450 lines
18 KiB
Python
Executable file
450 lines
18 KiB
Python
Executable file
#! /usr/bin/env python3
|
|
# encoding: utf-8
|
|
|
|
import json
|
|
import os
|
|
import over
|
|
import pathlib
|
|
import re
|
|
import tempfile
|
|
import time
|
|
import version
|
|
|
|
from aux import parse_fps, to_Path, update_cfg_context
|
|
|
|
Command = over.cmd.Command
|
|
|
|
# --------------------------------------------------
|
|
X264_BANNED_PIXFMTS = {'bgr24', 'yuv422p'}
|
|
|
|
# --------------------------------------------------
|
|
|
|
# see doc/command_assembler.png
|
|
command = over.types.ndict()
|
|
command.identify = Command('ffprobe', '-v', 'quiet', '-print_format', 'json', '-show_format', '-show_streams', 'INFILE')
|
|
command.normalize_prepass = Command('ffmpeg', '-i', 'INFILE', '-filter:a', 'volumedetect', '-f', 'null', '/dev/null')
|
|
command.encode_generic = Command('ffmpeg', '-i', 'INFILE', 'MAP', 'VIDEO', 'AUDIO', '-sn', 'OUTFILE')
|
|
command.sub_vorbis = Command('-codec:a', 'libvorbis', '-qscale:a', 'QUALITY', 'NORMALIZE')
|
|
command.sub_pcm = Command('-codec:a', 'pcm_s16le', 'NORMALIZE')
|
|
command.sub_theora = Command('-codec:v', 'libtheora', '-qscale:v', 'QUALITY', 'VFILTER')
|
|
command.sub_x264 = Command('PIXFMT', '-codec:v', 'libx264', '-preset', 'PRESET', '-crf', 'QUALITY', '-profile:v', 'high', '-level', '4.2', 'VFILTER')
|
|
command.sub_x265 = Command('PIXFMT', '-codec:v', 'libx265', '-preset', 'PRESET', '-crf', 'QUALITY', 'VFILTER')
|
|
command.sub_normalize = Command('-filter:a', 'VOLUME')
|
|
command.sub_vfilter = Command('-filter:v', 'ARGS')
|
|
command.force_yuv420p = Command('-pix_fmt', 'yuv420p')
|
|
command.sub_copy_audio = Command('-codec:a', 'copy')
|
|
command.sub_copy_video = Command('-codec:v', 'copy')
|
|
|
|
# --------------------------------------------------
|
|
|
|
if __name__ == '__main__':
|
|
main = over.app.Main('over-video', version.str, 'AO-JSL', features={'config_file': True})
|
|
main.add_option('audio', 'Audio codec to use, either <M>vorbis<.>, <M>pcm<.>, <M>copy<.> or <M>drop<.>.', str, ['vorbis'], abbr='a', count=1)
|
|
main.add_option('audio-quality', 'Audio encoding quality with <M>-1<.> being the worst and <M>10<.> being the best.', float, [4], abbr='q', count=1)
|
|
main.add_option('video', 'Video codec to use, either <M>x265<.>, <M>x264<.>, <M>theora<.>, <M>copy<.> or <M>drop<.>.', str, ['x264'], abbr='v', count=1)
|
|
main.add_option('video-preset', 'Video encoding preset, if supported by the selected encoder.', str, ['slow'], abbr='P', count=1)
|
|
main.add_option('video-quality', 'Video encoding quality (CRF). Use <M>0<.>-<M>10<.> for Theora (<M>0<.> being the lowest, <M>5<.>-<M>7<.> is generally watchable) and <M>0<.>-<M>51<.> for x264/5 (<M>0<.> being lossless, <M>18<.>-<M>28<.> is reasonable).', float, [22], abbr='Q', count=1)
|
|
main.add_option('context', 'Use .over-video file in CWD, if available, to remember encoding parameters per-directory.', bool, [True], abbr='C')
|
|
main.add_option('normalize', 'Normalize the audio track.', bool, [True], abbr='n')
