#!/usr/bin/python
"""
assembles raw cuts into final, titles, tweaks audio, encodes to format for upload.
"""
import os
import sys
import subprocess
import xml.etree.ElementTree
# import xml.etree.ElementTree as ET
from mk_mlt import mk_mlt
import pprint
from process import process
from main.models import Client, Show, Location, Episode, Raw_File, Cut_List
def time2s(time):
""" given 's.s' or 'h:m:s.s' returns s.s """
sec = reduce(lambda x, i: x * 60 + i,
map(float, time.split(':')))
return sec
def time2f(time, fps):
if time[-1] == 'f':
return int(time[:-1])
else:
return int(time2s(time) * fps)
def time2b(time, fps, bpf, default):
"""
converts the time stored in the db
(as blank, seconds, h:m:s.s )
to the byte offset in the file.
blank returns default, typically either 0 or filesize for start/end.
fps is 25.0 for pal and 29.9 ntsc.
bpf (bytes per frame) is 120000 for both.
(huh? doesn't matter, this func not used.)
"""
if time:
bytes = int(time2s(time) * fps) * bpf
else:
bytes = default
return bytes
class enc(process):
ready_state = 2
def mk_title_svg(self, raw_svg, texts):
"""
Make a title slide by filling in a pre-made svg with name/authors.
return: svg
"""
tree = xml.etree.ElementTree.XMLID(raw_svg)
for key in texts:
if self.options.verbose:
print "looking for:", key
# tollerate template where tokens have been removed
if tree[1].has_key(key):
if key == "license":
# CC license image
if self.options.verbose:
print "found in svg:", tree[1][key]
print "replacing with:", texts[key]
t = tree[1][key]
# import code; code.interact(local=locals())
if texts[key] is None:
# del(tree[1][key])
# print tree[1].has_key(key)
tree[1][key].clear()
else:
t.set('{http://www.w3.org/1999/xlink}href', texts[key])
else:
if self.options.verbose:
print "found in svg:", tree[1][key].text
print "replacing with:", texts[key] # .encode()
tree[1][key].text = texts[key]
# cooked_svg = xml.etree.ElementTree.tostring(tree[0])
# print "testing...", "license" in cooked_svg
if tree[1].has_key('presenternames'):
# some people like to add spiffy text near the presenter name(s)
if texts['authors']:
# prefix = u"Featuring" if "," in texts['authors'] else "By"
# tree[1]['presenternames'].text=u"%s %s" % (prefix,texts['authors'])
tree[1]['presenternames'].text = texts['authors']
else:
# remove the text (there is a placholder to make editing sane)
tree[1]['presenternames'].text = ""
cooked_svg = xml.etree.ElementTree.tostring(tree[0])
print "testing...", "license" in cooked_svg
return cooked_svg
def get_title_text(self, episode):
# lets try putting (stuff) on a new line
title = episode.name
authors = episode.authors
if episode.show.slug == 'wtd_na_2015':
title = title.upper()
authors = authors.upper()
if episode.show.slug != 'pygotham_2015' and len(title) > 80: # crazy long titles need all the lines
title2 = ''
elif ": " in title: # the space keeps 9:00 from breaking
pos = title.index(":") + 1
title, title2 = title[:pos], title[pos:].strip()
elif " - " in title:
# error if there is more than 1.
title, title2 = title.split(' - ')
elif " -- " in title:
# error if there is more than 1.
title, title2 = title.split(' -- ')
elif " (" in title:
pos = title.index(" (")
# +1 skip space in " ("
title, title2 = title[:pos], title[pos + 1:]
elif " using " in title:
pos = title.index(" using ")
title, title2 = title[:pos], title[pos + 1:]
elif ";" in title:
pos = title.index(";") + 1
title, title2 = title[:pos], title[pos:].strip()
elif "?" in title:
pos = title.index("?") + 1
title, title2 = title[:pos], title[pos:].strip()
else:
title2 = ""
if episode.license:
license = "cc/{}.svg".format(episode.license.lower())
else:
license = None
if episode.tags:
tags = episode.tags.split(',')
tag1 = tags[0]
else:
tags = []
tag1 = ''
"""
if ',' in authors:
authors = authors.split(', ')
author2 = ', '.join(authors[1:])
authors = authors[0].strip()
else:
author2 = ''
"""
author2 = ''
if episode.show.slug == 'pygotham_2015':
date = episode.start.strftime("%B %-d, %Y")
else:
date = episode.start.strftime("%Y-%m-%-d")
texts = {
'client': episode.show.client.name,
'show': episode.show.name,
'title': title,
'title2': title2,
'tag1': tag1,
'authors': authors,
'author2': author2,
'presentertitle': "",
'twitter_id': episode.twitter_id,
'date': date,
'time': episode.start.strftime("%H:%M"),
'license': license,
'room': episode.location.name,
}
return texts
def svg2png(self, svg_name, png_name, episode):
"""
Make a title slide png file.
