Transcode Mpeg2 to H264
From MythTV Official Wiki
This python script is designed to be used as a User Job of the form "/<path to script>/transcode-h264.py %JOBID%" and is known to work on Mythtv 0.27. I got the idea for this script from the shell script referred to in this thread:
http://www.gossamer-threads.com/lists/mythtv/users/578181
I used Raymond Wagner's transcode wrapper stub http://www.mythtv.org/wiki/Transcode_wrapper_stub as the starting point for converting the shell script to python. If a cutlist is found mythtranscode will be used to remove the marked segments and ffmpeg is used to transcode the result to h264. The original recording is replaced (example: 1501_20141014000000.mpg becomes 1501_20141014000000.mp4), the transcoded flag is set, the old preview files (*.png) are removed, and the seektable is cleared.
I compiled ffmpeg using these instructions https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu to make sure I had the latest features.
--Mike Stucky 21:09, 26 February 2015 (UTC)
I have updated my script based on input from this thread: http://www.gossamer-threads.com/lists/mythtv/users/586259
Dave discovered that ffmpeg appears to deinterlace SD content by default so I added "-filter:v yadif=0:-1:1" to the 'transcode' section, this should do a better job than the default deinterlacer. This will only deinterlace if the given frame is marked as interlaced so it should work for both interlaced and progressive content. This works well for my content which is all OTA from HDHomerun tuners.
I removed "-threads 0" since the h264 encoder uses all available threads by default.
And I removed "-c:a copy" and added "-strict -2" to use the built-in AAC encoder.
There are additional parameters referenced in the thread that you may want to add, but this works well for me.
--Mike Stucky (talk) 19:10, 11 May 2015 (UTC)
transcode-h264.py
#!/usr/bin/env python
# -*- coding: UTF-8 -*-
# 2015 Michael Stucky
# This script is based on Raymond Wagner's transcode wrapper stub.
# Designed to be a USERJOB of the form </path to script/transcode-h264.py %JOBID%>
from MythTV import Job, Recorded, System, MythDB, findfile, MythError, MythLog, datetime
from optparse import OptionParser
from glob import glob
from shutil import copyfile
import sys
import os
import errno
import time
transcoder = '/usr/bin/ffmpeg'
flush_commskip = True
build_seektable = True
def runjob(jobid=None, chanid=None, starttime=None):
db = MythDB()
if jobid:
job = Job(jobid, db=db)
chanid = job.chanid
starttime = job.starttime
rec = Recorded((chanid, starttime), db=db)
sg = findfile('/'+rec.basename, rec.storagegroup, db=db)
if sg is None:
print 'Local access to recording not found.'
sys.exit(1)
infile = os.path.join(sg.dirname, rec.basename)
tmpfile = '%s.tmp' % infile.rsplit('.',1)[0]
outfile = '%s.mp4' % infile.rsplit('.',1)[0]
