Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
// ZoneMinder Video Storage Implementation
|
|
|
|
// Written by Chris Wiggins
|
|
|
|
// http://chriswiggins.co.nz
|
|
|
|
// Modification by Steve Gilvarry
|
|
|
|
//
|
|
|
|
// This program is free software; you can redistribute it and/or
|
|
|
|
// modify it under the terms of the GNU General Public License
|
|
|
|
// as published by the Free Software Foundation; either version 2
|
|
|
|
// of the License, or (at your option) any later version.
|
|
|
|
//
|
|
|
|
// This program is distributed in the hope that it will be useful,
|
|
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
// GNU General Public License for more details.
|
|
|
|
//
|
|
|
|
// You should have received a copy of the GNU General Public License
|
|
|
|
// along with this program; if not, write to the Free Software
|
|
|
|
// Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
|
|
//
|
|
|
|
|
|
|
|
#define __STDC_FORMAT_MACROS 1
|
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
#include <inttypes.h>
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
|
|
|
|
#include "zm.h"
|
|
|
|
#include "zm_videostore.h"
|
|
|
|
|
2017-06-03 04:52:52 +08:00
|
|
|
extern "C" {
|
2017-08-23 22:37:32 +08:00
|
|
|
#include "libavutil/time.h"
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
VideoStore::CodecData VideoStore::codec_data[] = {
|
|
|
|
{ AV_CODEC_ID_H264, "h264", "h264_omx", AV_PIX_FMT_YUV420P },
|
|
|
|
{ AV_CODEC_ID_H264, "h264", "h264", AV_PIX_FMT_YUV420P },
|
|
|
|
{ AV_CODEC_ID_H264, "h264", "libx264", AV_PIX_FMT_YUV420P },
|
|
|
|
{ AV_CODEC_ID_MJPEG, "mjpeg", "mjpeg", AV_PIX_FMT_YUVJ422P },
|
|
|
|
};
|
|
|
|
|
2017-10-27 01:12:56 +08:00
|
|
|
VideoStore::VideoStore(
|
2017-11-13 00:42:34 +08:00
|
|
|
const char *filename_in,
|
|
|
|
const char *format_in,
|
2017-10-27 01:12:56 +08:00
|
|
|
AVStream *p_video_in_stream,
|
2017-11-13 00:42:34 +08:00
|
|
|
AVStream *p_audio_in_stream,
|
2017-12-05 07:06:42 +08:00
|
|
|
Monitor *p_monitor
|
2017-10-27 01:12:56 +08:00
|
|
|
) {
|
2017-12-05 07:06:42 +08:00
|
|
|
|
|
|
|
monitor = p_monitor;
|
2017-08-23 22:37:32 +08:00
|
|
|
video_in_stream = p_video_in_stream;
|
|
|
|
audio_in_stream = p_audio_in_stream;
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
filename = filename_in;
|
|
|
|
format = format_in;
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2017-10-27 01:12:56 +08:00
|
|
|
packets_written = 0;
|
|
|
|
frame_count = 0;
|
2017-12-05 07:06:42 +08:00
|
|
|
in_frame = NULL;
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2017-12-05 07:06:42 +08:00
|
|
|
converted_in_samples = NULL;
|
|
|
|
audio_out_codec = NULL;
|
|
|
|
audio_in_codec = NULL;
|
|
|
|
audio_in_ctx = NULL;
|
|
|
|
audio_out_stream = NULL;
|
|
|
|
out_frame = NULL;
|
|
|
|
#ifdef HAVE_LIBAVRESAMPLE
|
|
|
|
resample_ctx = NULL;
|
|
|
|
#endif
|
2017-12-02 03:48:30 +08:00
|
|
|
FFMPEGInit();
|
2017-12-01 03:37:36 +08:00
|
|
|
|
2017-12-09 04:17:45 +08:00
|
|
|
video_start_pts = 0;
|
2017-12-05 07:06:42 +08:00
|
|
|
audio_next_pts = 0;
|
|
|
|
audio_next_dts = 0;
|
|
|
|
} // VideoStore::VideoStore
|
|
|
|
|
|
|
|
bool VideoStore::open() {
|
2017-08-05 02:14:25 +08:00
|
|
|
Info("Opening video storage stream %s format: %s", filename, format);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2017-08-24 03:05:44 +08:00
|
|
|
ret = avformat_alloc_output_context2(&oc, NULL, NULL, filename);
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( ret < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
Warning(
|
|
|
|
"Could not create video storage stream %s as no out ctx"
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
" could be assigned based on filename: %s",
|
2017-08-23 22:37:32 +08:00
|
|
|
filename, av_make_error_string(ret).c_str());
|
2017-05-17 00:04:56 +08:00
|
|
|
} else {
|
2017-08-23 22:37:32 +08:00
|
|
|
Debug(4, "Success allocating out format ctx");
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
// Couldn't deduce format from filename, trying from format name
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( !oc ) {
|
2017-08-24 03:05:44 +08:00
|
|
|
avformat_alloc_output_context2(&oc, NULL, format, filename);
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( !oc ) {
|
2017-12-13 01:41:11 +08:00
|
|
|
Error(
|
2017-08-23 22:37:32 +08:00
|
|
|
"Could not create video storage stream %s as no out ctx"
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
" could not be assigned based on filename or format %s",
|
|
|
|
filename, format);
|
2017-12-13 04:30:47 +08:00
|
|
|
return false;
|
2017-08-11 03:44:20 +08:00
|
|
|
} else {
|
2017-10-27 01:12:56 +08:00
|
|
|
Debug(4, "Success alocating out ctx");
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
2017-12-01 08:26:51 +08:00
|
|
|
} // end if ! oc
|
2017-11-13 00:42:34 +08:00
|
|
|
Debug(2, "Success opening output contect");
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2017-05-17 00:04:56 +08:00
|
|
|
AVDictionary *pmetadata = NULL;
|
2017-12-01 08:26:51 +08:00
|
|
|
ret = av_dict_set(&pmetadata, "title", "Zoneminder Security Recording", 0);
|
|
|
|
if ( ret < 0 ) Warning("%s:%d: title set failed", __FILE__, __LINE__);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
|
|
|
oc->metadata = pmetadata;
|
2017-08-23 22:37:32 +08:00
|
|
|
out_format = oc->oformat;
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2017-12-17 06:08:21 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( video_in_stream ) {
|
|
|
|
video_in_stream_index = video_in_stream->index;
|
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
|
|
|
video_in_ctx = avcodec_alloc_context3(NULL);
|
2017-11-28 03:57:24 +08:00
|
|
|
Debug(2, "copy to video_in_context");
|
2017-12-13 04:29:19 +08:00
|
|
|
if ( (ret=avcodec_parameters_to_context(video_in_ctx, video_in_stream->codecpar)) < 0 ) {
|
|
|
|
Error("Couldn't copy params to context");
|
|
|
|
return false;
|
|
|
|
} else {
|
|
|
|
zm_dump_codecpar( video_in_stream->codecpar );
|
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
#else
|
|
|
|
video_in_ctx = video_in_stream->codec;
|
2017-12-01 08:26:51 +08:00
|
|
|
Debug(2,"Copied video context from input stream");
|
|
|
|
zm_dump_codec(video_in_ctx);
|
2017-11-13 00:42:34 +08:00
|
|
|
#endif
|
2017-11-18 04:37:26 +08:00
|
|
|
// Fix deprecated formats
|
|
|
|
switch ( video_in_ctx->pix_fmt ) {
|
|
|
|
case AV_PIX_FMT_YUVJ420P :
|
|
|
|
video_in_ctx->pix_fmt = AV_PIX_FMT_YUV420P;
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUVJ422P :
|
|
|
|
video_in_ctx->pix_fmt = AV_PIX_FMT_YUV422P;
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUVJ444P :
|
|
|
|
video_in_ctx->pix_fmt = AV_PIX_FMT_YUV444P;
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUVJ440P :
|
|
|
|
video_in_ctx->pix_fmt = AV_PIX_FMT_YUV440P;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
} else {
|
2017-11-13 00:42:34 +08:00
|
|
|
Debug(2, "No input ctx");
|
|
|
|
video_in_ctx = avcodec_alloc_context3(NULL);
|
|
|
|
video_in_stream_index = 0;
|
2017-06-01 08:44:23 +08:00
|
|
|
}
|
2017-08-23 22:37:32 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_ctx = avcodec_alloc_context3(NULL);
|
|
|
|
if ( oc->oformat->flags & AVFMT_GLOBALHEADER ) {
|
|
|
|
#if LIBAVCODEC_VERSION_CHECK(56, 35, 0, 64, 0)
|
|
|
|
video_out_ctx->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
|
|
|
|
#else
|
|
|
|
video_out_ctx->flags |= CODEC_FLAG_GLOBAL_HEADER;
|
2017-12-08 23:38:16 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
}
|
|
|
|
int wanted_codec = monitor->OutputCodec();
|
|
|
|
if ( ! wanted_codec ) {
|
|
|
|
// default to h264
|
|
|
|
wanted_codec = AV_CODEC_ID_H264;
|
2017-12-02 21:12:04 +08:00
|
|
|
}
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2017-12-09 03:22:38 +08:00
|
|
|
// // FIXME SHould check that we are set to passthrough
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( video_in_stream && ( video_in_ctx->codec_id == wanted_codec ) ) {
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2017-12-01 08:26:51 +08:00
|
|
|
ret = avcodec_parameters_to_context(video_out_ctx, video_in_stream->codecpar);
|
2017-12-02 21:12:04 +08:00
|
|
|
#else
|
|
|
|
ret = avcodec_copy_context( video_out_ctx, video_in_ctx );
|
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
// Copy params from instream to ctx
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( ret < 0 ) {
|
|
|
|
Error("Could not initialize ctx parameteres");
|
2017-12-05 07:06:42 +08:00
|
|
|
return false;
|
2017-11-13 00:42:34 +08:00
|
|
|
}
|
2017-12-15 04:56:22 +08:00
|
|
|
//video_out_ctx->time_base = (AVRational){1, 1000000}; // microseconds as base frame rate
|
|
|
|
video_out_ctx->time_base = video_in_ctx->time_base;
|
2017-11-18 04:37:26 +08:00