|
|
main.add_option('normalize-target', 'Target mean volume to target.', float, [-20.0], count=1)
|
|
main.add_option('normalize-override', 'Volume correction to use instead of computing the required value in a (lengthy) pre-pass.', float, [0.0], abbr='N', count=1, in_cfg_file=False)
|
|
main.add_option('ffmpeg-vfilter', 'Raw ffmpeg -filter:v options, e.g. "<M>scale=1280:trunc(ow/a/2)*2,transpose=dir=1<.>"', str, [''], abbr='F', count=1)
|
|
main.add_option('ffmpeg-map', 'Raw ffmpeg <c>-map<.> options, e.g. <W>--<g>map<.> <M>0:1<.> <W>--<g>map<.> <M>0:2<.>. This is a drop-in fix until we get proper stream selection.', str, abbr='M', overwrite=False, count=1)
|
|
main.add_option('move-source', 'Move source file to this directory after conversion. Use an empty string to disable.', str, ['processed'], count=1)
|
|
main.add_option('dump-commands', 'Print ffmpeg commands that would be executed. If <W>--<g>normalize<.> is in effect, the normalization pre-pass will still be performed so that the proper volume correction can be computed.', bool, [False], abbr='D', in_cfg_file=False)
|
|
main.add_option('probe', 'Print the raw JSON output of ffprobe and exit.', bool, [False], abbr='p', in_cfg_file=False)
|
|
main.add_option('armed', 'Perform the suggested action.', bool, [False], abbr='A', in_cfg_file=False)
|
|
|
|
main.add_doc('Description', ['Over-Video is a simple video converter.'])
|
|
main.add_doc('Good encoder settings', ['<W>x264<.>: <W>--<g>video<.> <M>x264<.> <W>--<g>video-preset<.> <M>slow<.> <W>--<g>video-quality<.> <M>22<.>', '<W>x265<.>: <W>--<g>video<.> <M>x265<.> <W>--<g>video-preset<.> <M>medium<.> <W>--<g>video-quality<.> <M>20<.>'])
|
|
|
|
main.setup()
|
|
|
|
# --------------------------------------------------
|
|
# cfg checks
|
|
|
|
files = over.types.ndict()
|
|
audio_words = []
|
|
video_words = []
|
|
files.container = 'mkv'
|
|
|
|
if main.cfg.context:
|
|
update_cfg_context(main, ["context", "armed", "probe", "dump-commands", "ffmpeg-map", "normalize-override"])
|
|
|
|
if main.cfg.audio in ('copy', 'drop'):
|
|
audio_words.append('<c>%s<.>' %(main.cfg.audio))
|
|
else:
|
|
audio_words.append('<g>codec<.>=<M>%s<.>' %(main.cfg.audio))
|
|
|
|
if main.cfg.audio == 'vorbis':
|
|
audio_words.append('<g>quality<.>=<M>%.1f<.>' %(main.cfg.audio_quality))
|
|
|
|
if main.cfg.normalize_override != 0:
|
|
audio_words.append('<g>adjust_volume<.>=<M>%.1f dB<.>' %(main.cfg.normalize_override))
|
|
elif main.cfg.normalize:
|
|
audio_words.append('<g>normalize<.>=<M>%.1f dB<.>' %(main.cfg.normalize_target))
|
|
|
|
if main.cfg.video in ('copy', 'drop'):
|
|
video_words.append('<c>%s<.>' %(main.cfg.video))
|
|
|
|
else:
|
|
video_words.append('<g>codec<.>=<M>%s<.>' %(main.cfg.video))
|
|
video_words.append('<g>quality<.>=<M>%.1f<.>' %(main.cfg.video_quality))
|
|
|
|
if main.cfg.video_preset and main.cfg.video in ('x264', 'x265'):
|
|
video_words.append('<g>preset<.>=<M>%s<.>' %(main.cfg.video_preset))
|
|
|
|
if main.cfg.ffmpeg_vfilter:
|
|
video_words.append('<g>vfilter<.>=<M>%s<.>' %(main.cfg.ffmpeg_vfilter))