# reformat 'starttime' for use with mythtranscode/ffmpeg/mythcommflag
starttime = str(starttime.utcisoformat().replace(u':', '').replace(u' ', '').replace(u'T', '').replace('-', ''))
# Lossless transcode to strip cutlist
if rec.cutlist == 1:
if jobid:
job.update({'status':4, 'comment':'Removing Cutlist'})
task = System(path='mythtranscode', db=db)
try:
output = task('--chanid "%s"' % chanid,
'--starttime "%s"' % starttime,
'--mpeg2',
'--honorcutlist',
'-o "%s"' % tmpfile,
'2> /dev/null')
except MythError, e:
print 'Command failed with output:\n%s' % e.stderr
if jobid:
job.update({'status':304, 'comment':'Removing Cutlist failed'})
sys.exit(e.retcode)
else:
copyfile('%s' % infile, '%s' % tmpfile)
# Transcode to mp4
if jobid:
job.update({'status':4, 'comment':'Transcoding to mp4'})
task = System(path=transcoder, db=db)
try:
output = task('-i "%s"' % tmpfile,
'-filter:v yadif=0:-1:1',
'-c:v libx264',
'-preset:v slow',
'-crf:v 18',
'-strict -2',
'-metadata:s:a:0',
'language="eng"',
'"%s"' % outfile,
'2> /dev/null')
except MythError, e:
print 'Command failed with output:\n%s' % e.stderr
if jobid:
job.update({'status':304, 'comment':'Transcoding to mp4 failed'})
sys.exit(e.retcode)
rec.basename = os.path.basename(outfile)
os.remove(infile)
# Cleanup the old *.png files
for filename in glob('%s*.png' % infile):
os.remove(filename)
os.remove(tmpfile)
try:
os.remove('%s.map' % tmpfile)
except OSError:
pass
rec.filesize = os.path.getsize(outfile)
rec.transcoded = 1
rec.seek.clean()
if flush_commskip:
for index,mark in reversed(list(enumerate(rec.markup))):
if mark.type in (rec.markup.MARK_COMM_START, rec.markup.MARK_COMM_END):
del rec.markup[index]
rec.bookmark = 0
rec.cutlist = 0
rec.markup.commit()
rec.update()
if jobid:
job.update({'status':4, 'comment':'Rebuilding seektable'})
if build_seektable:
task = System(path='mythcommflag')
task.command('--chanid %s' % chanid,
'--starttime %s' % starttime,
'--rebuild',
'2> /dev/null')
if jobid:
job.update({'status':272, 'comment':'Transcode Completed'})
def main():
parser = OptionParser(usage="usage: %prog [options] [jobid]")
parser.add_option('--chanid', action='store', type='int', dest='chanid',
help='Use chanid for manual operation')
parser.add_option('--starttime', action='store', type='int', dest='starttime',
help='Use starttime for manual operation')
parser.add_option('-v', '--verbose', action='store', type='string', dest='verbose',
help='Verbosity level')
opts, args = parser.parse_args()
if opts.verbose:
if opts.verbose == 'help':
print MythLog.helptext
sys.exit(0)
MythLog._setlevel(opts.verbose)
if len(args) == 1:
runjob(jobid=args[0])
elif opts.chanid and opts.starttime:
runjob(chanid=opts.chanid, starttime=opts.starttime)
else:
print 'Script must be provided jobid, or chanid and starttime.'
sys.exit(1)
if __name__ == '__main__':
main()
I took Mike's script and added some stuff to it. I thought that it might be useful so I've posted it here. Thanks to Mike for his initial post.
