45 | *fTo++ = 0x00; *fTo++ = 0x00; *fTo++ = 0x00; *fTo++ = 0x01; |
| 46 | fMaxSize -= 4; |
| 47 | } |
| 48 | |
| 49 | if (fInsertAccessUnitDelimiters && pictureEndMarker()) { |
| 50 | // Deliver an "access_unit_delimiter" NAL unit instead: |
| 51 | unsigned const audNALSize = fHNumber == 264 ? 2 : 3; |
| 52 | |
| 53 | if (audNALSize > fMaxSize) { // there's no space |
| 54 | fNumTruncatedBytes = audNALSize - fMaxSize; |
| 55 | handleClosure(); |
| 56 | return; |
| 57 | } |
| 58 | |
| 59 | if (fHNumber == 264) { |
| 60 | *fTo++ = 9; // "Access unit delimiter" nal_unit_type |
| 61 | *fTo++ = 0xF0; // "primary_pic_type" (7); "rbsp_trailing_bits()" |
| 62 | } else { // H.265 |
| 63 | *fTo++ = 35<<1; // "Access unit delimiter" nal_unit_type |
| 64 | *fTo++ = 0; // "nuh_layer_id" (0); "nuh_temporal_id_plus1" (0) (Is this correct??) |
| 65 | *fTo++ = 0x50; // "pic_type" (2); "rbsp_trailing_bits()" (Is this correct??) |
| 66 | } |
| 67 | |
| 68 | fFrameSize = (fIncludeStartCodeInOutput ? 4: 0) + audNALSize; |
| 69 | pictureEndMarker() = False; // for next time |
| 70 | afterGetting(this); // complete delivery to the downstream object |
| 71 | } else { |
| 72 | // Normal case: |
| 73 | // Arrange to read data (which should be a complete H.264 or H.265 NAL unit) |
| 74 | // from our data source, directly into the client's input buffer. |
| 75 | // After reading this, we'll do some parsing on the frame. |
| 76 | fInputSource->getNextFrame(fTo, fMaxSize, |
| 77 | afterGettingFrame, this, |
| 78 | FramedSource::handleClosure, this); |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | void H264or5VideoStreamDiscreteFramer |
| 83 | ::afterGettingFrame(void* clientData, unsigned frameSize, |
| 84 | unsigned numTruncatedBytes, |
| 85 | struct timeval presentationTime, |
| 86 | unsigned durationInMicroseconds) { |
| 87 | H264or5VideoStreamDiscreteFramer* source = (H264or5VideoStreamDiscreteFramer*)clientData; |
| 88 | source->afterGettingFrame1(frameSize, numTruncatedBytes, presentationTime, durationInMicroseconds); |
| 89 | } |
| 90 | |
| 91 | void H264or5VideoStreamDiscreteFramer |
| 92 | ::afterGettingFrame1(unsigned frameSize, unsigned numTruncatedBytes, |
| 93 | struct timeval presentationTime, |
| 94 | unsigned durationInMicroseconds) { |
| 95 | // Get the "nal_unit_type", to see if this NAL unit is one that we want to save a copy of: |
| 96 | u_int8_t nal_unit_type; |
| 97 | if (fHNumber == 264 && frameSize >= 1) { |
| 98 | nal_unit_type = fTo[0]&0x1F; |
| 99 | } else if (fHNumber == 265 && frameSize >= 2) { |
| 100 | nal_unit_type = (fTo[0]&0x7E)>>1; |
| 101 | } else { |
| 102 | // This is too short to be a valid NAL unit, so just assume a bogus nal_unit_type |
| 103 | nal_unit_type = 0xFF; |
| 104 | } |
| 105 | |
| 106 | // Begin by checking for a (likely) common error: NAL units that (erroneously) begin with a |
| 107 | // 0x00000001 or 0x000001 'start code'. (Those start codes should only be in byte-stream data; |
| 108 | // *not* data that consists of discrete NAL units.) |
| 109 | // Once again, to be clear: The NAL units that you feed to a "H264or5VideoStreamDiscreteFramer" |
| 110 | // MUST NOT include start codes. |
| 111 | if (frameSize >= 4 && fTo[0] == 0 && fTo[1] == 0 && ((fTo[2] == 0 && fTo[3] == 1) || fTo[2] == 1)) { |
| 112 | envir() << "H264or5VideoStreamDiscreteFramer error: MPEG 'start code' seen in the input\n" ; |
| 113 | } else if (isVPS(nal_unit_type)) { // Video parameter set (VPS) |
| 114 | saveCopyOfVPS(fTo, frameSize); |
| 115 | } else if (isSPS(nal_unit_type)) { // Sequence parameter set (SPS) |
| 116 | saveCopyOfSPS(fTo, frameSize); |
| 117 | } else if (isPPS(nal_unit_type)) { // Picture parameter set (PPS) |
| 118 | saveCopyOfPPS(fTo, frameSize); |
| 119 | } |
| 120 | |
| 121 | fPictureEndMarker = nalUnitEndsAccessUnit(nal_unit_type); |
| 122 | |
| 123 | // Finally, complete delivery to the client: |
| 124 | fFrameSize = fIncludeStartCodeInOutput ? (4+frameSize) : frameSize; |
| 125 | fNumTruncatedBytes = numTruncatedBytes; |
| 126 | fPresentationTime = presentationTime; |
| 127 | fDurationInMicroseconds = durationInMicroseconds; |
| 128 | afterGetting(this); |
| 129 | } |
| 130 | |
| 131 | Boolean H264or5VideoStreamDiscreteFramer::nalUnitEndsAccessUnit(u_int8_t nal_unit_type) { |
| 132 | // Check whether this NAL unit ends the current 'access unit' (basically, a video frame). |
| 133 | // Unfortunately, we can't do this reliably, because we don't yet know anything about the |
| 134 | // *next* NAL unit that we'll see. So, we guess this as best as we can, by assuming that |
| 135 | // if this NAL unit is a VCL NAL unit, then it ends the current 'access unit'. |
| 136 | // |
| 137 | // This will be wrong if you are streaming multiple 'slices' per picture. In that case, |
| 138 | // you can define a subclass that reimplements this virtual function to do the right thing. |
| 139 | |
| 140 | return isVCL(nal_unit_type); |
| 141 | } |
| 142 | |