khBQ%ddlZddlZddlmZddlmZmZmZmZm Z ddl Z ddl m Z ddl mZ e dd ZeZeddZGd d ZGd dZdeeefddfdZde j0de j0de j0de j0de j0de j0defdZde j0de j0deeefde j0fdZdd ddddded dddef ded ed!ed"ed#ed$ed%ed&eeefd'ed(ed)ed*edeeefd+edee j0e j0effd,Zdedeeeeeeffd-Zdedefd.Z dded9edee eeeefeeffd;Z%y#eef$rd ZYwxYw)?N)Fraction)DictListOptionalTupleUnion) _load_library) _raise_video_deprecation_warning video_readerTFc2eZdZeedZddgZdededdfdZy)Timebase numerator denominatorrrreturnNc ||_||_y)Nr)selfrrs U/var/www/teggl/fontify/venv/lib/python3.12/site-packages/torchvision/io/_video_opt.py__init__zTimebase.__init__s #&)__name__ __module__ __qualname__int__annotations__ __slots__rrrrrs7$'Yghqrshtgu v  'r vtimebasevfps vduration atimebase asample_rate adurationct}|jdkDrtt|dj t|dj |_|dj t |dj z }|jdkDr(d|_t |j |z|_|jdkDrt |j |_ |jdkDrtt|dj t|dj |_ |dj t |dj z }|jdkDr(d|_ t |j |z|_ |jdkDrt |j |_ |S)zE Build update VideoMetaData struct with info about the video rr T)r!numelrritemr#r,r"r$r%r'r&r(r))r1r2r3r4r5r6metatimebases r _fill_infor<Ns ?D1&s9Q<+<+<+>'?Yq\EVEVEXAYZQ<$$&y|/@/@/B)CC ?? q !DN"' (8"9H"DD  zz|atyy{+1&s9Q<+<+<+>'?Yq\EVEVEXAYZQ<$$&y|/@/@/B)CC ?? q !DN"' (8"9H"DD a!&|'8'8':!; Kraframes aframe_ptsaudio_pts_rangec|d|d}}|jd}t||z dzt|z }d}|}||dkrt|d|z |z }|ddk7r||dkDrt|d|z |z }|||ddfS)Nrr )sizer,r) r=r>r?startend num_samplesstep_per_aframes_idxe_idxs r_align_audio_framesrInsA 23E,,q/KC%K!O,u[/AAO E E q!!_Q'%/?BCqRC/!*<$<_Q'#-@A 5;> ""r?rrAfilenameseek_frame_marginread_video_stream video_width video_heightvideo_min_dimensionvideo_max_dimensionvideo_pts_ranger#read_audio_stream audio_samplesaudio_channelsr'ctt|t| tjjj ||d||||||d|d|j |j| | | | d| d| j | j}|\ }}}}}}}}}}t||||||}|jdkDr t||| }|||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 r0torchopsr read_video_from_filerrr<r8rI)rLrMrNrOrPrQrRrSr#rTrUrVr?r'resultvframes _vframe_ptsr1r2r3r=r>r4r5r6infos r_read_video_from_filer_}s@%&/"/" YY # # 8 8   ""  ""'F*qwmG[)T9gz9Vbdm iy)\9 UD}}%gz?K GT !!rc4tjjj|dddddddddddddddddd}|\ }}}}}}}} } } t |||| | | } |j j }|j j }||| 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 rA)rXrYr rZr<numpytolist) rLr[_vframes vframe_ptsr1r2r3_aframesr>r4r5r6r^s r _read_video_timestamps_from_filerfs YY # # 8 8                  'F*rxnHj)T9h IWcen iy)\9 UD!!#**,J!!