transcode-h264-v2.py
#!/usr/bin/env python
# -*- coding: UTF-8 -*-
# 2015 Michael Stucky
#
# This script is based on Raymond Wagner's transcode wrapper stub.
# Designed to be a USERJOB of the form </path to script/transcode-h264.py %JOBID%>
#
# Modifications - Drew 7/5/2015
#
# - chanid/starttime arguments needed a couple fixes to get it working
# - added ability to cut out commercials before transcode via generate_commcutlist=True
# - added ability to implement a compression ratio
# the compression ratio estimates the input streams encoding bitrate (all streams as a one bitrate)
# then computes an output bitrate as a factor of this bitrate, i.e., if compressionRatio=0.75
# then the output video will be encoded at 75% of the input video bitrate. Usually one sets
# compressionRatio < 1, resulting in a smaller output file.
# Note the estimated bitrate is derived from the video duration
# and file size hence it will over estimate the true video bitrate as it does not account for the
# encapsulation overhead of the input encoding scheme nor the bitrate of any included audio streams.
# To enable, set estimateBitrate = True and set compressionRatio to your desired value (I use 0.7).
# - added ability to change h264 encoding preset and constant rate factor (crf) settings for HD video
# - added loads of debug statements, I've kept them in to facilitate hacking -- sorry to the purists in advance
# - added status output from ffmpeg to the myth backend giving % complete, ETA and fps encode statistics.
# - added "smart" commercial detection - if needed it is run and on completion cancels any other mythcommflag jobs
# for the transcoded recording
# Modifications - Drew 1/25/2016
# - added fix for the markup data which is inaccurate, especially when commercials are removed
#
from MythTV import Job, Recorded, System, MythDB, findfile, MythError, MythLog, datetime
from optparse import OptionParser
from glob import glob
from shutil import copyfile
import sys
import os
import errno
import threading, time
from datetime import timedelta
import re, tempfile
import Queue # thread-safe
########## IMPORTANT #####################
#
# YOU WILL NEED TO EDIT THE SETTINGS BELOW
#
########## IMPORTANT #####################
transcoder = '/usr/bin/ffmpeg'
# flush_commskip
# True => (Default) the script will delete all commercial skip indices from the old file
# False => the transcode will leave the commercial skip indices from the old file "as is"
flush_commskip = True
# require_commflagged
# True => the script will ensure mythcommflag has run on the file before encoding it
# False => (Default) the transcode will process the video file "as is"
require_commflagged = False
# generate_commcutlist
# True => (Default) flagged commercials are removed from the output video file
# False => flagged commercials are NOT removed from the output video file
generate_commcutlist = True
# estimateBitrate
# True => (Default) the bitrate of the input file is estimated via size & duration
# ** Required True for "compressionRatio" option to work.
# False => The bitrate of the input file is unknown
estimateBitrate = True
# compressionRatio
# 0.0 - 1.0 => Set the approximate bitrate of the output relative to the
# detected bitrate of the input.
# One can think of this as the target compression rate, i.e., the
# compressionRatio = (output filesize)/(input filesize)
# h264 video quality is approximately equal to mpeg2 video quality
# at a compression ratio of 0.65-0.75
# * Note: When enabled, this value will determine the approximate
# relative size of the output file and input file
# (output filesize) = compressionRatio * (input filesize)
compressionRatio = 0.65
# enforce a max (do not exceed) bitrate for encoded HD video
# to disable set hd_max_bitrate=0
hdvideo_max_bitrate = 5500 # 0 = disable or (kBits_per_sec,kbps)
hdvideo_min_bitrate = 0 # 0 = disable or (kBits_per_sec,kbps)
# number of seconds of video that can be held in playing device video buffers (typically 2-5 secs)
NUM_SECS_VIDEO_BUF=3 # secs
device_bufsize = NUM_SECS_VIDEO_BUF*hdvideo_max_bitrate # (kBits_per_sec,kbps)
# enforce a target bitrate for the encoder to achieve approximately
#hdvideo_tgt_bitrate = 5000 # 0 = disable or (kBits_per_sec,kbps)
hdvideo_tgt_bitrate = 0 # 0 = disable or (kBits_per_sec,kbps)
# build_seektable
# True => Rebuild myth seek table.
# It allows accurate ffwd,rew / seeking on the transcoded output video
# False => (Default) Do not rebuild the myth seek table. Not working for mythtv on h264 content.
build_seektable = False
# Making this true enables a bunch of debug information to be printed as the script runs.
debug = False
# TODO - override buffer size (kB), only use when necessary for a specific target device
# bufsize_override=0 # 0 = disable or (kBits_per_sec,kbps)
# h264 encode preset
# ultrafast,superfast, veryfast, faster, fast, medium, slow, slower, veryslow
preset_HD = 'fast'
preset_nonHD = 'slow'
# h264 encode constant rate factor (used for non-HD) valid/sane values 18-28
# lower values -> higher quality, larger output files,
# higher values -> lower quality, smaller output files
crf = '21'
# if HD, copy input audio streams to the output audio streams
abitrate_param_HD='-c:a copy'
# if non-HD, encode audio to AAC with libfdk_aac at a bitrate of 128kbps
abitrate_param_nonHD = '-c:a libfdk_aac -b:a 128k'
# to convert non-HD audio to AAC using ffmpeg's aac encoder
#abitrate_param_nonHD='-strict -2'
# TODO use -crf 20 -maxrate 400k -bufsize 1835k
# effectively "target" crf 20, but if the output exceeds 400kb/s, it will degrade to something more than crf 20
# TODO detect and preserve ac3 5.1 streams typically found in HD content
# TODO detect and preserve audio streams by language
# TODO detect and preserve subtitle streams by language
# TODO is mp4 or mkv better for subtitle support in playback
# subtitle codecs for MKV containers: copy, ass, srt, ssa
# subtitle codecs for MP4 containers: copy, mov_text
# Languages for audio stream and subtitle selection
# eng - English
# fre - French
# ger - German
# ita - Italian
# spa - Spanish
language = 'eng'
