o V hBQ%@s(ddlZddlZddlmZddlmZmZmZmZm Z ddl Z ddl m Z ddl mZze dd ZWn eefyAd ZYnweZeddZGd d d ZGd ddZdeeefddfddZde jde jde jde jde jde jdefddZde jde jdeeefde jfddZd d ddddd!ed ddd!ef d"ed#ed$ed%ed&ed'ed(ed)eeefd*ed+ed,ed-edeeefd.edee je jeffd/d0Zd"edeeeeeeffd1d2Zd"edefd3d4Z  !   ! dHd5e jd#ed$ed%ed&ed'ed(ed)eeefd6ed7ed+ed,ed-edeeefd8ed9edee je jff"d:d;Z!d5e jdeeeeeeffdd?Z#  @dId"edAe eefdBee eefdCedee je jeeefff dDdEZ$ @dJd"edCedee eeeefeeffdFdGZ%dS)KN)Fraction)DictListOptionalTupleUnion) _load_library) _raise_video_deprecation_warning video_readerTFc@s4eZdZeedZddgZdededdfddZdS)Timebase numerator denominatorrrreturnNcCs||_||_dS)Nr)selfrrrM/var/www/vscode/kcb/lib/python3.10/site-packages/torchvision/io/_video_opt.py__init__s zTimebase.__init__)__name__ __module__ __qualname__int__annotations__ __slots__rrrrrr s r c @s4eZdZeeeeeeeedZgdZdddZdS) VideoMetaData) has_videovideo_timebasevideo_duration video_fps has_audioaudio_timebaseaudio_durationaudio_sample_raterNcCs@d|_tdd|_d|_d|_d|_tdd|_d|_d|_dS)NFrr g) rr rrr r!r"r#r$)rrrrr;s   zVideoMetaData.__init__)rN) rrrboolr floatrrrrrrrr%s  r pts_rangercCsB|d|dkrdkrndStd|dd|ddS)Nrr z=Start pts should not be smaller than end pts, got start pts: z and end pts: ) ValueError)r'rrr _validate_ptsFs "r) vtimebasevfps vduration atimebase asample_rate adurationcCst}|dkr;tt|dt|d|_|dt|d}|dkr;d|_t|||_|dkrHt||_ |dkrtt|dt|d|_ |dt|d}|dkrd|_ t|||_ |dkrt||_ |S)zE Build update VideoMetaData struct with info about the video rr T)rnumelr ritemrr&rrr r"r!r#r$)r*r+r,r-r.r/metatimebaserrr _fill_infoNs$ $   $  r4aframes aframe_ptsaudio_pts_rangec Cs|d|d}}|d}t||dt|}d}|}||dkr.t|d||}|ddkrD||dkrDt|d||}|||ddfS)Nrr )sizer&r) r5r6r7startend num_samplesstep_per_aframes_idxe_idxrrr_align_audio_framesns  r@?rr8filenameseek_frame_marginread_video_stream video_width video_heightvideo_min_dimensionvideo_max_dimensionvideo_pts_rangerread_audio_stream audio_samplesaudio_channelsr"cCstt|t| tjj||d||||||d|d|j|j| | | | d| d| j| j}|\ }}}}}}}}}}t||||||}| dkrQt ||| }|||fS)ab Reads a video from a file, returning both the video frames and the audio frames Args: filename (str): path to the video file seek_frame_margin (double, optional): seeking frame in the stream is imprecise. Thus, when video_start_pts is specified, we seek the pts earlier by seek_frame_margin seconds read_video_stream (int, optional): whether read video stream. If yes, set to 1. Otherwise, 0 video_width/video_height/video_min_dimension/video_max_dimension (int): together decide the size of decoded frames: - When video_width = 0, video_height = 0, video_min_dimension = 0, and video_max_dimension = 0, keep the original frame resolution - When video_width = 0, video_height = 0, video_min_dimension != 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that shorter edge size is video_min_dimension - When video_width = 0, video_height = 0, video_min_dimension = 0, and video_max_dimension != 0, keep the aspect ratio and resize the frame so that longer edge size is video_max_dimension - When video_width = 0, video_height = 0, video_min_dimension != 0, and video_max_dimension != 0, resize the frame so that shorter edge size is video_min_dimension, and longer edge size is video_max_dimension. The aspect ratio may not be preserved - When video_width = 0, video_height != 0, video_min_dimension = 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that frame video_height is $video_height - When