o 0iO%@sddlZddlZddlmZddlmZmZmZmZm Z ddl Z ddl m Z ze ddZ Wn eefy;dZ Ynwedd ZGd d d ZGd d d ZdeeefddfddZde jde jde jde jde jde jdefddZde jde jdeeefde jfddZddddddd edddd ef d!ed"ed#ed$ed%ed&ed'ed(eeefd)ed*ed+ed,edeeefd-edee je jeffd.d/Zd!edeeeeeeffd0d1Zd!edefd2d3Z  dGd4e jd"ed#ed$ed%ed&ed'ed(eeefd5ed6ed*ed+ed,edeeefd7ed8edee je jff"d9d:Zd4e jdeeeeeeffd;d<Zd4e jdefd=d>Z  ?dHd!ed@e eefdAee eefdBedee je jeeefff dCdDZ! ?dId!edBedee eeeefeeffdEdFZ"dS)JN)Fraction)DictListOptionalTupleUnion) _load_library video_readerTFc@s4eZdZeedZddgZdededdfddZdS)Timebase numerator denominatorrrreturnNcCs||_||_dS)Nr )selfrrrW/home/ubuntu/SoloSpeech/.venv/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 rrrr r!r"r#)rrrrr9s   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_ptsDs "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%rrrr!r r"r#)r)r*r+r,r-r.metatimebaserrr _fill_infoLs$ $   $  r3aframes aframe_ptsaudio_pts_rangec Cs|d|d}}|d}t||dt|}d}|}||dkr.t|d||}|ddkrD||dkrDt|d||}|||ddfS)Nrr )sizer%r) r4r5r6startend num_samplesstep_per_aframes_idxe_idxrrr_align_audio_framesls  r??rr7filenameseek_frame_marginread_video_stream video_width video_heightvideo_min_dimensionvideo_max_dimensionvideo_pts_rangerread_audio_stream audio_samplesaudio_channelsr!cCst|t| tjj||d||||||d|d|j|j| | | | d| d| j| j}|\ }}}}}}}}}}t||||||}|dkrNt ||| }|||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(torchopsr read_video_from_filerrr3r/r?)rBrCrDrErFrGrHrIrrJrKrLr6r!resultvframes _vframe_ptsr)r*r+r4r5r,r-r.inforrr_read_video_from_file{s8@   rTc 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 r7)rMrNr rOr3numpytolist) rBrP_vframes vframe_ptsr)r*r+_aframesr5r,r-r.rSrrr _read_video_timestamps_from_files4   rZc Cs4tjj|}|\}}}}}}t||||||}|S)zO Probe a video file and return VideoMetaData with info about the video )rMrNr probe_video_from_filer3) rBrPr)r*r+r,r-r.rSrrr_probe_video_from_filesr\ video_datavideo_timebase_numeratorvideo_timebase_denominatoraudio_timebase_numeratoraudio_timebase_denominatorcCst|t| t|tjs1ttjdddtj|tjd}Wdn1s,wYtj j ||d||||||d|d|| | | | | d| d||}|\ }}}}}}}}}}| dkrjt ||| }||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 $video_height, respectively video_pts_range (list(int), optional): the start and end presentation timestamp of video stream video_timebase_numerator / video_timebase_denominator (float, optional): a rational number which denotes timebase in video stream read_audio_stream (int, optional): whether read audio stream. If yes, set to 1. Otherwise, 0 audio_samples (int, optional): audio sampling rate audio_channels (int optional): audio audio_channels audio_pts_range (list(int), optional): the start and end presentation timestamp of audio stream audio_timebase_numerator / audio_timebase_denominator (float, optional): a rational number which denotes time base in audio stream Returns: vframes (Tensor[T, H, W, C]): the `T` video frames aframes (Tensor[L, K]): the audio frames, where `L` is the number of points and `K` is the number of channels ignore The given buffer is not writablemessagedtypeNrr )r( isinstancerMTensorwarningscatch_warningsfilterwarnings frombufferuint8rNr read_video_from_memoryr/r?)r]rCrDrErFrGrHrIr^r_rJrKrLr6r`rarPrQrRr)r*r+r4r5r,r-r.rrr_read_video_from_memorys@E    rpc Cst|tjs)ttjdddtj|tjd}Wdn1s$wYtjj |dddddddddddddddddd}|\ }}}}}}}} } } 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(...) rbrcrdrfNrr r7)rhrMrirjrkrlrmrnrNr ror3rUrV) r]rPrWrXr)r*r+rYr5r,r-r.rSrrr"_read_video_timestamps_from_memoryus>     rqc Cst|tjs)ttjdddtj|tjd}Wdn1s$wYtjj |}|\}}}}}}t ||||||}|S)zy Probe a video in memory and return VideoMetaData with info about the video This function is torchscriptable rbrcrdrfN) rhrMrirjrkrlrmrnrNr probe_video_from_memoryr3) r]rPr)r*r+r,r-r.rSrrr_probe_video_from_memorys  rspts start_ptsend_ptspts_unitc sdurtddkrtdt|}|j}|j}fdd}d}t} |r6t|jj |jj } || }d} t} |rIt|j j |j j } || } t |d|| d| | d\} } }i}|r`|j |d <|rg|j|d <| | |fS) NinfrtmThe pts_unit 'pts' gives wrong results and will be removed in a follow-up version. Please use pts_unit 'sec'.cs`}}dkr$ttd|}|tdkr$ttd|}|tdkr,d}||fS)Nsecr rxr7)rmathfloorr%ceil) time_base start_offset end_offsetrvrwrurrget_ptss  z_read_video..get_ptsrAT)rDrIrrJr6r!r audio_fps)r%rjwarnr\rr default_timebaserrrrr!rTrr#)rBrurvrwrSrr rrIrr6r!rQr4_inforrr _read_videosF      rcsd|dkr tdt|\}}}|dkr&t|jj|jjfdd|D}|jr,|jnd}||fS)Nrtryrzcsg|]}|qSrr).0xvideo_time_baserr sz*_read_video_timestamps..) rjrrZrrrrrr)rBrwrt_rSrrrr_read_video_timestampssr)r@r rrrrrArr r rrrArr )rNrt)rt)#r{rj fractionsrtypingrrrrrrM extensionr _HAS_VIDEO_OPT ImportErrorOSErrorrr rrr(rir3r?strr%r$rTrZr\rprqrsrrrrrrsP    !           $`#         m *   >