# interval between reads from the ffmpeg status file
# also defines the interval when waiting for a mythcommflag job to finish
POLL_INTERVAL=10 # secs
# mythtv automatically launched user jobs with nice level of 17
# this will add to that level (only positive values allowed unless run as root)
# e.g., NICELEVEL=1 will run with a nice level of 18. The max nicelevel is 19.
#NICELEVEL=5
NICELEVEL=0
class CleanExit:
pass
def runjob(jobid=None, chanid=None, starttime=None, tzoffset=None):
global estimateBitrate
db = MythDB()
if jobid:
job = Job(jobid, db=db)
chanid = job.chanid
utcstarttime = job.starttime
else:
job=None;
utcstarttime = datetime.strptime(starttime, "%Y%m%d%H%M%S")
utcstarttime = utcstarttime + timedelta(hours=tzoffset)
if debug:
print 'chanid "%s"' % chanid
print 'utcstarttime "%s"' % utcstarttime
rec = Recorded((chanid, utcstarttime), db=db);
utcstarttime = rec.starttime;
starttime_datetime = utcstarttime
# reformat 'starttime' for use with mythtranscode/ffmpeg/mythcommflag
starttime = str(utcstarttime.utcisoformat().replace(u':', '').replace(u' ', '').replace(u'T', '').replace('-', ''))
if debug:
print 'mythtv format starttime "%s"' % starttime
input_filesize = rec.filesize
if rec.commflagged:
if debug:
print 'Recording has been scanned to detect commerical breaks.'
waititer=1
keepWaiting = True
while keepWaiting == True:
keepWaiting=False;
for index,jobitem in reversed(list(enumerate(db.searchJobs(chanid=chanid, starttime=starttime_datetime)))):
if jobitem.type == jobitem.COMMFLAG: # Commercial flagging job
if debug:
print 'Commercial flagging job detected with status %s' % jobitem.status
if jobitem.status == jobitem.RUNNING: # status = RUNNING?
job.update({'status':job.PAUSED,
'comment':'Waited %d secs for the commercial flagging job' % (waititer*POLL_INTERVAL) \
+ ' currently running on this recording to complete.'})
if debug:
print 'Waited %d secs for the commercial flagging job' % (waititer*POLL_INTERVAL) \
+ ' currently running on this recording to complete.'
time.sleep(POLL_INTERVAL);
keepWaiting=True
waititer = waititer + 1
break
else:
if debug:
print 'Recording has not been scanned to detect/remove commercial breaks.'
if require_commflagged:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Required commercial flagging for this file is not found.'
+ 'Flagging commercials and cancelling any queued commercial flagging.'})
# cancel any queued job to flag commercials for this recording and run commercial flagging in this script
for index,jobitem in reversed(list(enumerate(db.searchJobs(chanid=chanid,starttime=starttime_datetime)))):
if debug:
if index==0:
print jobitem.keys()
print index,jobitem.id,jobitem.chanid
if jobitem.type == jobitem.COMMFLAG: # Commercial flagging job
if jobitem.status == jobitem.RUNNING: # status = RUNNING?
jobitem.cmds = jobitem.STOP # stop command from the frontend to stop the commercial flagging job
#jobitem.setStatus(jobitem.CANCELLED)
#jobitem.setComment('Cancelled: Transcode command ran commercial flagging for this recording.')
jobitem.update({'status':jobitem.CANCELLED,
'comment':'A user transcode job ran commercial flagging for'
+ ' this recording and cancelled this job.'})
if debug:
print 'Flagging Commercials...'