|
|
|
// Fix deprecated formats
|
|
|
|
switch ( video_out_ctx->pix_fmt ) {
|
|
|
|
case AV_PIX_FMT_YUVJ422P :
|
|
|
|
video_out_ctx->pix_fmt = AV_PIX_FMT_YUV422P;
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUVJ444P :
|
|
|
|
video_out_ctx->pix_fmt = AV_PIX_FMT_YUV444P;
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUVJ440P :
|
|
|
|
video_out_ctx->pix_fmt = AV_PIX_FMT_YUV440P;
|
|
|
|
break;
|
2017-12-15 04:56:22 +08:00
|
|
|
case AV_PIX_FMT_NONE :
|
|
|
|
case AV_PIX_FMT_YUVJ420P :
|
2017-11-18 04:37:26 +08:00
|
|
|
default:
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_ctx->pix_fmt = AV_PIX_FMT_YUV420P;
|
2017-11-18 04:37:26 +08:00
|
|
|
break;
|
|
|
|
}
|
2018-02-28 23:17:16 +08:00
|
|
|
// Only set orientation if doing passthrough, otherwise the frame image will be rotated
|
|
|
|
Monitor::Orientation orientation = monitor->getOrientation();
|
|
|
|
if ( orientation ) {
|
|
|
|
Debug(3, "Have orientation");
|
|
|
|
if ( orientation == Monitor::ROTATE_0 ) {
|
|
|
|
} else if ( orientation == Monitor::ROTATE_90 ) {
|
|
|
|
ret = av_dict_set(&video_out_stream->metadata, "rotate", "90", 0);
|
|
|
|
if ( ret < 0 ) Warning("%s:%d: title set failed", __FILE__, __LINE__);
|
|
|
|
} else if ( orientation == Monitor::ROTATE_180 ) {
|
|
|
|
ret = av_dict_set(&video_out_stream->metadata, "rotate", "180", 0);
|
|
|
|
if ( ret < 0 ) Warning("%s:%d: title set failed", __FILE__, __LINE__);
|
|
|
|
} else if ( orientation == Monitor::ROTATE_270 ) {
|
|
|
|
ret = av_dict_set(&video_out_stream->metadata, "rotate", "270", 0);
|
|
|
|
if ( ret < 0 ) Warning("%s:%d: title set failed", __FILE__, __LINE__);
|
|
|
|
} else {
|
|
|
|
Warning("Unsupported Orientation(%d)", orientation);
|
|
|
|
}
|
|
|
|
} // end if orientation
|
2017-10-27 01:12:56 +08:00
|
|
|
} else {
|
2017-11-13 00:42:34 +08:00
|
|
|
/** Create a new frame to store the */
|
2017-12-01 20:26:34 +08:00
|
|
|
if ( !(video_in_frame = zm_av_frame_alloc()) ) {
|
|
|
|
Error("Could not allocate video_in frame");
|
2017-12-05 07:06:42 +08:00
|
|
|
return false;
|
2017-10-27 01:12:56 +08:00
|
|
|
}
|
2018-02-28 23:17:16 +08:00
|
|
|
for (int i = 0; i < sizeof(codec_data) / sizeof(*codec_data); i++ ) {
|
|
|
|
if ( codec_data[i].codec_id != monitor->OutputCodec() )
|
|
|
|
continue;
|
2017-12-01 08:26:51 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_codec = avcodec_find_encoder_by_name(codec_data[i].codec_name);
|
|
|
|
if ( ! video_out_codec ) {
|
|
|
|
Debug(1, "Didn't find encoder for %s", codec_data[i].codec_name);
|
|
|
|
continue;
|
|
|
|
}
|
2017-11-14 14:59:15 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
Debug(1, "Using %s for codec", video_out_codec->name);
|
|
|
|
//video_out_ctx = avcodec_alloc_context3(video_out_codec);
|
|
|
|
if ( video_out_codec->id != video_out_ctx->codec_id ) {
|
|
|
|
Warning("Have to set codec_id?");
|
|
|
|
video_out_ctx->codec_id = AV_CODEC_ID_H264;
|
2017-11-13 00:42:34 +08:00
|
|
|
}
|
2017-11-14 14:59:15 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_ctx->pix_fmt = codec_data[i].pix_fmt;
|
|
|
|
video_out_ctx->level = 32;
|
|
|
|
|
|
|
|
// Don't have an input stream, so need to tell it what we are sending it, or are transcoding
|
|
|
|
video_out_ctx->width = monitor->Width();
|
|
|
|
video_out_ctx->height = monitor->Height();
|
|
|
|
video_out_ctx->codec_type = AVMEDIA_TYPE_VIDEO;
|
|
|
|
|
|
|
|
/* video time_base can be set to whatever is handy and supported by encoder */
|
|
|
|
//video_out_ctx->time_base = (AVRational){1, 1000000}; // microseconds as base frame rate
|
|
|
|
video_out_ctx->time_base = (AVRational){1, 30}; // microseconds as base frame rate
|
|
|
|
video_out_ctx->framerate = (AVRational){30,1};
|
|
|
|
//video_out_ctx->gop_size = 12;
|
|
|
|
//video_out_ctx->qmin = 10;
|
|
|
|
//video_out_ctx->qmax = 51;
|
|
|
|
//video_out_ctx->qcompress = 0.6;
|
|
|
|
video_out_ctx->bit_rate = 400*1024;
|
|
|
|
video_out_ctx->thread_count = 0;
|
|
|
|
|
|
|
|
if ( video_out_ctx->codec_id == AV_CODEC_ID_H264 ) {
|
|
|
|
video_out_ctx->max_b_frames = 1;
|
|
|
|
if ( video_out_ctx->priv_data ) {
|
|
|
|
av_opt_set(video_out_ctx->priv_data, "crf", "1", AV_OPT_SEARCH_CHILDREN);
|
|
|
|
//av_opt_set(video_out_ctx->priv_data, "preset", "ultrafast", 0);
|
|
|
|
} else {
|
|
|
|
Debug(2, "Not setting priv_data");
|
|
|
|
}
|
|
|
|
} else if (video_out_ctx->codec_id == AV_CODEC_ID_MPEG2VIDEO) {
|
|
|
|
/* just for testing, we also add B frames */
|
|
|
|
video_out_ctx->max_b_frames = 2;
|
|
|
|
} else if (video_out_ctx->codec_id == AV_CODEC_ID_MPEG1VIDEO) {
|
|
|
|
/* Needed to avoid using macroblocks in which some coeffs overflow.
|
|
|
|
* This does not happen with normal video, it just happens here as
|
|
|
|
* the motion of the chroma plane does not match the luma plane. */
|
|
|
|
video_out_ctx->mb_decision = 2;
|
2017-12-02 21:12:04 +08:00
|
|
|
}
|
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
AVDictionary *opts = 0;
|
|
|
|
std::string Options = monitor->GetEncoderOptions();
|
|
|
|
ret = av_dict_parse_string(&opts, Options.c_str(), "=", ",#\n", 0);
|
|
|
|
if ( ret < 0 ) {
|
|
|
|
Warning("Could not parse ffmpeg encoder options list '%s'\n", Options.c_str());
|
|
|
|
} else {
|
|
|
|
AVDictionaryEntry *e = NULL;
|
|
|
|
while ( (e = av_dict_get(opts, "", e, AV_DICT_IGNORE_SUFFIX)) != NULL ) {
|
|
|
|
Debug( 3, "Encoder Option %s=%s", e->key, e->value );
|
2017-11-14 14:59:15 +08:00
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
}
|
2018-02-28 23:17:16 +08:00
|
|
|
|
2017-12-02 21:12:04 +08:00
|
|
|
if ( (ret = avcodec_open2(video_out_ctx, video_out_codec, &opts)) < 0 ) {
|
2018-02-28 23:17:16 +08:00
|
|
|
Warning("Can't open video codec (%s) %s",
|
|
|
|
video_out_codec->name,
|
|
|
|
av_make_error_string(ret).c_str()
|
|
|
|
);
|
|
|
|
video_out_codec = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
AVDictionaryEntry *e = NULL;
|
|
|
|
while ( (e = av_dict_get(opts, "", e, AV_DICT_IGNORE_SUFFIX)) != NULL ) {
|
|
|
|
Warning( "Encoder Option %s not recognized by ffmpeg codec", e->key);
|
|
|
|
}
|
|
|
|
av_dict_free(&opts);
|
|
|
|
if ( video_out_codec ) break;
|
|
|
|
|
|
|
|
} // end foreach codec
|
|
|
|
|
|
|
|
if ( ! video_out_codec ) {
|
|
|
|
Error("Can't open video codec!");
|
2017-12-05 07:06:42 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
// We allocate and copy in newer ffmpeg, so need to free it
|
|
|
|
avcodec_free_context(&video_out_ctx);
|
2017-12-05 07:06:42 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_ctx = NULL;
|
|
|
|
|
|
|
|
return false;
|
2017-12-09 02:46:02 +08:00
|
|
|
} // end if can't open codec
|
2018-02-28 23:17:16 +08:00
|
|
|
|
2017-12-02 21:12:04 +08:00
|
|
|
Debug(2,"Sucess opening codec");
|
2017-11-15 00:53:29 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
} // end if copying or trasncoding
|
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( !video_out_ctx->codec_tag ) {
|
|
|
|
video_out_ctx->codec_tag =
|
|
|
|
av_codec_get_tag(oc->oformat->codec_tag, video_out_ctx->codec_id );
|
|
|
|
Debug(2, "No codec_tag, setting to h264 ? ");
|
|
|
|
}
|
|
|
|
|
2017-10-27 01:12:56 +08:00
|
|
|
video_out_stream = avformat_new_stream(oc, video_out_codec);
|
2017-11-18 03:30:38 +08:00
|
|
|
if ( ! video_out_stream ) {
|
2017-12-15 04:56:22 +08:00
|
|
|
Error("Unable to create video out stream");
|
|
|
|
return false;
|
2017-06-01 08:44:23 +08:00
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2017-11-28 00:28:36 +08:00
|
|
|
ret = avcodec_parameters_from_context(video_out_stream->codecpar, video_out_ctx);
|
2017-10-27 01:12:56 +08:00
|
|
|
if ( ret < 0 ) {
|
2017-11-13 00:42:34 +08:00
|
|
|
Error("Could not initialize stream parameteres");
|
2017-12-05 07:06:42 +08:00
|
|
|
return false;
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
#else
|
2017-11-28 00:28:36 +08:00
|
|
|
avcodec_copy_context(video_out_stream->codec, video_out_ctx);
|
2017-11-15 04:41:25 +08:00
|
|
|
#endif
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( audio_in_stream ) {
|
|
|
|
audio_in_stream_index = audio_in_stream->index;
|
2017-06-16 01:02:54 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2017-08-24 03:05:44 +08:00
|
|
|
audio_in_ctx = avcodec_alloc_context3(NULL);
|
|
|
|
ret = avcodec_parameters_to_context(audio_in_ctx,
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_in_stream->codecpar);
|
2017-06-03 04:52:52 +08:00
|
|
|
#else
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_in_ctx = audio_in_stream->codec;
|
2017-06-03 04:52:52 +08:00
|
|
|
#endif
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( audio_in_ctx->codec_id != AV_CODEC_ID_AAC ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
static char error_buffer[256];
|
|
|
|