|
|
|
|
if main.cfg.video == 'drop':
|
|
if main.cfg.audio == 'pcm':
|
|
files.container = 'wav'
|
|
elif main.cfg.audio == 'vorbis':
|
|
files.container = 'ogg'
|
|
|
|
main.print('settings', main.print.tl.start, end=':\n')
|
|
main.print('audio: %s' %(', '.join(audio_words)))
|
|
main.print('video: %s' %(', '.join(video_words)))
|
|
main.print('container: <g>type<.>=<M>%s<.>' %(files.container))
|
|
|
|
if main.cfg.move_source:
|
|
main.print('move source files to <W>%s<.>/' %(main.cfg.move_source))
|
|
|
|
if main.cfg.audio not in ('drop', 'copy', 'pcm', 'vorbis'):
|
|
raise ValueError('unknown audio codec: %s' %(main.cfg.audio))
|
|
|
|
if main.cfg.video not in ('drop', 'copy', 'theora', 'x264', 'x265'):
|
|
raise ValueError('unknown video codec: %s' %(main.cfg.video))
|
|
|
|
if not main.targets:
|
|
main.print('no files specified', main.print.tl.warn)
|
|
|
|
for tgt in main.targets:
|
|
print()
|
|
|
|
files.infile = to_Path(tgt)
|
|
files.tmpfile = to_Path(tempfile.mktemp(suffix='.' + files.container, dir='.'))
|
|
files.outfile = files.infile.parent / (str(files.infile.stem) + '.' + files.container)
|
|
files.move_infile_to = to_Path(main.cfg.move_source) / files.infile.name if main.cfg.move_source else None
|
|
|
|
if not os.path.exists(tgt) or os.path.isdir(tgt):
|
|
main.print('target <y>%s<.> <r>is not a readable file<.>, skipping' %(tgt), main.print.tl.fail)
|
|
continue
|
|
|
|
original_filesize = over.text.Unit(files.infile.stat().st_size, 'o')
|
|
main.print('processing <W>%s<.> (%s)' %(tgt, original_filesize), main.print.tl.start)
|
|
|
|
# --------------------------------------------------
|
|
# identify the input file
|
|
|
|
command.identify.reset()
|
|
command.identify.INFILE = 'file:' + str(files.infile)
|
|
command.identify.run()
|
|
identify_raw = command.identify.get_all_output().decode('utf-8')
|
|
identify_dict = json.loads(identify_raw)
|
|
|
|
if main.cfg.probe:
|
|
print(identify_raw)
|
|
continue
|
|
|
|
info = over.types.ndict()
|
|
|
|
try:
|
|
info.duration = float(identify_dict['format']['duration'])
|
|
|
|
video_streams = [s for s in identify_dict['streams'] if s['codec_type'] == 'video']
|
|
audio_streams = [s for s in identify_dict['streams'] if s['codec_type'] == 'audio']
|
|
|
|
amount_vs = len(video_streams)
|
|
amount_as = len(audio_streams)
|
|
|
|
if amount_vs > 1:
|
|
main.print('detected <r>%d<.> video streams, picking the "best" one (see man 1 ffmpeg, section STREAM SELECTION)' %(amount_vs), main.print.tl.warn)
|
|
|
|
if amount_as > 1:
|
|
main.print('detected <y>%d<.> audio streams, picking the "best" one (see man 1 ffmpeg, section STREAM SELECTION)' %(amount_as), main.print.tl.warn)
|
|
|
|
if video_streams:
|
|
# ffmpeg picks the stream with the highest pixel count and then the lowest index
|
|
video_streams.sort(key=lambda s: s['width'] * s['height'], reverse=True)
|
|
video = video_streams[0]
|
|
info.video_codec = video['codec_name']
|
|
info.video_size_x = video['width']
|
|
info.video_size_y = video['height']
|
|