video_width != 0, video_height == 0, video_min_dimension = 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that frame video_width is $video_width - When video_width != 0, video_height != 0, video_min_dimension = 0, and video_max_dimension = 0, resize the frame so that frame video_width and video_height are set to $video_width and $video_height, respectively video_pts_range (list(int), optional): the start and end presentation timestamp of video stream video_timebase (Fraction, optional): a Fraction rational number which denotes timebase in video stream read_audio_stream (int, optional): whether read audio stream. If yes, set to 1. Otherwise, 0 audio_samples (int, optional): audio sampling rate audio_channels (int optional): audio channels audio_pts_range (list(int), optional): the start and end presentation timestamp of audio stream audio_timebase (Fraction, optional): a Fraction rational number which denotes time base in audio stream Returns vframes (Tensor[T, H, W, C]): the `T` video frames aframes (Tensor[L, K]): the audio frames, where `L` is the number of points and `K` is the number of audio_channels info (Dict): metadata for the video and audio. Can contain the fields video_fps (float) and audio_fps (int) rr ) r r)torchopsr read_video_from_filerrr4r0r@)rCrDrErFrGrHrIrJrrKrLrMr7r"resultvframes _vframe_ptsr*r+r,r5r6r-r.r/inforrr_read_video_from_file}s:@   rUc Cs~tjj|dddddddddddddddddd}|\ }}}}}}}} } } t|||| | | } |}|}||| fS)z Decode all video- and audio frames in the video. Only pts (presentation timestamp) is returned. The actual frame pixel data is not copied. Thus, it is much faster than read_video(...) rr r8)rNrOr rPr4numpytolist) rCrQ_vframes vframe_ptsr*r+r,_aframesr6r-r.r/rTrrr _read_video_timestamps_from_files4   r[c Cs:ttjj|}|\}}}}}}t||||||}|S)zO Probe a video file and return VideoMetaData with info about the video )r rNrOr probe_video_from_filer4) rCrQr*r+r,r-r.r/rTrrr_probe_video_from_files r] video_datavideo_timebase_numeratorvideo_timebase_denominatoraudio_timebase_numeratoraudio_timebase_denominatorcCstt|t| t|tjs4ttjdddtj|tj d}Wdn1s/wYtj j ||d||||||d|d|| | | | | d| d||}|\ }}}}}}}}}}| dkrmt||| }||fS)a Reads a video from memory, returning both the video frames as the audio frames This function is torchscriptable. Args: video_data (data type could be 1) torch.Tensor, dtype=torch.int8 or 2) python bytes): compressed video content stored in either 1) torch.Tensor 2) python bytes seek_frame_margin (double, optional): seeking frame in the stream is imprecise. Thus, when video_start_pts is specified, we seek the pts earlier by seek_frame_margin seconds read_video_stream (int, optional): whether read video stream. If yes, set to 1. Otherwise, 0 video_width/video_height/video_min_dimension/video_max_dimension (int): together decide the size of decoded frames: - When video_width = 0, video_height = 0, video_min_dimension = 0, and video_max_dimension = 0, keep the original frame resolution - When video_width = 0, video_height = 0, video_min_dimension != 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that shorter edge size is video_min_dimension - When video_width = 0, video_height = 0, video_min_dimension = 0, and video_max_dimension != 0, keep the aspect ratio and resize the frame so that longer edge size is video_max_dimension - When video_width = 0, video_height = 0, video_min_dimension != 0, and video_max_dimension != 0, resize