avcodec_string(error_buffer, sizeof(error_buffer), audio_in_ctx,
|
2018-02-28 23:17:16 +08:00
|
|
|
0);
|
2017-08-23 22:37:32 +08:00
|
|
|
Debug(2, "Got something other than AAC (%s)", error_buffer);
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( !setup_resampler() ) {
|
2017-12-05 07:06:42 +08:00
|
|
|
return false;
|
2017-05-17 00:04:56 +08:00
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
} else {
|
2017-08-23 22:37:32 +08:00
|
|
|
Debug(3, "Got AAC");
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_stream =
|
2017-11-27 07:06:12 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
avformat_new_stream(oc, (const AVCodec *)(audio_in_ctx->codec));
|
2017-11-27 07:06:12 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
avformat_new_stream(oc, (AVCodec *)audio_in_ctx->codec);
|
2017-11-27 07:06:12 +08:00
|
|
|
#endif
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( !audio_out_stream ) {
|
2017-05-17 00:04:56 +08:00
|
|
|
Error("Unable to create audio out stream\n");
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_stream = NULL;
|
2017-05-17 00:04:56 +08:00
|
|
|
} else {
|
2017-06-01 08:44:23 +08:00
|
|
|
Debug(2, "setting parameters");
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2017-06-16 01:02:54 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2017-08-24 03:05:44 +08:00
|
|
|
audio_out_ctx = avcodec_alloc_context3(audio_out_codec);
|
2017-08-23 22:37:32 +08:00
|
|
|
// Copy params from instream to ctx
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( (ret = avcodec_parameters_to_context(audio_out_ctx, audio_in_stream->codecpar) ) < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
Error("Unable to copy audio params to ctx %s\n",
|
2018-02-28 23:17:16 +08:00
|
|
|
av_make_error_string(ret).c_str());
|
2017-06-14 05:06:50 +08:00
|
|
|
}
|
2017-12-01 08:26:51 +08:00
|
|
|
// Then from ctx to out_stream
|
|
|
|
ret = avcodec_parameters_from_context(audio_out_stream->codecpar, audio_out_ctx);
|
|
|
|
if ( ret < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
Error("Unable to copy audio params to stream %s\n",
|
2018-02-28 23:17:16 +08:00
|
|
|
av_make_error_string(ret).c_str());
|
2017-06-14 05:06:50 +08:00
|
|
|
}
|
|
|
|
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( !audio_out_ctx->codec_tag ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_ctx->codec_tag = av_codec_get_tag(
|
|
|
|
oc->oformat->codec_tag, audio_in_ctx->codec_id);
|
2017-12-01 08:26:51 +08:00
|
|
|
Debug(2, "Setting audio codec tag to %d", audio_out_ctx->codec_tag);
|
2017-06-14 05:06:50 +08:00
|
|
|
}
|
2017-06-01 08:44:23 +08:00
|
|
|
#else
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_ctx = audio_out_stream->codec;
|
2017-08-24 21:53:40 +08:00
|
|
|
ret = avcodec_copy_context(audio_out_ctx, audio_in_ctx);
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_ctx->codec_tag = 0;
|
2017-06-01 08:44:23 +08:00
|
|
|
#endif
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( ret < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
Error("Unable to copy audio ctx %s\n",
|
2018-02-28 23:17:16 +08:00
|
|
|
av_make_error_string(ret).c_str());
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_stream = NULL;
|
2017-05-17 00:04:56 +08:00
|
|
|
} else {
|
2017-12-01 08:26:51 +08:00
|
|
|
if ( audio_out_ctx->channels > 1 ) {
|
2017-05-17 00:04:56 +08:00
|
|
|
Warning("Audio isn't mono, changing it.");
|
2017-08-23 22:37:32 +08:00
|
|
|
audio_out_ctx->channels = 1;
|
2017-05-17 00:04:56 +08:00
|
|
|
} else {
|
|
|
|
Debug(3, "Audio is mono");
|
|
|
|
}
|
2017-08-23 22:37:32 +08:00
|
|
|
}
|
|
|
|
} // end if audio_out_stream
|
2018-02-28 23:17:16 +08:00
|
|
|
} // end if is AAC
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( audio_out_stream ) {
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
if (oc->oformat->flags & AVFMT_GLOBALHEADER) {
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(56, 35, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_ctx->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
|
2017-11-13 00:42:34 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_ctx->flags |= CODEC_FLAG_GLOBAL_HEADER;
|
2017-11-13 00:42:34 +08:00
|
|
|
#endif
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
|
|
|
}
|
2017-08-23 22:37:32 +08:00
|
|
|
} // end if audio_in_stream
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
/* open the out file, if needed */
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( !(out_format->flags & AVFMT_NOFILE) ) {
|
|
|
|
if ( (ret = avio_open2(&oc->pb, filename, AVIO_FLAG_WRITE, NULL, NULL) ) < 0 ) {
|
2017-09-26 03:28:14 +08:00
|
|
|
Error("Could not open out file '%s': %s\n", filename,
|
2018-02-28 23:17:16 +08:00
|
|
|
av_make_error_string(ret).c_str());
|
2017-09-26 03:28:14 +08:00
|
|
|
return false;
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
zm_dump_stream_format(oc, 0, 0, 1);
|
|
|
|
if (audio_out_stream) zm_dump_stream_format(oc, 1, 0, 1);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2017-08-23 22:37:32 +08:00
|
|
|
AVDictionary *opts = NULL;
|
2017-12-08 23:38:16 +08:00
|
|
|
av_dict_set(&opts, "movflags", "frag_keyframe+empty_moov", 0);
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( (ret = avformat_write_header(oc, &opts)) < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
// if ((ret = avformat_write_header(oc, &opts)) < 0) {
|
2017-05-17 00:04:56 +08:00
|
|
|
Warning("Unable to set movflags to frag_custom+dash+delay_moov");
|
|
|
|
/* Write the stream header, if any. */
|
|
|
|
ret = avformat_write_header(oc, NULL);
|
2017-11-13 00:42:34 +08:00
|
|
|
} else if ( av_dict_count(opts) != 0 ) {
|
2017-05-17 00:04:56 +08:00
|
|
|
Warning("some options not set\n");
|
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( ret < 0 ) {
|
2017-08-23 22:37:32 +08:00
|
|
|
Error("Error occurred when writing out file header to %s: %s\n",
|
2018-02-28 23:17:16 +08:00
|
|
|
filename, av_make_error_string(ret).c_str());
|
2017-09-26 03:28:14 +08:00
|
|
|
return false;
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
if ( opts ) av_dict_free(&opts);
|
2017-12-17 06:08:21 +08:00
|
|
|
|
2017-12-15 04:56:22 +08:00
|
|
|
zm_dump_stream_format(oc, 0, 0, 1);
|
|
|
|
if (audio_out_stream) zm_dump_stream_format(oc, 1, 0, 1);
|
2017-09-26 03:28:14 +08:00
|
|
|
return true;
|
2018-02-28 23:17:16 +08:00
|
|
|
} // end bool VideoStore::open()
|
|
|
|
|
|
|
|
void VideoStore::write_audio_packet( AVPacket &pkt ) {
|
|
|
|
//Debug(2, "writing audio packet pts(%d) dts(%d) duration(%d)", pkt.pts,
|
|
|
|
//pkt.dts, pkt.duration);
|
|
|
|
pkt.pts = audio_next_pts;
|
|
|
|
pkt.dts = audio_next_dts;
|
|
|
|
|
|
|
|
Debug(2, "writing audio packet pts(%d) dts(%d) duration(%d)", pkt.pts, pkt.dts, pkt.duration);
|
|
|
|
if ( pkt.duration > 0 ) {
|
|
|
|
pkt.duration =
|
|
|
|
av_rescale_q(pkt.duration, audio_out_ctx->time_base,
|
|
|
|
audio_out_stream->time_base);
|
|
|
|
}
|
|
|
|
audio_next_pts += pkt.duration;
|
|
|
|
audio_next_dts += pkt.duration;
|
|
|
|
|
|
|
|
Debug(2, "writing audio packet pts(%d) dts(%d) duration(%d)", pkt.pts, pkt.dts, pkt.duration);
|
|
|
|
pkt.stream_index = audio_out_stream->index;
|
|
|
|
av_interleaved_write_frame(oc, &pkt);
|
|
|
|
} // end void VideoStore::Write_audio_packet( AVPacket &pkt )
|
|
|
|
|
|
|
|
VideoStore::~VideoStore() {
|
|
|
|
if ( oc->pb ) {
|
|
|
|
if ( ( video_out_ctx->codec_id != video_in_ctx->codec_id ) || audio_out_codec ) {
|
|
|
|
Debug(2,"Different codecs between in and out");
|
2018-01-05 23:06:18 +08:00
|
|
|
// The codec queues data. We need to send a flush command and out
|
|
|
|
// whatever we get. Failures are not fatal.
|
|
|
|
AVPacket pkt;
|
2017-12-01 08:26:51 +08:00
|
|
|
// WIthout these we seg fault I don't know why.
|
|
|
|
pkt.data = NULL;
|
|
|
|
pkt.size = 0;
|
|
|
|
av_init_packet(&pkt);
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// I got crashes if the codec didn't do DELAY, so let's test for it.
|
|
|
|
if ( video_out_ctx->codec && ( video_out_ctx->codec->capabilities &
|
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
|
|
|
AV_CODEC_CAP_DELAY
|
|
|
|
#else
|
|
|
|
CODEC_CAP_DELAY
|
|
|
|
#endif
|
|
|
|
) ) {
|
|
|
|
|
2017-06-16 01:02:54 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
// Put encoder into flushing mode
|
|
|
|
avcodec_send_frame(video_out_ctx, NULL);
|
|
|
|
while (1) {
|
|
|
|
ret = avcodec_receive_packet(video_out_ctx, &pkt);
|
|
|
|
if (ret < 0) {
|
|
|
|
if (AVERROR_EOF != ret) {
|
|
|
|
Error("ERror encoding audio while flushing (%d) (%s)", ret,
|
|
|
|
av_err2str(ret));
|
|
|
|
}
|
|
|
|
break;
|
2018-01-05 23:06:18 +08:00
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
while (1) {
|
|
|
|
// WIthout these we seg fault I don't know why.