info.video_fps = over.text.Unit(parse_fps(video['r_frame_rate']), 'Hz')
|
|
|
|
if 'bit_rate' in video:
|
|
info.video_bitrate = over.text.Unit(video['bit_rate'], 'b/s')
|
|
elif 'tags' in video and 'BPS' in video['tags']:
|
|
info.video_bitrate = over.text.Unit(int(video['tags']['BPS']), 'b/s')
|
|
else:
|
|
info.video_bitrate = '<R>??<.>'
|
|
info.pixel_fmt = video['pix_fmt']
|
|
else:
|
|
info.video_fps = 30 # faked for progress bars
|
|
|
|
if audio_streams:
|
|
# ffmpeg picks the stream with the most channels and then the lowest index
|
|
audio_streams.sort(key=lambda s: s['channels'], reverse=True)
|
|
audio = audio_streams[0]
|
|
info.audio_codec = audio['codec_name']
|
|
info.audio_channels = audio['channels']
|
|
info.audio_samplerate = over.text.Unit(audio['sample_rate'], 'Hz')
|
|
info.audio_language = audio['tags']['language'] if 'tags' in audio and 'language' in audio['tags'] else 'und'
|
|
info.audio_bitrate = over.text.Unit(audio['bit_rate'], 'b/s') if 'bit_rate' in audio else '<R>??<.>'
|
|
|
|
except:
|
|
main.print('exception while reading identify_dict, dump follows', main.print.tl.fail)
|
|
print(identify_dict)
|
|
raise
|
|
|
|
if video_streams:
|
|
main.print('<m>video<.>: size=<M>%d<.>x<M>%d<.> px, framerate=%s, codec=%s, bitrate=%s' %(info.video_size_x, info.video_size_y, info.video_fps, info.video_codec, info.video_bitrate))
|
|
else:
|
|
main.print('<m>video<.>: <y>None<.>', main.print.tl.warn)
|
|
|
|
if audio_streams:
|
|
main.print('<c>audio<.>: channels=<C>%d<.>, samplerate=%s, codec=%s, bitrate=%s, language=%s' %(info.audio_channels, info.audio_samplerate, info.audio_codec, info.audio_bitrate, info.audio_language))
|
|
else:
|
|
main.print('<c>audio<.>: <y>None<.>', main.print.tl.warn)
|
|
|
|
# --------------------------------------------------
|
|
# normalization pre-pass
|
|
|
|
if audio_streams and main.cfg.normalize and (main.cfg.armed or main.cfg.dump_commands) and (not main.cfg.audio == 'drop'):
|
|
if main.cfg.normalize_override == 0.0:
|
|
main.print('running normalization pre-pass')
|
|
|
|
command.normalize_prepass.reset()
|
|
command.normalize_prepass.INFILE = 'file:' + str(files.infile)
|
|
command.normalize_prepass.run(stderr=True)
|
|
|
|
pb = over.text.ProgressBar(
|
|
"§%a [§=a>§ A] §sa (Trem=§TA)",
|
|
{
|
|
"a": {
|
|
"unit": "s",
|
|
"top": info.duration,
|
|
"precision": 1,
|
|
"min_width_raw": 0,
|
|
"min_width_rate": 0,
|
|
"min_width_time": 0
|
|
}
|
|
}
|
|
)
|
|
|
|
pb.render()
|
|
output_buffer = []
|
|
|
|
while True:
|
|
time.sleep(.25)
|
|
|
|
out = command.normalize_prepass.get_output()
|
|
|
|
if out:
|
|
output_buffer.append(out)
|
|
|
|
if b'frame=' in out:
|
|
frame_id = re.findall(b'frame= *(\d+) ', out)[0]
|
|
pb.set("a", int(frame_id) / info.video_fps.value)
|
|
pb.render()
|
|
|
|
elif out is None:
|
|
break
|
|
|
|
pb.end()
|
|
|
|
output = b''.join(output_buffer)
|
|
|
|
if b'mean_volume: ' in output:
|
|
info.mean_volume = float(re.findall(b'mean_volume: (-?\d+\.\d+) dB', output)[0])
|
|
info.max_correction = -float(re.findall(b'max_volume: (-?\d+\.\d+) dB', output)[0])