# Call "mythcommflag --chanid $CHANID --starttime $STARTTIME"
task = System(path='mythcommflag', db=db)
try:
output = task('--chanid "%s"' % chanid,
'--starttime "%s"' % starttime,
'2> /dev/null')
except MythError, e:
# it seems mythcommflag always exits with an decoding error "eno: Unknown error 541478725 (541478725)"
pass
#print 'Command failed with output:\n%s' % e.stderr
#if jobid:
# job.update({'status':304, 'comment':'Flagging commercials failed'})
#sys.exit(e.retcode)
sg = findfile('/'+rec.basename, rec.storagegroup, db=db)
if sg is None:
print 'Local access to recording not found.'
sys.exit(1)
infile = os.path.join(sg.dirname, rec.basename)
tmpfile = '%s.tmp' % infile.rsplit('.',1)[0]
# tmpfile = infile
outfile = '%s.mp4' % infile.rsplit('.',1)[0]
if debug:
print 'tmpfile "%s"' % tmpfile
clipped_bytes=0;
# If selected, create a cutlist to remove commercials via mythtranscode by running:
# mythutil --gencutlist --chanid $CHANID --starttime $STARTTIME
if generate_commcutlist:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Generating Cutlist for commercial removal'})
task = System(path='mythutil', db=db)
try:
output = task('--gencutlist',
'--chanid "%s"' % chanid,
'--starttime "%s"' % starttime)
# '--loglevel debug',
# '2> /dev/null')
except MythError, e:
print 'Command "mythutil --gencutlist" failed with output:\n%s' % e.stderr
if jobid:
job.update({'status':job.ERRORED, 'comment':'Generation of commercial Cutlist failed'})
sys.exit(e.retcode)
# Lossless transcode to strip cutlist
if generate_commcutlist or rec.cutlist==1:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Removing Cutlist'})
task = System(path='mythtranscode', db=db)
try:
output = task('--chanid "%s"' % chanid,
'--starttime "%s"' % starttime,
'--mpeg2',
'--honorcutlist',
'-o "%s"' % tmpfile,
'1>&2')
# '2> /dev/null')
clipped_filesize = os.path.getsize(tmpfile)
clipped_bytes = input_filesize - clipped_filesize
clipped_compress_pct = float(clipped_bytes)/input_filesize
rec.commflagged = 0
except MythError, e:
print 'Command "mythtranscode --honorcutlist" failed with output:\n%s' % e.stderr
if jobid:
job.update({'status':job.ERRORED, 'comment':'Removing Cutlist failed. Copying file instead.'})
# sys.exit(e.retcode)
copyfile('%s' % infile, '%s' % tmpfile)
clipped_filesize = input_filesize
clipped_bytes = 0
clipped_compress_pct = 0
pass
else:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Creating temporary file for transcoding.'})
copyfile('%s' % infile, '%s' % tmpfile)
clipped_filesize = input_filesize
clipped_bytes = 0
clipped_compress_pct = 0
duration_secs = 0
# Estimate bitrate, and detect duration and number of frames
if estimateBitrate:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Estimating bitrate; detecting frames per second, and resolution.'})
duration_secs, e = get_duration(db, rec, transcoder, tmpfile);
if duration_secs>0:
bitrate = int(clipped_filesize*8/(1024*duration_secs))
else:
print 'Estimate bitrate failed falling back to constant rate factor encoding.\n'
estimateBitrate = False
duration_secs = 0
# get framerate of mpeg2 video stream and detect if stream is HD
r = re.compile('mpeg2video (.*?) fps,')
m = r.search(e.stderr)
strval = m.group(1)
if debug:
print strval
isHD = False
if "1920x1080" in strval or "1280x720" in strval or "2560x1440" in strval:
if debug:
print 'Stream is HD'
isHD = True
else:
if debug:
print 'Stream is not HD'
framerate = float(m.group(1).split(' ')[-1])
if debug:
print 'Framerate %s' % framerate
# Setup transcode video bitrate and quality parameters
# if estimateBitrate is true and the input content is HD:
# encode 'medium' preset and vbitrate = inputfile_bitrate*compressionRatio
# else:
# encode at user default preset and constant rate factor ('slow' and 20)
preset = preset_nonHD