|
|
|
|
pkt.data = NULL;
|
|
|
|
pkt.size = 0;
|
|
|
|
av_init_packet(&pkt);
|
|
|
|
int got_packet = 0;
|
|
|
|
ret = avcodec_encode_video2(video_out_ctx, &pkt, NULL, &got_packet);
|
|
|
|
if ( ret < 0 ) {
|
|
|
|
Error("ERror encoding video while flushing (%d) (%s)", ret, av_err2str(ret));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (!got_packet) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
write_video_packet(pkt);
|
|
|
|
zm_av_packet_unref(&pkt);
|
|
|
|
} // while have buffered frames
|
|
|
|
} // end if have delay capability
|
|
|
|
|
|
|
|
if ( audio_out_codec ) {
|
|
|
|
// The codec queues data. We need to send a flush command and out
|
|
|
|
// whatever we get. Failures are not fatal.
|
|
|
|
AVPacket pkt;
|
|
|
|
// WIthout these we seg fault I don't know why.
|
|
|
|
pkt.data = NULL;
|
|
|
|
pkt.size = 0;
|
|
|
|
av_init_packet(&pkt);
|
|
|
|
|
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
|
|
|
// Put encoder into flushing mode
|
|
|
|
avcodec_send_frame(audio_out_ctx, NULL);
|
|
|
|
while (1) {
|
|
|
|
if ( (ret = avcodec_receive_packet(audio_out_ctx, &pkt) ) < 0 ) {
|
|
|
|
if ( AVERROR_EOF != ret ) {
|
|
|
|
Error("ERror encoding audio while flushing (%d) (%s)", ret, av_err2str(ret));
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
while (1) {
|
|
|
|
pkt.data = NULL;
|
|
|
|
pkt.size = 0;
|
|
|
|
av_init_packet(&pkt);
|
|
|
|
int got_packet = 0;
|
|
|
|
if ( (ret = avcodec_encode_audio2(audio_out_ctx, &pkt, NULL, &got_packet)) < 0 ) {
|
|
|
|
Error("ERror encoding audio while flushing (%d) (%s)", ret, av_err2str(ret));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
Debug(1, "Have audio encoder, need to flush it's out");
|
|
|
|
if (!got_packet) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
write_audio_packet(pkt);
|
|
|
|
zm_av_packet_unref(&pkt);
|
|
|
|
} // while have buffered frames
|
|
|
|
} // end if audio_out_codec
|
|
|
|
} // end if buffers
|
|
|
|
|
|
|
|
// Flush Queues
|
|
|
|
av_interleaved_write_frame(oc, NULL);
|
|
|
|
|
|
|
|
/* Write the trailer before close */
|
|
|
|
if ( int rc = av_write_trailer(oc) ) {
|
|
|
|
Error("Error writing trailer %s", av_err2str(rc));
|
|
|
|
} else {
|
|
|
|
Debug(3, "Sucess Writing trailer");
|
2018-01-05 23:06:18 +08:00
|
|
|
}
|
2018-02-28 23:17:16 +08:00
|
|
|
|
|
|
|
// WHen will be not using a file ?
|
|
|
|
if ( !(out_format->flags & AVFMT_NOFILE) ) {
|
|
|
|
/* Close the out file. */
|
|
|
|
Debug(2, "Closing");
|
|
|
|
if (int rc = avio_close(oc->pb)) {
|
|
|
|
oc->pb = NULL;
|
|
|
|
Error("Error closing avio %s", av_err2str(rc));
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
Debug(3, "Not closing avio because we are not writing to a file.");
|
2018-01-05 23:06:18 +08:00
|
|
|
}
|
2018-02-28 23:17:16 +08:00
|
|
|
} // end if oc->pb
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// I wonder if we should be closing the file first.
|
|
|
|
// I also wonder if we really need to be doing all the ctx
|
|
|
|
// allocation/de-allocation constantly, or whether we can just re-use it.
|
|
|
|
// Just do a file open/close/writeheader/etc.
|
|
|
|
// What if we were only doing audio recording?
|
|
|
|
|
|
|
|
if ( video_out_stream ) {
|
2017-11-28 22:50:09 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
// We allocate and copy in newer ffmpeg, so need to free it
|
|
|
|
avcodec_free_context(&video_in_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
video_in_ctx=NULL;
|
2017-11-28 22:50:09 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
avcodec_close(video_out_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
avcodec_free_context(&video_out_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
video_out_ctx = NULL;
|
|
|
|
Debug(4, "Success freeing video_out_ctx");
|
|
|
|
}
|
|
|
|
if ( audio_out_stream ) {
|
|
|
|
if ( audio_in_codec ) {
|
|
|
|
avcodec_close(audio_in_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
// We allocate and copy in newer ffmpeg, so need to free it
|
|
|
|
avcodec_free_context(&audio_in_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_in_ctx = NULL;
|
|
|
|
audio_in_codec = NULL;
|
|
|
|
} // end if audio_in_codec
|
2017-11-28 22:50:09 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
avcodec_close(audio_out_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
avcodec_free_context(&audio_out_ctx);
|
2017-11-28 22:50:09 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_ctx = NULL;
|
2017-05-17 00:04:56 +08:00
|
|
|
#ifdef HAVE_LIBAVRESAMPLE
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( resample_ctx ) {
|
|
|
|
avresample_close(resample_ctx);
|
|
|
|
avresample_free(&resample_ctx);
|
|
|
|
}
|
|
|
|
if ( out_frame ) {
|
|
|
|
av_frame_free(&out_frame);
|
|
|
|
out_frame = NULL;
|
|
|
|
}
|
|
|
|
if ( converted_in_samples ) {
|
|
|
|
av_free(converted_in_samples);
|
|
|
|
converted_in_samples = NULL;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
}
|
2017-11-13 06:14:21 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( video_in_ctx ) {
|
|
|
|
avcodec_free_context(&video_in_ctx);
|
|
|
|
video_in_ctx = NULL;
|
|
|
|
}
|
|
|
|
if ( video_out_ctx ) {
|
|
|
|
avcodec_close(video_out_ctx);
|
|
|
|
avcodec_free_context(&video_out_ctx);
|
|
|
|
video_out_ctx = NULL;
|
|
|
|
}
|
2017-11-13 06:14:21 +08:00
|
|
|
#endif
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
/* free the stream */
|
|
|
|
avformat_free_context(oc);
|
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
bool VideoStore::setup_resampler() {
|
|
|
|
//I think this is unneccessary, we should be able to just pass in the decoder from the input.
|
2017-05-17 00:04:56 +08:00
|
|
|
#ifdef HAVE_LIBAVRESAMPLE
|
2018-02-28 23:17:16 +08:00
|
|
|
// Newer ffmpeg wants to keep everything separate... so have to lookup our own
|
|
|
|
// decoder, can't reuse the one from the camera.
|
|
|
|
AVCodec *audio_in_codec = avcodec_find_decoder(
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_in_stream->codecpar->codec_id
|
2017-06-03 22:13:39 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_in_ctx->codec_id
|
2017-06-03 22:13:39 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
);
|
|
|
|
if ( (ret = avcodec_open2(audio_in_ctx, audio_in_codec, NULL)) < 0 ) {
|
|
|
|
Error("Can't open in codec!");
|
|
|
|
return false;
|
|
|
|
}
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_codec = avcodec_find_encoder(AV_CODEC_ID_AAC);
|
|
|
|
if ( !audio_out_codec ) {
|
|
|
|
Error("Could not find codec for AAC");
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// Now copy them to the out stream
|
|
|
|
audio_out_stream = avformat_new_stream(oc, audio_out_codec);
|
2017-12-01 11:18:33 +08:00
|
|
|
|
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_ctx = avcodec_alloc_context3(audio_out_codec);
|
|
|
|
if ( !audio_out_ctx ) {
|
|
|
|
Error("could not allocate codec ctx for AAC\n");
|
|
|
|
audio_out_stream = NULL;
|
|
|
|
return false;
|
|
|
|
}
|
2017-12-01 11:18:33 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_ctx = audio_out_stream->codec;
|
2017-12-01 11:18:33 +08:00
|
|
|
#endif
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
/* put sample parameters */
|
|
|
|
audio_out_ctx->bit_rate = audio_in_ctx->bit_rate;
|
|
|
|
audio_out_ctx->sample_rate = audio_in_ctx->sample_rate;
|
|
|
|
audio_out_ctx->sample_fmt = audio_in_ctx->sample_fmt;
|
|
|
|
audio_out_ctx->channels = audio_in_ctx->channels;
|
|
|
|
audio_out_ctx->channel_layout = audio_in_ctx->channel_layout;
|
|
|
|
|
|
|
|
if ( audio_out_codec->supported_samplerates ) {
|
|
|
|
int found = 0;
|
|
|
|
for ( int i=0; audio_out_codec->supported_samplerates[i]; i++ ) {
|
|
|
|
if ( audio_out_ctx->sample_rate ==
|
|
|
|
audio_out_codec->supported_samplerates[i]) {
|
|
|
|
found = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if ( found ) {
|
|
|
|
Debug(4, "Sample rate is good");
|
|
|
|
} else {
|
|
|
|
audio_out_ctx->sample_rate =
|
|
|
|
audio_out_codec->supported_samplerates[0];
|
|
|
|
Debug(1, "Sampel rate is no good, setting to (%d)",
|
|
|
|
audio_out_codec->supported_samplerates[0]);
|
|
|
|
}
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
/* check that the encoder supports s16 pcm in */
|
|
|
|
if ( !check_sample_fmt(audio_out_codec, audio_out_ctx->sample_fmt) ) {
|
|
|
|
Debug(3, "Encoder does not support sample format %s, setting to FLTP",
|
|
|
|
av_get_sample_fmt_name(audio_out_ctx->sample_fmt));
|
|
|
|
audio_out_ctx->sample_fmt = AV_SAMPLE_FMT_FLTP;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// Example code doesn't set the codec tb. I think it just uses whatever defaults
|
|
|
|
//audio_out_ctx->time_base = (AVRational){1, audio_out_ctx->sample_rate};
|
2017-06-03 04:52:52 +08:00
|
|
|
|
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
AVDictionary *opts = NULL;
|
|
|
|
// Needed to allow AAC
|
|
|
|
if ( (ret = av_dict_set(&opts, "strict", "experimental", 0)) < 0 ) {
|
|
|
|
Error("Couldn't set experimental");
|
|
|
|
}
|
|
|
|
ret = avcodec_open2(audio_out_ctx, audio_out_codec, &opts);
|
|
|
|
av_dict_free(&opts);
|
|
|
|
if ( ret < 0 ) {
|
|
|
|
Fatal("could not open codec (%d) (%s)\n", ret, av_make_error_string(ret).c_str());
|
|
|
|
audio_out_codec = NULL;
|
|
|
|
audio_out_ctx = NULL;
|
|
|
|
audio_out_stream = NULL;
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
audio_out_stream->time_base = (AVRational){1, audio_out_ctx->sample_rate};
|
2017-12-09 02:46:02 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( (ret = avcodec_parameters_from_context(
|
|
|
|
audio_out_stream->codecpar,
|
|
|
|
audio_out_ctx)) < 0 ) {
|
|
|
|
Error("Could not initialize stream parameteres");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