|
|
info.volume_correction = main.cfg.normalize_target - info.mean_volume
|
|
else:
|
|
main.print('<r>unexpected ffmpeg output<.>, dump follows', main.print.tl.fail, suffix=':\n')
|
|
print(output.decode('utf-8'))
|
|
raise RuntimeError
|
|
|
|
main.print('detected volume %.1f dB, correction %.1f dB, max. correction %.1f dB' %(info.mean_volume, info.volume_correction, info.max_correction))
|
|
|
|
if info.volume_correction > info.max_correction:
|
|
d = info.volume_correction - info.max_correction
|
|
main.print("suggested correction is %.1f dB above the stream's maximum and will cause clipping" %(d), main.print.tl.warn)
|
|
|
|
else:
|
|
info.volume_correction = main.cfg.normalize_override
|
|
main.print('using user-supplied volume correction <M>%.1f dB<.>' %(info.volume_correction))
|
|
|
|
info.normalize_command = command.sub_normalize
|
|
info.normalize_command.reset()
|
|
info.normalize_command.VOLUME = 'volume=%.2fdB' %(info.volume_correction)
|
|
|
|
else:
|
|
info.normalize_command = None
|
|
|
|
# --------------------------------------------------
|
|
# main command assembly
|
|
|
|
encode_cmd = command.encode_generic
|
|
encode_cmd.reset()
|
|
|
|
encode_cmd.INFILE = 'file:' + str(files.infile)
|
|
encode_cmd.OUTFILE = files.tmpfile
|
|
|
|
if main.cfg.audio == 'copy':
|
|
encode_cmd.AUDIO = command.sub_copy_audio
|
|
elif main.cfg.audio == 'drop':
|
|
encode_cmd.AUDIO = '-an'
|
|
elif main.cfg.audio == 'pcm':
|
|
command.sub_pcm.reset()
|
|
command.sub_pcm.NORMALIZE = info.normalize_command
|
|
|
|
encode_cmd.AUDIO = command.sub_pcm
|
|
elif main.cfg.audio == 'vorbis':
|
|
command.sub_vorbis.reset()
|
|
command.sub_vorbis.QUALITY = main.cfg.audio_quality
|
|
command.sub_vorbis.NORMALIZE = info.normalize_command
|
|
|
|
encode_cmd.AUDIO = command.sub_vorbis
|
|
|
|
if main.cfg.ffmpeg_vfilter:
|
|
info.vfilter_command = command.sub_vfilter
|
|
info.vfilter_command.reset()
|
|
info.vfilter_command.ARGS = main.cfg.ffmpeg_vfilter
|
|
else:
|
|
info.vfilter_command = None
|
|
|
|
if main.cfg.ffmpeg_map:
|
|
info.map_command = []
|
|
|
|
for m in main.cfg.ffmpeg_map:
|
|
info.map_command.append("-map")
|
|
info.map_command.append(m)
|
|
else:
|
|
info.map_command = None
|
|
|
|
encode_cmd.MAP = info.map_command
|
|
|
|
if main.cfg.video == 'copy':
|
|
encode_cmd.VIDEO = command.sub_copy_video
|
|
elif main.cfg.video == 'drop':
|
|
encode_cmd.VIDEO = '-vn'
|
|
elif main.cfg.video == 'theora':
|
|
command.sub_theora.reset()
|
|
command.sub_theora.QUALITY = main.cfg.video_quality
|
|
command.sub_theora.VFILTER = info.vfilter_command
|
|
|
|
encode_cmd.VIDEO = command.sub_theora
|
|
elif main.cfg.video == 'x264':
|
|
command.sub_x264.reset()
|
|
command.sub_x264.QUALITY = main.cfg.video_quality
|
|
command.sub_x264.PRESET = main.cfg.video_preset
|
|
command.sub_x264.VFILTER = info.vfilter_command
|
|
|
|
if info.pixel_fmt in X264_BANNED_PIXFMTS:
|
|
main.print('source pixel format <r>%s<.> is incompatible with x264, forcing <y>yuv420p<.>' %(info.pixel_fmt), main.print.tl.warn)
|
|
command.sub_x264.PIXFMT = command.force_yuv420p