the frame so that shorter edge size is video_min_dimension, and longer edge size is video_max_dimension. The aspect ratio may not be preserved - When video_width = 0, video_height != 0, video_min_dimension = 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that frame video_height is $video_height - When video_width != 0, video_height == 0, video_min_dimension = 0, and video_max_dimension = 0, keep the aspect ratio and resize the frame so that frame video_width is $video_width - When video_width != 0, video_height != 0, video_min_dimension = 0, and video_max_dimension = 0, resize the frame so that frame video_width and video_height are set to $video_width and $video_height, respectively video_pts_range (list(int), optional): the start and end presentation timestamp of video stream video_timebase_numerator / video_timebase_denominator (float, optional): a rational number which denotes timebase in video stream read_audio_stream (int, optional): whether read audio stream. If yes, set to 1. Otherwise, 0 audio_samples (int, optional): audio sampling rate audio_channels (int optional): audio audio_channels audio_pts_range (list(int), optional): the start and end presentation timestamp of audio stream audio_timebase_numerator / audio_timebase_denominator (float, optional): a rational number which denotes time base in audio stream Returns: vframes (Tensor[T, H, W, C]): the `T` video frames aframes (Tensor[L, K]): the audio frames, where `L` is the number of points and `K` is the number of channels ignore The given buffer is not writablemessagedtypeNrr )r r) isinstancerNTensorwarningscatch_warningsfilterwarnings frombufferuint8rOr read_video_from_memoryr0r@)r^rDrErFrGrHrIrJr_r`rKrLrMr7rarbrQrRrSr*r+r,r5r6r-r.r/rrr_read_video_from_memory sBE    rqc Cst|tjs)ttjdddtj|tjd}Wdn1s$wYtjj |dddddddddddddddddd}t |\ }}}}}}}} } } t |||| | | } | }| }||| fS) z Decode all frames in the video. Only pts (presentation timestamp) is returned. The actual frame pixel data is not copied. Thus, read_video_timestamps(...) is much faster than read_video(...) rcrdrergNrr r8)rirNrjrkrlrmrnrorOr rpr r4rVrW) r^rQrXrYr*r+r,rZr6r-r.r/rTrrr"_read_video_timestamps_from_memoryzs@     rrc Cstt|tjs,ttjdddtj|tjd}Wdn1s'wYtj j |}|\}}}}}}t ||||||}|S)zy Probe a video in memory and return VideoMetaData with info about the video This function is torchscriptable rcrdrergN) r rirNrjrkrlrmrnrorOr probe_video_from_memoryr4) r^rQr*r+r,r-r.r/rTrrr_probe_video_from_memorys  rtpts start_ptsend_ptspts_unitc stdur tddkrtdt|}|j}|j}fdd}d}t} |r9t|j j |j j } || }d} t} |rLt|j j |j j } || } t |d|| d| | d\} } }i}|rc|j|d <|rj|j|d <| | |fS) NinfrumThe pts_unit 'pts' gives wrong results and will be removed in a follow-up version. Please use pts_unit 'sec'.cs`}}dkr$ttd|}|tdkr$ttd|}|tdkr,d}||fS)Nsecr ryr8)rmathfloorr&ceil) time_base start_offset end_offsetrwrxrvrrget_ptss  z_read_video..get_ptsrBT)rErJrrKr7r"r audio_fps)r r&rkwarnr]rr!default_timebaserrrrr"rUr r$)rCrvrwrxrTrr!rrJrr7r"rRr5_inforrr _read_videosH      rcsjt|dkr tdt|\}}}|dkr)t|jj|jjfdd|D}|jr/|j nd}||fS)Nrurzr{csg|]}|qSrr).0xvideo_time_baserr sz*_read_video_timestamps..) r rkrr[rrrrrr )rCrxru_rTr rrr_read_video_timestampssr)rAr rrrrrBrr r rrrBrr )rNru)ru)&r|rk fractionsrtypingrrrrrrN extensionr _video_deprecation_warningr _HAS_CPU_VIDEO_DECODER ImportErrorOSError_HAS_VIDEO_OPTrr rrr)rjr4r@strr&r%rUr[r]rqrrrtrrrrrrsT     !           $a#         n +   ?