//audio_out_stream->codecpar->frame_size = audio_out_ctx->frame_size;
|
|
|
|
//audio_out_stream->codecpar->bit_rate = audio_out_ctx->bit_rate;
|
2017-12-09 02:46:02 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
avcodec_copy_context( audio_out_stream->codec, audio_out_ctx );
|
2017-12-09 02:46:02 +08:00
|
|
|
#endif
|
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
Debug(1,
|
|
|
|
"Audio out context bit_rate (%d) sample_rate(%d) channels(%d) fmt(%d) "
|
|
|
|
"layout(%d) frame_size(%d)",
|
|
|
|
audio_out_ctx->bit_rate, audio_out_ctx->sample_rate,
|
|
|
|
audio_out_ctx->channels, audio_out_ctx->sample_fmt,
|
|
|
|
audio_out_ctx->channel_layout, audio_out_ctx->frame_size);
|
2017-08-10 00:50:46 +08:00
|
|
|
|
2017-12-01 20:26:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
Debug(1,
|
|
|
|
"Audio out stream bit_rate (%d) sample_rate(%d) channels(%d) fmt(%d) "
|
|
|
|
"layout(%d) frame_size(%d)",
|
|
|
|
audio_out_stream->codecpar->bit_rate, audio_out_stream->codecpar->sample_rate,
|
|
|
|
audio_out_stream->codecpar->channels, audio_out_stream->codecpar->format,
|
|
|
|
audio_out_stream->codecpar->channel_layout, audio_out_stream->codecpar->frame_size);
|
2017-12-01 20:26:34 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
Debug(1,
|
|
|
|
"Audio out bit_rate (%d) sample_rate(%d) channels(%d) fmt(%d) "
|
|
|
|
"layout(%d) frame_size(%d)",
|
|
|
|
audio_out_stream->codec->bit_rate, audio_out_stream->codec->sample_rate,
|
|
|
|
audio_out_stream->codec->channels, audio_out_stream->codec->sample_fmt,
|
|
|
|
audio_out_stream->codec->channel_layout, audio_out_stream->codec->frame_size);
|
2017-12-01 20:26:34 +08:00
|
|
|
#endif
|
2017-12-01 11:18:33 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
/** Create a new frame to store the audio samples. */
|
|
|
|
if ( ! in_frame ) {
|
|
|
|
if (!(in_frame = zm_av_frame_alloc())) {
|
|
|
|
Error("Could not allocate in frame");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
/** Create a new frame to store the audio samples. */
|
|
|
|
if ( !(out_frame = zm_av_frame_alloc()) ) {
|
|
|
|
Error("Could not allocate out frame");
|
|
|
|
av_frame_free(&in_frame);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
out_frame->sample_rate = audio_out_ctx->sample_rate;
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// Setup the audio resampler
|
|
|
|
resample_ctx = avresample_alloc_context();
|
|
|
|
if ( !resample_ctx ) {
|
|
|
|
Error("Could not allocate resample ctx\n");
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
uint64_t mono_layout = av_get_channel_layout("mono");
|
|
|
|
// Some formats (i.e. WAV) do not produce the proper channel layout
|
|
|
|
if ( audio_in_ctx->channel_layout == 0 ) {
|
|
|
|
av_opt_set_int(resample_ctx, "in_channel_layout", mono_layout, 0);
|
|
|
|
Debug(1, "Bad channel layout. Need to set it to mono (%d).", mono_layout);
|
|
|
|
} else {
|
|
|
|
Debug(1, "channel layout. set it to mono (%d).", audio_in_ctx->channel_layout);
|
|
|
|
av_opt_set_int(resample_ctx, "in_channel_layout",
|
|
|
|
audio_in_ctx->channel_layout, 0);
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
av_opt_set_int(resample_ctx, "in_sample_fmt", audio_in_ctx->sample_fmt, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "in_sample_rate", audio_in_ctx->sample_rate, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "in_channels", audio_in_ctx->channels, 0);
|
|
|
|
// av_opt_set_int( resample_ctx, "out_channel_layout",
|
|
|
|
// audio_out_ctx->channel_layout, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "out_channel_layout", mono_layout, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "out_sample_fmt",
|
|
|
|
audio_out_ctx->sample_fmt, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "out_sample_rate",
|
|
|
|
audio_out_ctx->sample_rate, 0);
|
|
|
|
av_opt_set_int(resample_ctx, "out_channels",
|
|
|
|
audio_out_ctx->channels, 0);
|
|
|
|
|
|
|
|
if ( (ret = avresample_open(resample_ctx)) < 0 ) {
|
|
|
|
Error("Could not open resample ctx\n");
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
out_frame->nb_samples = audio_out_ctx->frame_size;
|
|
|
|
out_frame->format = audio_out_ctx->sample_fmt;
|
|
|
|
out_frame->channel_layout = audio_out_ctx->channel_layout;
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// The codec gives us the frame size, in samples, we calculate the size of the
|
|
|
|
// samples buffer in bytes
|
|
|
|
unsigned int audioSampleBuffer_size = av_samples_get_buffer_size(
|
|
|
|
NULL, audio_out_ctx->channels, audio_out_ctx->frame_size,
|
|
|
|
audio_out_ctx->sample_fmt, 0);
|
|
|
|
converted_in_samples = (uint8_t *)av_malloc(audioSampleBuffer_size);
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( !converted_in_samples ) {
|
|
|
|
Error("Could not allocate converted in sample pointers\n");
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// Setup the data pointers in the AVFrame
|
|
|
|
if ( avcodec_fill_audio_frame(out_frame, audio_out_ctx->channels,
|
|
|
|
audio_out_ctx->sample_fmt,
|
|
|
|
(const uint8_t *)converted_in_samples,
|
|
|
|
audioSampleBuffer_size, 0) < 0 ) {
|
|
|
|
Error("Could not allocate converted in sample pointers\n");
|
|
|
|
return false;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
return true;
|
2017-05-17 00:04:56 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
Error(
|
|
|
|
"Not built with libavresample library. Cannot do audio conversion to "
|
|
|
|
"AAC");
|
|
|
|
return false;
|
2017-05-17 00:04:56 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
} // end bool VideoStore::setup_resampler()
|
2017-05-17 00:04:56 +08:00
|
|
|
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
int VideoStore::writePacket( ZMPacket *ipkt ) {
|
|
|
|
if ( ipkt->packet.stream_index == video_in_stream_index ) {
|
|
|
|
return writeVideoFramePacket( ipkt );
|
|
|
|
} else if ( ipkt->packet.stream_index == audio_in_stream_index ) {
|
|
|
|
return writeAudioFramePacket( ipkt );
|
|
|
|
}
|
|
|
|
Error("Unknown stream type in packet (%d) out input video stream is (%d) and audio is (%d)",
|
|
|
|
ipkt->packet.stream_index, video_in_stream_index, ( audio_in_stream ? audio_in_stream_index : -1 )
|
|
|
|
);
|
|
|
|
return 0;
|
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
int VideoStore::writeVideoFramePacket( ZMPacket * zm_packet ) {
|
|
|
|
frame_count += 1;
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// if we have to transcode
|
|
|
|
if ( video_out_ctx->codec_id != video_in_ctx->codec_id ) {
|
|
|
|
//Debug(3, "Have encoding video frame count (%d)", frame_count);
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( ! zm_packet->out_frame ) {
|
|
|
|
//Debug(3, "Have no out frame");
|
|
|
|
AVFrame *out_frame = zm_packet->out_frame = zm_av_frame_alloc();
|
|
|
|
if ( ! out_frame ) {
|
|
|
|
Error("Unable to allocate a frame");
|
|
|
|
return 0;
|
|
|
|
}
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVUTIL_VERSION_CHECK(54, 6, 0, 6, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
int codec_imgsize = av_image_get_buffer_size(
|
|
|
|
video_out_ctx->pix_fmt,
|
|
|
|
video_out_ctx->width,
|
|
|
|
video_out_ctx->height, 1);
|
|
|
|
zm_packet->buffer = (uint8_t *)av_malloc(codec_imgsize);
|
|
|
|
av_image_fill_arrays(
|
|
|
|
out_frame->data,
|
|
|
|
out_frame->linesize,
|
|
|
|
zm_packet->buffer,
|
|
|
|
video_out_ctx->pix_fmt,
|
|
|
|
video_out_ctx->width,
|
|
|
|
video_out_ctx->height,
|
|
|
|
1);
|
2017-11-13 00:42:34 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
int codec_imgsize = avpicture_get_size(
|
|
|
|
video_out_ctx->pix_fmt,
|
|
|
|
video_out_ctx->width,
|
|
|
|
video_out_ctx->height);
|
|
|
|
zm_packet->buffer = (uint8_t *)av_malloc(codec_imgsize);
|
|
|
|
avpicture_fill(
|
|
|
|
(AVPicture *)out_frame,
|
|
|
|
zm_packet->buffer,
|
|
|
|
video_out_ctx->pix_fmt,
|
|
|
|
video_out_ctx->width,
|
|
|
|
video_out_ctx->height
|
|
|
|
);
|
2017-11-13 00:42:34 +08:00
|
|
|
#endif
|
2017-08-23 22:37:32 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
out_frame->width = video_out_ctx->width;
|
|
|
|
out_frame->height = video_out_ctx->height;
|
|
|
|
out_frame->format = video_out_ctx->pix_fmt;
|
|
|
|
//out_frame->pkt_duration = 0;
|
|
|
|
|
|
|
|
if ( ! zm_packet->in_frame ) {
|
|
|
|
//Debug(2,"Have no in_frame");
|
|
|
|
if ( zm_packet->packet.size ) {
|
|
|
|
//Debug(2,"Decoding");
|
|
|
|
if ( ! zm_packet->decode( video_in_ctx ) ) {
|
|
|
|
Debug(2, "unable to decode yet.");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
//Go straight to out frame
|
|
|
|
swscale.Convert( zm_packet->in_frame, out_frame );
|
|
|
|
} else if ( zm_packet->image ) {
|
|
|
|
//Debug(2,"Have an image, convert it");
|
|
|
|
//Go straight to out frame
|
|
|
|
swscale.Convert(
|
|
|
|
zm_packet->image,
|
|
|
|
zm_packet->buffer,
|
|
|
|
codec_imgsize,
|
|
|
|
(AVPixelFormat)zm_packet->image->AVPixFormat(),
|
|
|
|
video_out_ctx->pix_fmt,
|
|
|
|
video_out_ctx->width,
|
|
|
|
video_out_ctx->height
|
|
|
|
);
|
|
|
|
|
|
|
|
} else {
|
|
|
|
Error("Have neither in_frame or image!");
|
|
|
|
return 0;
|
|
|
|
} // end if has packet or image
|
|
|
|
} else {
|
|
|
|
// Have in_frame.... may need to convert it to out_frame
|
|
|
|
swscale.Convert(zm_packet->in_frame, zm_packet->out_frame);
|
|
|
|
} // end if no in_frame
|
|
|
|
} // end if no out_frame
|
|
|
|
|
|
|
|
zm_packet->out_frame->coded_picture_number = frame_count;
|
|
|
|
zm_packet->out_frame->display_picture_number = frame_count;
|
|
|
|
zm_packet->out_frame->sample_aspect_ratio = (AVRational){ 0, 1 };
|
2017-12-15 23:47:44 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
zm_packet->out_frame->pkt_duration = 0;