#**,J z4 ''rcttjjj |}|\}}}}}}t ||||||}|S)zO Probe a video file and return VideoMetaData with info about the video )r rXrYr probe_video_from_filer<) rLr[r1r2r3r4r5r6r^s r_probe_video_from_filerisO%& YY # # 9 9( CFEKBItY < iy)\9 UD Kr video_datavideo_timebase_numeratorvideo_timebase_denominatoraudio_timebase_numeratoraudio_timebase_denominatorc0tt|t| t|tjsYt j 5t jddtj|tj}dddtjjj||d||||||d|d|| | | | | d| d||}|\ }}}}}}}}}}|jdkDr t||| }||fS#1swYxYw)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 r0 isinstancerXTensorwarningscatch_warningsfilterwarnings frombufferuint8rYr read_video_from_memoryr8rI)rjrMrNrOrPrQrRrSrkrlrTrUrVr?rmrnr[r\r]r1r2r3r=r>r4r5r6s r_read_video_from_memoryr~ s&J%&/"/" j%,, /  $ $ & I  # #H6X Y))*EKKHJ I YY # # : :  " "'F,qwmG[)T9gz9Vbdm}}%gz?K G C I Is =D  DcFt|tjsYtj5tj ddtj |tj}dddtjjj|dddddddddddddddddd}t|\ }}}}}}}} } } t|||| | | } |jj}|jj}||| fS#1swYxYw) 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(...) rprqrrrtNrr rA)rvrXrwrxryrzr{r|rYr r}r r<rarb) rjr[rcrdr1r2r3rer>r4r5r6r^s r"_read_video_timestamps_from_memoryrzs& j%,, /  $ $ & I  # #H6X Y))*EKKHJ IYY # # : :                  'F*%&qwnHj)T9h IWcen iy)\9 UD!!#**,J!!#**,J z4 ''? I Is =DD ctt|tjsYt j 5t j ddtj|tj}dddtjjj|}|\}}}}}}t||||||}|S#1swYMxYw)zy Probe a video in memory and return VideoMetaData with info about the video This function is torchscriptable rprqrrrtN) r rvrXrwrxryrzr{r|rYr probe_video_from_memoryr<) rjr[r1r2r3r4r5r6r^s r_probe_video_from_memoryrs%& j%,, /  $ $ & I  # #H6X Y))*EKKHJ IYY # # ; ;J GFEKBItY < iy)\9 UD K I Is =CC  start_ptsend_ptspts_unitc Tt tddk(rtjdt |}|j }|j }fd}d}t} |r.get_ptssu  u tzz)q9}*EFGLU5\) 7a)m+D!EF u %JZ''rrKT)rNrSr#rTr?r'r% audio_fps)r r,rxwarnrir"r&default_timebaserr#rrr'r_r%r))rLrrrr^r"r&rrSr#r?r'r\r=_infos ``` r _read_videors+ %&,5  > "( +DII (O%N!$"5"5"?"?ATATA`A`a!.1O%N!$"5"5"?"?ATATA`A`a!.12'%'%GWd E!^^k!33k GU ""rcDt|dk(rtjdt|\}}}|dk(rFt |j j |j j}|Dcgc]}||z }}|jr |jnd}||fScc}w)Nrrr) r rxrrfrr#rrr"r%)rLrr_r^video_time_basexr%s r_read_video_timestampsrs%&5  > 4H=LCD5"4#6#6#@#@$BUBUBaBab,/0qq?"00"&..dI  > 1s1 B)rJr rrrrrKrr r rrrKrr )rNr)r)&rrx fractionsrtypingrrrrrrX extensionr _video_deprecation_warningr _HAS_CPU_VIDEO_DECODER ImportErrorOSError_HAS_VIDEO_OPTrrr!rr0rwr<rIstrr,r+r_rfrir~rrrrrrrrsj 55 %H#.!!(Aq> ' '%%B U38_  || ,,||||  ,,  || @ # \\ #',|| #FKCQTHo # \\ #" $"  './"'./^"^"^"^" ^"  ^"  ^"^"38_^"^"^"^"^"38_^"^" 5<<} 45^"B (s (uT#YS S`=`7a (FS] $  '.$%&''.$%&'!k kkk k  k  kk38_k"k!$kkkk38_k"k !$!k" 5<< %&#k\(( (( 49d3i ./((V *)*04 ;#;#UH_%;#eE8O, -;# ;#  5<<tCJ'7 78 ;#~$)  5cDN* +Xe_ <=O W#"#s I II