|
|
else:
|
|
command.sub_x264.PIXFMT = None
|
|
|
|
encode_cmd.VIDEO = command.sub_x264
|
|
elif main.cfg.video == 'x265':
|
|
command.sub_x265.reset()
|
|
command.sub_x265.QUALITY = main.cfg.video_quality
|
|
command.sub_x265.PRESET = main.cfg.video_preset
|
|
command.sub_x265.VFILTER = info.vfilter_command
|
|
command.sub_x265.PIXFMT = None
|
|
|
|
encode_cmd.VIDEO = command.sub_x265
|
|
|
|
# --------------------------------------------------
|
|
# run the command iff armed
|
|
|
|
if main.cfg.dump_commands or main.cfg.armed:
|
|
cmd = ' '.join(encode_cmd.dump(pretty=True))
|
|
|
|
if main.cfg.armed:
|
|
main.print('executing <W>%s<.>' %(cmd), main.print.tl.start)
|
|
else:
|
|
main.print('will execute <W>%s<.>' %(cmd))
|
|
else:
|
|
main.print('will encode into <W>%s<.>' %(files.tmpfile))
|
|
|
|
if main.cfg.armed:
|
|
pb = over.text.ProgressBar(
|
|
"§%f §rs [§=f>§ F] §sf (§ss) (Sest=§zs, Trem=§TF)",
|
|
{
|
|
"f": {
|
|
"unit": "f",
|
|
"top": int(info.video_fps.value * info.duration),
|
|
"precision": 1,
|
|
"min_width_rate": 9
|
|
},
|
|
"s": {
|
|
"unit": "o", # octets are cool
|
|
"top": None, # size is unknown at the start but will be estimated during updates
|
|
"precision": 1,
|
|
"min_width_raw": 9,
|
|
"min_width_rate": 11
|
|
}
|
|
}
|
|
)
|
|
|
|
encode_cmd.run(stderr=True)
|
|
|
|
while True:
|
|
time.sleep(.25)
|
|
|
|
out = encode_cmd.get_output()
|
|
|
|
if out:
|
|
if b'frame=' in out:
|
|
frame_id = re.findall(b'frame= *(\d+) ', out)[0]
|
|
pb.set("f", int(frame_id))
|
|
|
|
elif out is None:
|
|
break
|
|
|
|
try:
|
|
pb.set("s", files.tmpfile.stat().st_size)
|
|
except FileNotFoundError: # a race condition with ffmpeg
|
|
pass
|
|
|
|
pb.render()
|
|
|
|
new_filesize = over.text.Unit(files.tmpfile.stat().st_size, 'o')
|
|
|
|
pb.end()
|
|
|
|
if encode_cmd.returncode == 0:
|
|
main.print('encoding finished: %s -> %s' %(original_filesize, new_filesize), main.print.tl.done)
|
|
else:
|
|
main.print('<r>encoding failed<.>, ffmpeg returned <y>%d<.>' %(encode_cmd.returncode), main.print.tl.fail)
|
|
raise RuntimeError
|
|
|
|
# --------------------------------------------------
|
|
# shuffle files around
|
|
|
|
if main.cfg.move_source:
|
|
move_to_dir = pathlib.Path(main.cfg.move_source)
|
|
|
|
if not move_to_dir.is_dir():
|
|
if main.cfg.armed:
|
|
main.print('creating directory <W>%s<.>' %(move_to_dir), main.print.tl.start)
|
|
move_to_dir.mkdir()
|
|
else:
|
|
main.print('will create directory <W>%s<.>' %(move_to_dir))
|
|
|
|
if files.move_infile_to:
|
|
if main.cfg.armed:
|
|
main.print('moving <W>%s<.> -> <W>%s<.>' %(files.infile, files.move_infile_to), main.print.tl.start)
|
|
files.infile.rename(files.move_infile_to)
|
|
else:
|
|
main.print('will move <W>%s<.> -> <W>%s<.>' %(files.infile, files.move_infile_to))
|
|
|
|
if main.cfg.armed:
|
|
main.print('moving <W>%s<.> -> <W>%s<.>' %(files.tmpfile, files.outfile), main.print.tl.start)
|
|
files.tmpfile.rename(files.outfile)
|
|
else:
|
|
main.print('will move <W>%s<.> -> <W>%s<.>' %(files.tmpfile, files.outfile))
|