|
2017-12-15 23:47:44 +08:00
|
|
|
#endif
|
2017-12-06 05:16:52 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( ! video_start_pts ) {
|
|
|
|
uint64_t temp = zm_packet->timestamp->tv_sec*(uint64_t)1000000;
|
|
|
|
video_start_pts = temp + zm_packet->timestamp->tv_usec;
|
|
|
|
Debug(2, "No video_lsat_pts, set to (%" PRId64 ") secs(%d=>%" PRId64 ") usecs(%d)",
|
|
|
|
video_start_pts, zm_packet->timestamp->tv_sec, temp, zm_packet->timestamp->tv_usec );
|
|
|
|
Debug(2, "No video_lsat_pts, set to (%" PRId64 ") secs(%d) usecs(%d)",
|
|
|
|
video_start_pts, zm_packet->timestamp->tv_sec, zm_packet->timestamp->tv_usec );
|
|
|
|
zm_packet->out_frame->pts = 0;
|
|
|
|
zm_packet->out_frame->coded_picture_number = 0;
|
|
|
|
} else {
|
|
|
|
uint64_t seconds = ( zm_packet->timestamp->tv_sec*(uint64_t)1000000 + zm_packet->timestamp->tv_usec ) - video_start_pts;
|
|
|
|
zm_packet->out_frame->pts = av_rescale_q( seconds, video_in_stream->time_base, video_out_ctx->time_base);
|
2018-02-23 00:43:04 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
//zm_packet->out_frame->pkt_duration = zm_packet->out_frame->pts - video_start_pts;
|
|
|
|
Debug(2, " Setting pts for frame(%d), set to (%" PRId64 ") from (start %" PRIu64 " - %" PRIu64 " - secs(%d) usecs(%d)",
|
|
|
|
frame_count, zm_packet->out_frame->pts, video_start_pts, seconds, zm_packet->timestamp->tv_sec, zm_packet->timestamp->tv_usec );
|
|
|
|
}
|
|
|
|
if ( zm_packet->keyframe ) {
|
|
|
|
//Debug(2, "Setting keyframe was (%d)", zm_packet->out_frame->key_frame );
|
|
|
|
zm_packet->out_frame->key_frame = 1;
|
|
|
|
//Debug(2, "Setting keyframe (%d)", zm_packet->out_frame->key_frame );
|
|
|
|
} else {
|
|
|
|
Debug(2, "Not Setting keyframe");
|
|
|
|
}
|
2017-11-13 06:02:11 +08:00
|
|
|
|
2017-11-13 00:42:34 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
// Do this to allow the encoder to choose whether to use I/P/B frame
|
|
|
|
zm_packet->out_frame->pict_type = AV_PICTURE_TYPE_NONE;
|
|
|
|
Debug(4, "Sending frame");
|
|
|
|
if ( (ret = avcodec_send_frame(video_out_ctx, zm_packet->out_frame)) < 0 ) {
|
|
|
|
Error("Could not send frame (error '%s')", av_make_error_string(ret).c_str());
|
|
|
|
return -1;
|
|
|
|
}
|
2017-11-15 07:32:37 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
av_init_packet(&opkt);
|
|
|
|
opkt.data = NULL;
|
|
|
|
opkt.size = 0;
|
|
|
|
if ( (ret = avcodec_receive_packet(video_out_ctx, &opkt)) < 0 ) {
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
if ( AVERROR(EAGAIN) == ret ) {
|
|
|
|
// THe codec may need more samples than it has, perfectly valid
|
|
|
|
Debug(3, "Could not recieve packet (error '%s')",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
return 0;
|
|
|
|
} else {
|
|
|
|
Error("Could not recieve packet (error %d = '%s')", ret,
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
}
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
//Debug(2, "Got packet using receive_packet, dts:%" PRId64 ", pts:%" PRId64 ", keyframe:%d", opkt.dts, opkt.pts, opkt.flags & AV_PKT_FLAG_KEY );
|
2017-10-27 01:12:56 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
av_init_packet(&opkt);
|
|
|
|
opkt.data = NULL;
|
|
|
|
opkt.size = 0;
|
|
|
|
int data_present;
|
|
|
|
if ( (ret = avcodec_encode_video2(
|
|
|
|
video_out_ctx, &opkt, zm_packet->out_frame, &data_present)) < 0) {
|
|
|
|
Error("Could not encode frame (error '%s')",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if ( !data_present ) {
|
|
|
|
Debug(2, "Not ready to out a frame yet.");
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 0;
|
|
|
|
}
|
2017-10-27 01:12:56 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
// Need to adjust pts/dts values from codec time to stream time
|
|
|
|
if ( opkt.pts != AV_NOPTS_VALUE)
|
|
|
|
opkt.pts = av_rescale_q(opkt.pts, video_out_ctx->time_base, video_out_stream->time_base);
|
|
|
|
if ( opkt.dts != AV_NOPTS_VALUE)
|
|
|
|
opkt.dts = av_rescale_q(opkt.dts, video_out_ctx->time_base, video_out_stream->time_base);
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
} else {
|
|
|
|
AVPacket *ipkt = &zm_packet->packet;
|
|
|
|
Debug(3, "Doing passthrough, just copy packet");
|
|
|
|
// Just copy it because the codec is the same
|
|
|
|
av_init_packet(&opkt);
|
|
|
|
opkt.data = ipkt->data;
|
|
|
|
opkt.size = ipkt->size;
|
|
|
|
opkt.flags = ipkt->flags;
|
|
|
|
if ( ! video_start_pts ) {
|
|
|
|
// If video has b-frames, pts can be AV_NOPTS_VALUE so use dts intead
|
|
|
|
video_start_pts = ipkt->dts;
|
|
|
|
Debug(2, "No video_lsat_pts, set to (%" PRId64 ")", video_start_pts );
|
|
|
|
opkt.dts = opkt.pts = 0;
|
|
|
|
} else {
|
|
|
|
dumpPacket(ipkt);
|
|
|
|
if ( ipkt->pts != AV_NOPTS_VALUE )
|
|
|
|
opkt.pts = av_rescale_q( ipkt->pts - video_start_pts, video_in_stream->time_base, video_out_stream->time_base );
|
|
|
|
else
|
|
|
|
opkt.pts = 0;
|
|
|
|
|
|
|
|
if ( ipkt->dts != AV_NOPTS_VALUE )
|
|
|
|
opkt.dts = av_rescale_q( ipkt->dts - video_start_pts, video_in_stream->time_base, video_out_stream->time_base );
|
|
|
|
else
|
|
|
|
opkt.dts = 0;
|
|
|
|
Debug(2, "out_stream_time_base(%d/%d) in_stream_time_base(%d/%d) video_start_pts(%" PRId64 ")",
|
|
|
|
video_out_stream->time_base.num, video_out_stream->time_base.den,
|
|
|
|
video_in_stream->time_base.num, video_in_stream->time_base.den,
|
|
|
|
video_start_pts
|
|
|
|
);
|
|
|
|
|
|
|
|
dumpPacket(&opkt);
|
|
|
|
|
|
|
|
opkt.duration = av_rescale_q( opkt.duration, video_in_stream->time_base, video_out_stream->time_base);
|
|
|
|
}
|
|
|
|
}
|
2017-08-10 00:50:46 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
//opkt.duration = 0;
|
2017-12-08 23:38:16 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
write_video_packet( opkt );
|
|
|
|
zm_av_packet_unref(&opkt);
|
2017-10-27 01:12:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
return 1;
|
|
|
|
} // end int VideoStore::writeVideoFramePacket( AVPacket *ipkt )
|
2017-10-27 01:12:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
void VideoStore::write_video_packet( AVPacket &opkt ) {
|
2017-10-27 01:12:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( opkt.dts > opkt.pts ) {
|
|
|
|
Debug(1,
|
|
|
|
"opkt.dts(%" PRId64 ") must be <= opkt.pts(%" PRId64 "). Decompression must happen "
|
|
|
|
"before presentation.",
|
|
|
|
opkt.dts, opkt.pts);
|
|
|
|
opkt.dts = opkt.pts;
|
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
opkt.stream_index = video_out_stream->index;
|
|
|
|
//av_packet_rescale_ts( &opkt, video_out_ctx->time_base, video_out_stream->time_base );
|
2017-12-15 04:56:22 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
dumpPacket(&opkt, "writing video packet");
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( (opkt.data == NULL) || (opkt.size < 1) ) {
|
|
|
|
Warning("%s:%d: Mangled AVPacket: discarding frame", __FILE__, __LINE__);
|
|
|
|
//dumpPacket(&opkt);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
//} else if ((video_next_dts > 0) && (video_next_dts > opkt.dts)) {
|
|
|
|
//Warning("%s:%d: DTS out of order: next:%lld \u226E opkt.dts %lld; discarding frame",
|
|
|
|
//__FILE__, __LINE__, video_next_dts, opkt.dts);
|
|
|
|
//video_next_dts = opkt.dts;
|
|
|
|
//dumpPacket(&opkt);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
} else {
|
|
|
|
if ( (ret = av_interleaved_write_frame(oc, &opkt)) < 0 ) {
|
|
|
|
// There's nothing we can really do if the frame is rejected, just drop it
|
|
|
|
// and get on with the next
|
|
|
|
Warning(
|
|
|
|
"%s:%d: Writing frame [av_interleaved_write_frame()] failed: %s(%d) "
|
|
|
|
" ",
|
|
|
|
__FILE__, __LINE__, av_make_error_string(ret).c_str(), ret);
|
|
|
|
|
|
|
|
//dumpPacket(&safepkt);
|
2017-07-06 01:09:00 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( video_in_stream )
|
|
|
|
zm_dump_codecpar(video_in_stream->codecpar);
|
|
|
|
zm_dump_codecpar(video_out_stream->codecpar);
|
2017-07-06 01:09:00 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
} else {
|
|
|
|
packets_written += 1;
|
|
|
|
}
|
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
} // end void VideoStore::write_video_packet
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
int VideoStore::writeAudioFramePacket(ZMPacket *zm_packet) {
|
|
|
|
Debug(4, "writeAudioFrame");
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
AVPacket *ipkt = &zm_packet->packet;
|
2017-11-13 00:42:34 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( !audio_out_stream ) {
|
|
|
|
Debug(1, "Called writeAudioFramePacket when no audio_out_stream");
|
|
|
|
return 0; // FIXME -ve return codes do not free packet in ffmpeg_camera at
|
|
|
|
// the moment
|
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( audio_out_codec ) {
|
|
|
|
Debug(3, "Have audio codec");
|
2017-05-17 00:04:56 +08:00
|
|
|
#ifdef HAVE_LIBAVRESAMPLE
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2017-06-16 01:02:54 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( (ret = avcodec_send_packet(audio_in_ctx, ipkt)) < 0 ) {
|
|
|
|
Error("avcodec_send_packet fail %s", av_make_error_string(ret).c_str());
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if ( (ret = avcodec_receive_frame(audio_in_ctx, in_frame)) < 0 ) {
|
|
|
|
Error("avcodec_receive_frame fail %s", av_make_error_string(ret).c_str());
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
Debug(2,
|
|
|
|
"Input Frame: samples(%d), format(%d), sample_rate(%d), channel "
|
|
|
|
"layout(%d)",
|
|
|
|
in_frame->nb_samples, in_frame->format,
|
|
|
|
in_frame->sample_rate, in_frame->channel_layout);
|
2017-05-17 00:04:56 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
/**
|
|
|
|
* Decode the audio frame stored in the packet.