if estimateBitrate:
if isHD:
h264_bitrate = int(bitrate*compressionRatio)
# HD coding with specified target bitrate (CRB encoding)
if hdvideo_tgt_bitrate > 0 and h264_bitrate > hdvideo_tgt_bitrate:
h264_bitrate = hdvideo_tgt_bitrate;
vbitrate_param = '-b:v %dk' % h264_bitrate
else: # HD coding with disabled or acceptable target bitrate (CRF encoding)
vbitrate_param = '-crf:v %s' % crf
preset = preset_HD
else: # non-HD encoding (CRF encoding)
vbitrate_param = '-crf:v %s' % crf
else:
vbitrate_param = '-crf:v %s' % crf
if hdvideo_min_bitrate > 0:
vbitrate_param = vbitrate_param + ' -minrate %sk' % hdvideo_min_bitrate
if hdvideo_max_bitrate > 0:
vbitrate_param = vbitrate_param + ' -maxrate %sk' % hdvideo_max_bitrate
if hdvideo_max_bitrate > 0 or hdvideo_min_bitrate > 0:
vbitrate_param = vbitrate_param + ' -bufsize %sk' % device_bufsize
if debug:
print 'Video bitrate parameter "%s"' % vbitrate_param
print 'Video h264 preset parameter "%s"' % preset
# Setup transcode audio bitrate and quality parameters
# Right now, the setup is as follows:
# if input is HD:
# copy audio streams to output, i.e., input=output audio
# else:
# output is libfdk_aac encoded at 128kbps
if isHD:
abitrate_param = abitrate_param_HD # preserve 5.1 audio
else:
abitrate_param = abitrate_param_nonHD
if debug:
print 'Audio bitrate parameter "%s"' % abitrate_param
# Transcode to mp4
# if jobid:
# job.update({'status':4, 'comment':'Transcoding to mp4'})
# ffmpeg output is redirected to the temporary file tmpstatusfile and
# a second thread continuously reads this file while
# the transcode is in-process. see while loop below for the monitoring thread
tf = tempfile.NamedTemporaryFile()
tmpstatusfile = tf.name
# tmpstatusfile = '/tmp/ffmpeg-transcode.txt'
if debug:
print 'Using temporary file "%s" for ffmpeg status updates.' % tmpstatusfile
res = []
# create a thread to perform the encode
ipq = Queue.Queue()
t = threading.Thread(target=wrapper, args=(encode,
(jobid, db, job, ipq, preset, vbitrate_param, abitrate_param,
tmpfile, outfile, tmpstatusfile,), res))
t.start()
# wait for ffmpeg to open the file and emit its initialization information
# before we start the monitoring process
time.sleep(1)
# open the temporary file having the ffmeg output text and process it to generate status updates
hangiter=0;
with open(tmpstatusfile) as f:
# read all the opening ffmpeg status/analysis lines
lines = f.readlines()
# set initial progress to -1
prev_progress=-1
framenum=0
fps=1.0
while t.is_alive():
# read all output since last readline() call
lines = f.readlines()
if len(lines) > 0:
# every ffmpeg output status line ends with a carriage return '\r'
# split the last read line at these locations
lines=lines[-1].split('\r')
# if debug:
# print lines;
hangiter=0
if len(lines) > 1 and lines[-2].startswith('frame'):
# since typical reads will have the last line ending with \r the last status
# message is at index=[-2] start processing this line
# replace multiple spaces with one space
lines[-2] = re.sub(' +',' ',lines[-2])
# remove any spaces after equals signs
lines[-2] = re.sub('= +','=',lines[-2])
# split the fields at the spaces the first two fields for typical
# status lines will be framenum=XXXX and fps=YYYY parse the values
values = lines[-2].split(' ')
if len(values) > 1:
if debug:
print 'values %s' % values
prev_framenum = framenum
prev_fps = fps
try:
# framenum = current frame number being encoded
framenum = int(values[0].split('=')[1])
# fps = frames per second for the encoder
fps = float(values[1].split('=')[1])
except ValueError, e:
print 'ffmpeg status parse exception: "%s"' % e
framenum = prev_framenum
fps = prev_fps
pass
# progress = 0-100 represent percent complete for the transcode
progress = int((100*framenum)/(duration_secs*framerate))
