现在我可以说,除了09年当时的H.263文件头拼接没有做(实在没意义了),其他所有Windows上的黑科技都已经成功移植到了Linux和OSX上。
基于小丸的办法,但是我连音频一起做——防止有天谁抽风了想起来查最短轨道。
美中不足的是略慢。
真是到家了。
3个小时候看大夫去。。。。。。。。
地址:
https://github.com/cnbeining/videoblacker
代码扔下面。
#!/usr/bin/env python #coding:utf-8 # Author: Beining --<ACICFG> # Contact: http://www.cnbeining.com/ |https://github.com/cnbeining/videoblacker # Purpose: Patch video's both part to bypass Letvcloud's transcode. # Created: 08/29/2014 # LICENSE: GNU v2 import sys import os import os, sys, subprocess, shlex, re from subprocess import call import uuid import math import shutil import getopt #---------------------------------------------------------------------- def probe_file(filename): cmnd = ['ffprobe', '-show_format', '-pretty', '-loglevel', 'quiet', filename] p = subprocess.Popen(cmnd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) #print filename out, err = p.communicate() #print out if err: print err return None return out #---------------------------------------------------------------------- def time_to_sec(time_raw): """ str->int ignore .*.""" hr = int(time_raw.split(':')[0]) * 3600 minute = int(time_raw.split(':')[1]) * 60 sec = int(float(time_raw.split(':')[2])) return int(hr + minute + sec) #---------------------------------------------------------------------- def get_abspath(filename): """str->str""" return str(os.path.abspath(filename)) #---------------------------------------------------------------------- def get_dimention_mediainfo(filename): """str->str path to dimention""" return str(os.popen('mediainfo \'--Inform=Video;%Width%x%Height%\' \'' + filename + '\'').read()).replace('x', '*').strip() #---------------------------------------------------------------------- def process(filename, target_bitrate, audiobitrate, safe, outputfile): """str,int,int,str->? filename,outputfile comes with the path. safe=0: directly convert the file to m4a""" tmpdir = '/tmp/Audioblacker-' + str(uuid.uuid4()) if not os.path.exists(tmpdir): os.makedirs(tmpdir) audio_format = '' audio_duration = '' audio_bitrate = '' video_format = '' video_duration = '' video_bitrate = '' ff = '' if safe == 1: #directly to mp4 print('INFO: Pre-processing video ...') os.system('ffmpeg -i \'' + filename + '\' -c copy ' + tmpdir+'/orivideo.mp4' +'> /dev/null 2>&1') filename = tmpdir+'/orivideo.mp4' elif safe == 2: #All the f**king dramas...... print('INFO: Pre-processing video ...') os.system('ffmpeg -i \'' + filename + '\' -c copy ' + tmpdir+'/orivideo1.mp4' +'> /dev/null 2>&1') os.system('ffmpeg -i ' + tmpdir+'/orivideo1.mp4 -vbsf h264_mp4toannexb -an -c:v copy ' + tmpdir+'/orivideo.h264' +'> /dev/null 2>&1') os.system('ffmpeg -i ' + tmpdir+'/orivideo1.mp4 -vn -f adts -c:a copy ' + tmpdir+'/oriaudio.aac' +'> /dev/null 2>&1') os.system('ffmpeg -i ' + tmpdir+'/orivideo.h264 -i '+ tmpdir+'/oriaudio.aac -bsf:a aac_adtstoasc -c copy ' + tmpdir+'/orivideo.mp4' +'> /dev/null 2>&1') filename = tmpdir+'/orivideo.mp4' try: os.remove(tmpdir+'/orivideo.h264') os.remove(tmpdir+'/oriaudio.aac') pass except: print('WARNING: Cannot delete temp file, watch out your free space!') filename = tmpdir+'/orivideo.mp4' print('INFO: Detecting video info...') os.system('ffmpeg -i \'' + filename + '\' -vn -c:a copy ' + tmpdir+'/audio.m4a') # +' > /dev/null 2>&1') try: for line in probe_file(tmpdir+'/audio.m4a').split('\n'): if 'duration' in line: audio_duration = str(line.split('=')[1]) break for line in probe_file(filename).split('\n'): if 'duration' in line: video_duration = str(line.split('=')[1]) break except: print('ERROR: Cannot read video file!') shutil.rmtree(tmpdir) exit() #Calc... #time_by_video print('INFO: Doing calculation...') try: video_duration_sec = time_to_sec(video_duration) video_size_byte = int(os.path.getsize(filename)) audio_duration_sec = time_to_sec(audio_duration) audio_size_byte = int(os.path.getsize(tmpdir+'/audio.m4a')) print(video_duration_sec) except: print('ERROR: Cannot calculate time, did you input a bitrate too high?') shutil.rmtree(tmpdir) exit() #!!!!!FLOAT DUE TO ACCURACY!!!!! target_byterate = target_bitrate / 8.0 target_audiobyterate = audiobitrate / 8.0 time_video = int(math.ceil((video_size_byte - target_byterate * video_duration_sec) / (target_byterate - 470))) time_audio = int(math.ceil((audio_size_byte - target_audiobyterate * audio_duration_sec) / (target_audiobyterate - 470))) if time_audio < 0 and time_video < 0: print('ERROR: Cannot calculate target, your target bitrate is higher than the original file!') shutil.rmtree(tmpdir) exit() if time_audio == 0 and time_video == 0: time_patch = 60 elif time_audio > time_video: time_patch = time_audio + 10 else: time_patch = time_video + 10 #Make audio patch print('INFO: Making ' + str(time_patch) + ' secs of blank video, can be slow...') os.system('ffmpeg -filter_complex \'color=black:s=' + get_dimention_mediainfo(filename) + ':d=' + str(time_patch) + '\' -c:v libx264 -filter_complex \'aevalsrc=0:d=' + str(time_patch) + '\' -strict experimental -c:a aac ' + tmpdir + '/patch.mp4'+'> /dev/null 2>&1') f = open(tmpdir + '/patch.txt', 'w') ff = 'file \'' + filename + '\'' + '\n' + 'file \'' + tmpdir + '/patch.mp4\'' py_path = sys.path[0] os.chdir(py_path) f.write(ff) f.close() print('INFO: Making final output file...') os.system('ffmpeg -f concat -i '+ tmpdir + '/patch.txt -c copy ' + outputfile +'> /dev/null 2>&1') print('Done!') #clean up try: shutil.rmtree(tmpdir) except: print('ERROR: Cannot remove temp dir, do it by yourself!') #---------------------------------------------------------------------- def usage(): """""" print('''Usage: python videoblacker.py (-h) (-i input.mp4) (-o output.mp4) (-b 1900000) (-a 110000) (-s 1) -h: Default: None Help. -i: Default: Blank Input file. If the file and audioblacker are not under the same path, it is suggested to use absolute path to avoid possible failure. -o Default: input-filename.black.mp4 Output file. Would be in the same folder with the original file if not specified. -b: Default: 1900000 Target bitrate. -a: Default: 110000 Target audio bitrate. videoblacker would calculate both of the required black time, and choose the larger one to make sure your convert is successful. Please notice that if your original video/audio bitrate is too small, videoblacker would throw you an ERROR and quit. -s: Default: 1 Use safe mode. Disabling would save you some (can be a lot!) space and time, if you know what you are doing. - 1: videoblacker would remux the file to MP4 at the very first. Good enough for most of the files. - 2: videoblacker would remux the file to h264 and aac at the very first, then remux them to MP4. I put it there in case you come across some videos-from-hell. ''') #---------------------------------------------------------------------- if __name__=='__main__': argv_list = [] argv_list = sys.argv[1:] filename = '' target_bitrate = 1900000 outputfile = '' safe = 1 audiobitrate = 110000 try: opts, args = getopt.getopt(argv_list, "hi:b:a:o:s:", ['help', 'input','bitrate', 'audiobitrate' 'outputfile', 'safe']) except getopt.GetoptError: usage() exit() for o, a in opts: if o in ('-h', '--help'): usage() exit() elif o in ('-i', '--input'): filename = a try: argv_list.remove('-i') except: break elif o in ('-b', '--bitrate'): target_bitrate = int(a) try: argv_list.remove('-b') except: break elif o in ('-a', '--audiobitrate'): audiobitrate = int(a) try: argv_list.remove('-a') except: break elif o in ('-o', '--outputfile'): outputfile = a try: argv_list.remove('-o') except: break elif o in ('-s', '--safe'): safe = int(a) try: argv_list.remove('-s') except: break if filename == '': print('ERROR: No input file!') exit() if outputfile == '': outputfile = filename.split('.')[0] for i in filename.split('.')[1:-1]: outputfile = outputfile + '.' + i outputfile = outputfile + '.black.mp4' process(filename, target_bitrate, audiobitrate, safe, outputfile) exit()
赞。加油的说。
对了。A投B似乎会有弹幕池问题,你有遇到吗?
恩。。。是这样
B站的弹幕池编号就是cid。只要你提交了投稿,就每个P给你分配一个cid。
A站的弹幕池编号嘛。。。可以是vid,ykid,qid,tid,vu等等等等。。。
我个人没遇见过弹幕池问题,因为理论上cid不撞就不会出事。
但是我还真不知道B站怎么共用弹幕池。。。。。你知道么?
好像B站的直传弹幕池有点乱。我也见到我直传的弹幕管理里有两个弹幕池的,一个是乐视vid号,一个是cid号。
而且我也见过用黑科技投稿结果不能发弹幕的。
那个黑科技是上传视频后如果无限转码就删掉视频新建一个把投稿代码贴上去。然后就发现弹幕进入原视频弹幕池了。
除非cid一样了,否则不应该出现这种问题啊。。。。
似乎B站对直传的视频的弹幕池不是用cid标记的而是用vid标记的。
有待确认。
反正我是不想冒险投两个稿填同样的vid看cid是否相同。
如果是我写,cid应该是自增长的。
但是现在咱们自己看不见cid,这很麻烦。。。。。
av7 的 cid 是 475
av9 的 cid 是 297
除非 av7 在 av9 之后上传的,要么这不科学。
可以科学。
例如,重投或者加源,都会产生新的cid。这就科学了。
我觉得你这个评论框很不科学(笑)
怎么说呢。。。因为垃圾评论太多 3月份给我刷了6.8W条,现在一个月在2~5K左右。
要是有留言显示审核的,我肯定可以看见。
但是嵌套问题和转义就没啥办法了。。。反XSS啊。