|
|
|
|
* The in audio stream decoder is used to do this.
|
|
|
|
* If we are at the end of the file, pass an empty packet to the decoder
|
|
|
|
* to flush it.
|
|
|
|
*/
|
|
|
|
int data_present;
|
|
|
|
if ( (ret = avcodec_decode_audio4(audio_in_ctx, in_frame,
|
|
|
|
&data_present, ipkt)) < 0 ) {
|
|
|
|
Error("Could not decode frame (error '%s')\n",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
dumpPacket(ipkt);
|
|
|
|
av_frame_free(&in_frame);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if ( !data_present ) {
|
|
|
|
Debug(2, "Not ready to transcode a frame yet.");
|
|
|
|
return 0;
|
|
|
|
}
|
2017-06-01 08:44:23 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
int frame_size = out_frame->nb_samples;
|
|
|
|
in_frame->pts = audio_next_pts;
|
|
|
|
|
|
|
|
// Resample the in into the audioSampleBuffer until we proceed the whole
|
|
|
|
// decoded data
|
|
|
|
if ((ret = avresample_convert(resample_ctx, NULL, 0, 0, in_frame->data,
|
|
|
|
0, in_frame->nb_samples)) < 0) {
|
|
|
|
Error("Could not resample frame (error '%s')\n",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
av_frame_unref(in_frame);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
av_frame_unref(in_frame);
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
int samples_available = avresample_available(resample_ctx);
|
|
|
|
if ( samples_available < frame_size ) {
|
|
|
|
Debug(1, "Not enough samples yet (%d)", samples_available);
|
|
|
|
return 0;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
Debug(3, "Output_frame samples (%d)", out_frame->nb_samples);
|
|
|
|
// Read a frame audio data from the resample fifo
|
|
|
|
if ( avresample_read(resample_ctx, out_frame->data, frame_size) != frame_size) {
|
|
|
|
Warning("Error reading resampled audio: ");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
Debug(2,
|
|
|
|
"Frame: samples(%d), format(%d), sample_rate(%d), channel layout(%d)",
|
|
|
|
out_frame->nb_samples, out_frame->format,
|
|
|
|
out_frame->sample_rate, out_frame->channel_layout);
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
av_init_packet(&opkt);
|
|
|
|
Debug(5, "after init packet");
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2017-06-16 01:02:54 +08:00
|
|
|
#if LIBAVCODEC_VERSION_CHECK(57, 64, 0, 64, 0)
|
2018-02-28 23:17:16 +08:00
|
|
|
if ((ret = avcodec_send_frame(audio_out_ctx, out_frame)) < 0) {
|
|
|
|
Error("Could not send frame (error '%s')",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 0;
|
|
|
|
}
|
2017-06-03 04:52:52 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// av_frame_unref( out_frame );
|
2017-08-15 06:22:52 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( (ret = avcodec_receive_packet(audio_out_ctx, &opkt)) < 0 ) {
|
|
|
|
if ( AVERROR(EAGAIN) == ret ) {
|
|
|
|
// THe codec may need more samples than it has, perfectly valid
|
|
|
|
Debug(3, "Could not recieve packet (error '%s')",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
} else {
|
|
|
|
Error("Could not recieve packet (error %d = '%s')", ret,
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
}
|
|
|
|
//zm_av_packet_unref(&opkt);
|
|
|
|
av_frame_unref(in_frame);
|
|
|
|
// av_frame_unref( out_frame );
|
|
|
|
return 0;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( (ret = avcodec_encode_audio2(audio_out_ctx, &opkt, out_frame,
|
|
|
|
&data_present)) < 0) {
|
|
|
|
Error("Could not encode frame (error '%s')",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if ( !data_present ) {
|
|
|
|
Debug(2, "Not ready to out a frame yet.");
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 0;
|
|
|
|
}
|
2017-06-03 04:52:52 +08:00
|
|
|
#endif
|
2017-05-17 00:04:56 +08:00
|
|
|
|
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
opkt.duration = out_frame->nb_samples;
|
|
|
|
} else {
|
|
|
|
av_init_packet(&opkt);
|
|
|
|
Debug(5, "after init packet");
|
|
|
|
opkt.data = ipkt->data;
|
|
|
|
opkt.size = ipkt->size;
|
|
|
|
opkt.duration = ipkt->duration;
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// PTS is difficult, because of the buffering of the audio packets in the
|
|
|
|
// resampler. So we have to do it once we actually have a packet...
|
|
|
|
// audio_last_pts is the pts of ipkt, audio_next_pts is the last pts of the
|
|
|
|
// out
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
// Scale the PTS of the outgoing packet to be the correct time base
|
2017-08-10 00:50:46 +08:00
|
|
|
#if 0
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( ipkt->pts != AV_NOPTS_VALUE ) {
|
|
|
|
if ( !audio_last_pts ) {
|
|
|
|
opkt.pts = 0;
|
|
|
|
Debug(1, "No audio_last_pts");
|
|
|
|
} else {
|
|
|
|
if ( audio_last_pts > ipkt->pts ) {
|
|
|
|
Debug(1, "Resetting audio_start_pts from (%d) to (%d)", audio_last_pts, ipkt->pts);
|
|
|
|
opkt.pts = audio_next_pts + av_rescale_q(ipkt->pts, audio_in_stream->time_base, audio_out_stream->time_base);
|
|
|
|
} else {
|
|
|
|
opkt.pts = audio_next_pts + av_rescale_q(ipkt->pts - audio_last_pts, audio_in_stream->time_base, audio_out_stream->time_base);
|
|
|
|
}
|
|
|
|
Debug(2, "audio opkt.pts = %d from ipkt->pts(%d) - last_pts(%d)", opkt.pts, ipkt->pts, audio_last_pts);
|
|
|
|
}
|
|
|
|
audio_last_pts = ipkt->pts;
|
|
|
|
} else {
|
|
|
|
Debug(2, "opkt.pts = undef");
|
|
|
|
opkt.pts = AV_NOPTS_VALUE;
|
|
|
|
}
|
2017-08-10 00:50:46 +08:00
|
|
|
#else
|
2018-02-28 23:17:16 +08:00
|
|
|
opkt.pts = audio_next_pts;
|
|
|
|
opkt.dts = audio_next_dts;
|
2017-08-15 06:22:52 +08:00
|
|
|
#endif
|
2017-08-10 00:50:46 +08:00
|
|
|
|
|
|
|
#if 0
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( ipkt->dts == AV_NOPTS_VALUE ) {
|
|
|
|
// So if the in has no dts assigned... still need an out dts... so we use cur_dts?
|
2017-05-17 00:04:56 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
if ( audio_last_dts >= audio_in_stream->cur_dts ) {
|
|
|
|
Debug(1, "Resetting audio_last_dts from (%d) to cur_dts (%d)", audio_last_dts, audio_in_stream->cur_dts);
|
|
|
|
opkt.dts = audio_next_dts + av_rescale_q( audio_in_stream->cur_dts, AV_TIME_BASE_Q, audio_out_stream->time_base);
|
|
|
|
} else {
|
|
|
|
opkt.dts = audio_next_dts + av_rescale_q( audio_in_stream->cur_dts - audio_last_dts, AV_TIME_BASE_Q, audio_out_stream->time_base);
|
|
|
|
}
|
|
|
|
audio_last_dts = audio_in_stream->cur_dts;
|
|
|
|
Debug(2, "opkt.dts = %d from video_in_stream->cur_dts(%d) - last_dts(%d)", opkt.dts, audio_in_stream->cur_dts, audio_last_dts);
|
|
|
|
} else {
|
|
|
|
if ( audio_last_dts >= ipkt->dts ) {
|
|
|
|
Debug(1, "Resetting audio_last_dts from (%d) to (%d)", audio_last_dts, ipkt->dts );
|
|
|
|
opkt.dts = audio_next_dts + av_rescale_q(ipkt->dts, audio_in_stream->time_base, audio_out_stream->time_base);
|
|
|
|
} else {
|
|
|
|
opkt.dts = audio_next_dts + av_rescale_q(ipkt->dts - audio_last_dts, audio_in_stream->time_base, audio_out_stream->time_base);
|
|
|
|
Debug(2, "opkt.dts = %d from previous(%d) + ( ipkt->dts(%d) - last_dts(%d) )", opkt.dts, audio_next_dts, ipkt->dts, audio_last_dts );
|
|
|
|
}
|
|
|
|
}
|
2017-05-17 00:04:56 +08:00
|
|
|
}
|
2017-08-10 00:50:46 +08:00
|
|
|
#endif
|
2018-02-28 23:17:16 +08:00
|
|
|
// audio_last_dts = ipkt->dts;
|
|
|
|
if ( opkt.dts > opkt.pts ) {
|
|
|
|
Debug(1,
|
|
|
|
"opkt.dts(%" PRId64 ") must be <= opkt.pts(%" PRId64 "). Decompression must happen "
|
|
|
|
"before presentation.",
|
|
|
|
opkt.dts, opkt.pts);
|
|
|
|
opkt.dts = opkt.pts;
|
|
|
|
}
|
Feature h264 videostorage (#1882)
* Moved writing of configure options from Controller to Model. Fixes #191.