# eta_secs = estimated number of seconds until transcoding is complete
eta_secs = int((float(duration_secs*framerate)-framenum)/fps)
# pct_realtime = how many real seconds it takes to encode 1 second of video
pct_realtime = float(fps/framerate)
if debug:
print 'framenum = %d fps = %.2f' % (framenum, fps)
if progress != prev_progress:
if debug:
print 'Progress %d%% encoding %.1f frames per second ETA %d mins' \
% ( progress, fps, float(eta_secs)/60)
if jobid:
progress_str = 'Transcoding to mp4 %d%% complete ETA %d mins fps=%.1f.' \
% ( progress, float(eta_secs)/60, fps)
job.update({'status':job.RUNNING, 'comment': progress_str})
prev_progress = progress
elif len(lines) > 1:
if debug:
print 'Read pathological output %s' % lines[-2]
else:
if debug:
print 'Read no lines of ffmpeg output for %s secs. Possible hang?' % (POLL_INTERVAL*hangiter)
hangiter = hangiter + 1
if jobid:
progress_str = 'Read no lines of ffmpeg output for %s secs. Possible hang?' % (POLL_INTERVAL*hangiter)
job.update({'status':job.RUNNING, 'comment': progress_str})
time.sleep(POLL_INTERVAL)
if debug:
print 'res = "%s"' % res
t.join(1)
try:
if ipq.get_nowait() == CleanExit:
sys.exit()
except Queue.Empty:
pass
if flush_commskip:
task = System(path='mythutil')
task.command('--chanid %s' % chanid,
'--starttime %s' % starttime,
'--clearcutlist',
'2> /dev/null')
task = System(path='mythutil')
task.command('--chanid %s' % chanid,
'--starttime %s' % starttime,
'--clearskiplist',
'2> /dev/null')
if flush_commskip:
for index,mark in reversed(list(enumerate(rec.markup))):
if mark.type in (rec.markup.MARK_COMM_START, rec.markup.MARK_COMM_END):
del rec.markup[index]
rec.bookmark = 0
rec.cutlist = 0
rec.markup.commit()
# tf.close();
# os.remove(tmpstatusfile);
rec.basename = os.path.basename(outfile)
rec.filesize = os.path.getsize(outfile)
# rec.commflagged = 0
rec.transcoded = 1
rec.seek.clean()
rec.update()
os.remove(infile)
# Cleanup the old *.png files
for filename in glob('%s*.png' % infile):
os.remove(filename)
os.remove(tmpfile)
try:
os.remove('%s.map' % tmpfile)
except OSError:
pass
output_filesize = rec.filesize
if duration_secs > 0:
output_bitrate = int(output_filesize*8/(1024*duration_secs)) # kbps
actual_compression_ratio = 1 - float(output_filesize)/clipped_filesize
compressed_pct = 1 - float(output_filesize)/input_filesize
if build_seektable:
if jobid:
job.update({'status':job.RUNNING, 'comment':'Rebuilding seektable'})
task = System(path='mythcommflag')
task.command('--chanid %s' % chanid,
'--starttime %s' % starttime,
'--rebuild',
'2> /dev/null')
# fix during in the recorded markup table this will be off if commercials are removed
duration_msecs, e = get_duration(db, rec, transcoder, outfile)
duration_msecs = 1000*duration_msecs
for index,mark in reversed(list(enumerate(rec.markup))):
# find the duration markup entry and correct any error in the video duration that might be there
if mark.type == 33:
if debug:
print 'Markup Duration in milliseconds "%s"' % mark.data
error = mark.data - duration_msecs
if error != 0:
if debug:
print 'Markup Duration error is "%s"msecs' % error
mark.data = duration_msecs
#rec.bookmark = 0
#rec.cutlist = 0
rec.markup.commit()
if jobid:
if output_bitrate:
job.update({'status':job.FINISHED, 'comment':'Transcode Completed @ %dkbps, compressed file by %d%% (clipped %d%%, transcoder compressed %d%%)' % (output_bitrate,int(compressed_pct*100),int(clipped_compress_pct*100),int(actual_compression_ratio*100))})
else:
job.update({'status':job.FINISHED, 'comment':'Transcode Completed'})
def get_duration(db=None, rec=None, transcoder='/usr/bin/ffmpeg', filename=None):
task = System(path=transcoder, db=db)
if filename is None:
return -1
try:
output = task('-i "%s"' % filename, '1>&2')
except MythError, e:
pass
r = re.compile('Duration: (.*?), start')
m = r.search(e.stderr)
if m:
duration = m.group(1).split(':')
duration_secs = float((int(duration[0])*60+int(duration[1]))*60+float(duration[2]))
duration_msecs = int(1000*duration_secs)
if debug:
print 'Duration %s' % m.group(1)
print 'Duration %s' % duration
print 'Duration in seconds "%s"' % duration_secs
print 'Duration in milliseconds "%s"' % duration_msecs
return duration_secs, e
return -1, e
def encode(jobid=None, db=None, job=None,
procqueue=None, preset='slow',
vbitrate_param='-crf:v 18',
abitrate_param='-c:a libfdk_aac -b:a 128k',
tmpfile=None, outfile=None, statusfile=None):
# task = System(path=transcoder, db=db)
task = System(path='nice', db=db)
try:
output = task(
'-n %s' % NICELEVEL,
'%s' % transcoder,
'-i "%s"' % tmpfile,
# parameter to overwrite output file if present without prompt
'-y',
# parameter de-interlacing filter
'-filter:v yadif=0:-1:1',
# parameter to allow streaming content
'-movflags faststart',
# parameter needed when hdhomerun prime mpeg2 files sometime repeat timestamps
'-vsync passthrough',
# h264 video codec
'-c:v libx264',
# presets for h264 encode that effect encode speed/output filesize
'-preset:v %s' % preset,
# ########## IMPORTANT ############
# ffmpeg versions after 08-18-2015 include a change to force explicit IDR frames,
# setting this flag helps/corrects myth seektable indexing h264-encoded files
# uncomment the line below if you have a recent version of ffmpeg that supports this option
# '-forced-idr 1',
# parameters to determine video encode target bitrate
vbitrate_param,
# parameters to determine audio encode target bitrate
abitrate_param,
# parameter to encode all input audio streams into the output
# '-map 0:a',
# parameters to set the first output audio stream
# to be an audio stream having the specified language (default=eng -> English)
# '-metadata:s:a:0',
# 'language=%s' % language,
# parameter to copy input subtitle streams into the output
'-c:s copy',
# '-c:s mov_text',
# parameters to set the first output subtitle stream
# to be an english subtitle stream
# '-metadata:s:s:0',
# 'language=%s' % language,
# we can control the number of encode threads (disabled)
# '-threads 2',
# output file parameter
'"%s"' % outfile,
# redirection of output to temporaryfile
'> %s 2>&1 < /dev/null' % statusfile)
except MythError, e:
print 'Command failed with output:\n%s' % e.stderr
if jobid:
job.update({'status':job.ERRORED, 'comment':'Transcoding to mp4 failed'})
procqueue.put(CleanExit)
sys.exit(e.retcode)
def wrapper(func, args, res):
res.append(func(*args))
def main():
parser = OptionParser(usage="usage: %prog [options] [jobid]")
parser.add_option('--chanid', action='store', type='int', dest='chanid',
help='Use chanid with both starttime and tzoffset for manual operation')
parser.add_option('--starttime', action='store', type='string', dest='starttime',
help='Use starttime with both chanid and tzoffset for manual operation')
parser.add_option('--tzoffset', action='store', type='int', dest='tzoffset',
help='Use tzoffset with both chanid and starttime for manual operation')
parser.add_option('-v', '--verbose', action='store', type='string', dest='verbose',
help='Verbosity level')
opts, args = parser.parse_args()
if opts.verbose:
if opts.verbose == 'help':
print MythLog.helptext
sys.exit(0)
MythLog._setlevel(opts.verbose)
if len(args) == 1:
runjob(jobid=args[0])
elif opts.chanid and opts.starttime and opts.tzoffset is not None:
runjob(chanid=opts.chanid, starttime=opts.starttime, tzoffset=opts.tzoffset)
else:
print 'Script must be provided jobid, or chanid, starttime and timezone offset.'
sys.exit(1)
if __name__ == '__main__':
main()