* Initial commit for saving events as videos :)
* Add zm_video.cpp to autotools
* Add zm_video.h to autotools
* Search for MP4V2 header file 3 times: mp4v2/mp4v2.h, mp4v2.h, mp4.h
* Fix serve memory leak
* Few minor code improvements
* Added the ability to override preset, tune, profile and few other improvements
* Correctly write SPS & PPS from x264 encoder headers
* Remove unnessecary SPS & PPS writing code
* Imported missing files from master to feature-h264-videostorage
* Audio support including fixes for dts/pts, split on keyframe and update to mkv extension to prevent ffmpeg problems writing rtsp audio to mp4 containter (header problem)
* Updates to make gcc happy
* Add html5 video control to timeline and event to support mkv playback
* Add zm_videostore.cpp to CMakeLists.txt
* Remove Modern Branch for now
* Fix minor bug
* Option handled added in master, removing duplicate declaration
* Add CaptureandRecord from zm_camera.h
* Putting placeholder in for CaptureAndRecord function
* Removed duplicate code and brackets
* add digest auth file for cmake
Conflicts:
src/CMakeLists.txt
* Add web dir back into Makefile.am
Revert "Removed web from SUBDIRS in Makefile.am"
This reverts commit d9bbcdf3a94cba4d8955fcd03bd965ed2772f34d.
* Add CaptureAndRecord to vlc, still need to make it record
* Resolve SegFault on videostore
* Swap to mp4 container
* mp4 changes
* spaces to tabs, hide video stuff if video writer is turned off
* Make timeline open event.mp4 instead of mkv
* Missed mkv in timeline.js
* Fix some issues from the merge conflict
* Resolve post merge build issues with braces
* Fix whitespace
* Update Jpeg and Video options for passthrough options
* Whitespace fix zm_camera.h
* Fix array mkssing comma
* Add support for Jpeg save options for h264 branch snapshot. Might remove altogether if snapshots not needed
* Update VideoStoreData memory size comment
* Change from config.use_mkv_storage to per monitor option VideoWriter from video branch
* Fix bracket issues post merge
* Clean up comments and add av_free_packet
* Convert from event_directory to event file as per Video branch
* Testing videojs for video playback
* Fixed a missing bracket post merge and also SQL_values now used for EventID and Monitors
* bring recent improvements in ffmpeg capture function into captureandrecord
* Remove pict from writeAudioFramePacket as not used
* Add translate options for h264 Storage options in Monitor and update en_gb file
* Cherry-Pick from iconnor - make it compile on ubuntu 15.04. Which is libav 56.1.0
Conflicts:
src/zm_ffmpeg.cpp
src/zm_remote_camera_rtsp.cpp
Conflicts:
distros/ubuntu1204/changelog
* Clean up videostore code and remove lots of unused code
* proof of concept for dynamic/automatic video rotation using video-js plugin zoomrotate
Conflicts:
web/skins/classic/views/event.php
* removed redundant field in sql query
Conflicts:
web/skins/classic/views/event.php
* local storage of video js plugin
* Beautify!
Make the code somewhat readable.
* added missing videojs.zoomrotate.js file
added missing videojs.zoomrotate.js file
* Typo
added missing "
* Added missing brackets
* fix to display thumbnails when only storing snapshot.jpg
* added control for video playback rate
Conflicts:
web/skins/classic/views/event.php
* dynamically create jpegs from video file for viewing in browser
* fix timeline view for SaveJPEGs monitors (without enabled VideoWriter)
* only expose monitor info which are being used in client
* fix segmentation fault in zma with ubuntu 14.04 and ffmpeg 2.5.8 (gcc 4.8)
when libx264 is not installed
* better way of detecting showing image or video in timeline and event view
instead of Monitor.VideoWriter, Event.DefaultVideo is used, so even if
VideoWriter/SaveJPEG option is changed, a valid image or video will always be
displayed for historical events in both timeline and event view
this also fixes loading videos in timeline view
* Fixes problem of crashing zmc when bad packet arrives causing av_interleaved_write_frame() to return non-zero (-22). Prefilters common packet issues. Add metadata title to generated video file
* Remove syslog.h
* fixed SaveJPEGs are not working
which is caused in errors introduced when merging with master
* Update README.md
* Fix build warnings specific to h264 branch, unused FrameImg, unused ret and int64_t snprintf issues
* Fix PRId64 issue in travis, builds locally fine, but I can see a gcc version issue here
* Fix PRId64 issue in travis, another try
* Try "STDC_FORMAT_MACROS" to see if that helps Travis on gcc 4.6.3
* Revert space removal around PRId64
* video branch ffmpeg 2.9 fixes
ffmpeg 2.9 patched removed SSE2 CPU
* Add FFMPEGInit back
* use webvvt to overlay timestamp (honoring Monitor.LabelFormat) to videos in timeline and event
also fixed bug which prevented seeking in timeline video preview
* ffmpeg 3.0 API build failure fixes
* Update README.md
* merge all the commits from the messed up iconnor_video branch
* fix whitespace
* revert
* whitespace fixes
* spelling fix
* put back some text
* add these back
* fix spelling mistake
* Steal some packet dumping routines from ffmpeg. Convert them to use our logging routines
* add a test and error message if the codec is not h264
* these have been removed in master
* add a view to check auth and just send the video
* add some comments, and dump filename and AVFormatContext on failure to write header
* add the toggle for RecordAudio so that the checkbox works to turn off Audio
* Must init videoStore in constuctor
* more debug and comments, return checking
* Fix dropped part of sql query.
* fix extra else and some whitespace
* Fix missing } from merge that was preventing building.
* fix tabs
* get rid of use of separator, just use \n
* Restore lost fixes for deprecation
* Why are these failing
* Respect record_audio flag when setting up video file so dont try and initiliase mp4 with unsupported audio
* Forgot that I was trying to solve case of stream is true and record_audio
is false.
* Pass swscale_ctx back in to getCachedContext or it will create new
context every frame and leak memory like a mofo.
* Add libx264-dev and libmp4v2-dev to build requires to save hassle of
ensuring they are installed before build.
* Merge my Rotation/Orientation work and fixes for bad h264 streams
* need arpa/inet for reverse lookups
* pull in the new byte range code for viewing videos
* Move our recording flag deeper into closeevent
* add braces and only call closeEvent if there is an event
* deprecate the z_frame_rate stuff which is deprecated in ffmpeg
* remark out some debugging
* fix for video on stream 1
* fix audio_stream to audio_st
* Ignore bad decodes
* fix problems with content-length causing viewing to not work in chrome/android
* change logic of sending file contents to handle an off by one and be more readable
* Some fixes pointed out by Maxim Romanov. Also simply the loading of events to not join the Monitors table
* fix to sql for timeline
* added RecordAudio to sql in README
* Use sub queries instead of joins to fix errors when using new mysql defaults.
* fix sql queries
* Dockerfile to build feature-h264-videostorage
* Must cast codec
* add php-acpu as a dependency
* require php5-acpu
* fix typo
* remove extra /
* Add a line for out-of-tree builds to do api/lib/Cake/bootstrap.php
* delete merge conflict files
* delete merge conflict files
2017-05-16 10:02:48 +08:00
|
|
|
|
2018-02-28 23:17:16 +08:00
|
|
|
//opkt.duration = out_frame ? out_frame->nb_samples : ipkt->duration;
|
|
|
|
// opkt.duration = av_rescale_q(ipkt->duration, audio_in_stream->time_base,
|
|
|
|
// audio_out_stream->time_base);
|
|
|
|
dumpPacket(&opkt);
|
|
|
|
|
|
|
|
// pkt.pos: byte position in stream, -1 if unknown
|
|
|
|
opkt.pos = -1;
|
|
|
|
opkt.stream_index = audio_out_stream->index;
|
|
|
|
audio_next_dts = opkt.dts + opkt.duration;
|
|
|
|
audio_next_pts = opkt.pts + opkt.duration;
|
|
|
|
|
|
|
|
AVPacket safepkt;
|
|
|
|
memcpy(&safepkt, &opkt, sizeof(AVPacket));
|
|
|
|
ret = av_interleaved_write_frame(oc, &opkt);
|
|
|
|
if ( ret != 0 ) {
|
|
|
|
Error("Error writing audio frame packet: %s\n",
|
|
|
|
av_make_error_string(ret).c_str());
|
|
|
|
dumpPacket(&safepkt);
|
|
|
|
} else {
|
|
|
|
Debug(2, "Success writing audio frame");
|
|
|
|
}
|
|
|
|
zm_av_packet_unref(&opkt);
|
|
|
|
return 1;
|
|
|
|
} // end int VideoStore::writeAudioFramePacket( AVPacket *ipkt )
|
|
|
|
|
|
|
|
int VideoStore::write_packets( zm_packetqueue &queue ) {
|
|
|
|
// Need to write out all the frames from the last keyframe?
|
|
|
|
// No... need to write out all frames from when the event began. Due to PreEventFrames, this could be more than since the last keyframe.
|
|
|
|
unsigned int packet_count = 0;
|
|
|
|
ZMPacket *queued_packet;
|
|
|
|
|
|
|
|
while ( ( queued_packet = queue.popPacket() ) ) {
|
|
|
|
AVPacket *avp = queued_packet->av_packet();
|
|
|
|
|
|
|
|
packet_count += 1;
|
|
|
|
//Write the packet to our video store
|
|
|
|
Debug(2, "Writing queued packet stream: %d KEY %d, remaining (%d)",
|
|
|
|
avp->stream_index, avp->flags & AV_PKT_FLAG_KEY, queue.size() );
|
|
|
|
int ret = this->writePacket( queued_packet );
|
|
|
|
if ( ret < 0 ) {
|
|
|
|
//Less than zero and we skipped a frame
|
|
|
|
}
|
|
|
|
delete queued_packet;
|
|
|
|
} // end while packets in the packetqueue
|
|
|
|
Debug(2, "Wrote %d queued packets", packet_count );
|
|
|
|
return packet_count;
|
|
|
|
} // end int VideoStore::write_packets( PacketQueue &queue ) {
|