melt uses librsvg which doesn't support flow,
wich is needed for long titles, so render it to a .png using inkscape
"""
# create png file
# inkscape does not return an error code on failure
# so clean up previous run and
# check for the existance of a new png
if os.path.exists(png_name):
os.remove(png_name)
# cmd=["inkscape", svg_name, "--export-png", png_name]
cmd = ["inkscape", svg_name,
"--export-png", png_name,
# "--export-width", "720",
]
ret = self.run_cmds(episode, [cmd])
ret = os.path.exists(png_name)
# if self.options.verbose: print cooked_svg
if self.options.verbose:
print png_name
if not ret:
print "svg:", svg_name
png_name = None
return png_name
def mk_title(self, episode):
# make a title slide
# if we find titles/custom/(slug).svg, use that
# else make one from the tempalte
custom_svg_name = os.path.join(
self.show_dir, "titles", "custom", episode.slug + ".svg")
if self.options.verbose: print "custom:", custom_svg_name
if os.path.exists(custom_svg_name):
cooked_svg_name = custom_svg_name
else:
svg_name = episode.show.client.title_svg
print svg_name
template = os.path.join(
os.path.split(os.path.abspath(__file__))[0],
"bling",
svg_name)
# happy_filename = episode.slug.encode('utf-8')
happy_filename = episode.slug
# happy_filename = ''.join([c for c in happy_filename if c.isalpha()])
title_base = os.path.join(self.show_dir, "titles", happy_filename)
raw_svg = open(template).read()
# tree=xml.etree.ElementTree.XMLID(raw_svg)
texts = self.get_title_text(episode)
cooked_svg = self.mk_title_svg(raw_svg, texts)
# save svg to a file
# strip 'broken' chars because inkscape can't handle the truth
# output_base=''.join([ c for c in output_base if c.isalpha()])
# output_base=''.join([ c for c in output_base if ord(c)<128])
# output_base=output_base.encode('utf-8','ignore')
cooked_svg_name = os.path.join(
self.show_dir, "titles", '%s.svg' % episode.slug)
open(cooked_svg_name, 'w').write(cooked_svg)
png_name = os.path.join(
self.show_dir, "titles", '%s.png' % episode.slug)
title_img = self.svg2png(cooked_svg_name, png_name, episode)
# title_img=png_name
if title_img is None:
print "missing title png"
return False
return title_img
def mkmlt_1(self, title_img, credits_img, episode, cls, rfs):
"""
assemble a mlt playlist from:
mlt template, title screen image,
filter parameters (currently just audio)
and cutlist+raw filenames
"""
# output file name
# this gets used twice:
# once to get melt to scan the files and count total frames of content
# then the trailer gets added to the xml
# and a final version gets written out. whacky.
mlt_pathname = os.path.join(self.work_dir, "%s.mlt" % episode.slug)
# parse the xml into a tree of nodes
# mlt_template_name = os.path.join(self.show_dir,"bling/chiweb/chiweb.mlt")
mlt_template_name = os.path.join(self.show_dir,
os.path.split(
os.path.abspath(__file__))[0],
"bling",
"chiweb/chiweb.mlt")
# print mlt_template_name
# mlt = open(mlt_template_name).read()
# print mlt
tree = xml.etree.ElementTree.XMLID(mlt)
# set the title to the title slide we just made
# import code; code.interact(local=locals())
title = tree[1]['title']
title.attrib['resource'] = title_img
# get the dvfile placeholder and remove it from the tree
dvfile = tree[1]['producer0']
tree[0].remove(dvfile)
# add in the dv files
pos = 1
for rf in rfs:
if self.options.load_temp:
src_pathname = os.path.join(self.episode_dir, rf.filename)
dst_path = os.path.join(
self.tmp_dir, episode.slug, os.path.dirname(rf.filename))
rawpathname = os.path.join(
self.tmp_dir, episode.slug, rf.filename)
# self.tmp_dir,episode.slug,rf.filename.replace(':','_'))
cmds = [['mkdir', '-p', dst_path],
['rsync', '--progress', '--size-only',
src_pathname, rawpathname]]
self.run_cmds(episode, cmds)
else:
if rf.filename.startswith('\\'):
rawpathname = rf.filename
else:
rawpathname = os.path.join(self.episode_dir, rf.filename)
# check for missing input file
# typically due to incorrect fs mount
if not os.path.exists(rawpathname):
print "can't find rawpathname", rawpathname
return False
dvfile.attrib['id'] = "producer%s" % rf.id
dvfile.attrib['resource'] = rawpathname
new = xml.etree.ElementTree.Element('producer', dvfile.attrib)
tree[0].insert(pos, new)
pos += 1
# get the dv clip placeholder, remove it from the playlist
clip = tree[1]['clip']
playlist = tree[1]['playlist1']
playlist.remove(clip)
# add in the clips
pos = 1
for cl in cls:
print cl
if cl.raw_file.duration:
print "duration:", cl.raw_file.duration
clip.attrib['id'] = "clip%s" % cl.id
clip.attrib['producer'] = "producer%s" % cl.raw_file.id
# set start/end on the clips if they are set in the db
# else remove them,
# ignoroe the error if they are not there to remove
if cl.start:
in_frame = time2f(cl.start, self.fps)
clip.attrib['in'] = str(in_frame)
else:
try:
del(clip.attrib['in'])
except KeyError:
pass
if self.options.test:
out_frame = "300"
clip.attrib['out'] = str(out_frame)
elif cl.end:
out_frame = time2f(cl.end, self.fps)
clip.attrib['out'] = str(out_frame)
else:
try:
del(clip.attrib['out'])
except KeyError:
pass
# add the new clip to the tree
print clip.attrib
new = xml.etree.ElementTree.Element('entry', clip.attrib)
playlist.insert(pos, new)
pos += 1
# write out the xml we have so far
# then pass it to melt to calc total frames
mlt_xml = xml.etree.ElementTree.tostring(tree[0])
open(mlt_pathname, 'w').write(mlt_xml)
p = subprocess.Popen(
['melt', mlt_pathname, '-consumer', 'xml'],
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
out, err = p.communicate()
# hack to remove "Plugin 1046 exists in both ...."
# out = '\n'.join(
# l for l in out.split('\n') if not l.startswith('Plugin') )
print out
print "err", err
t = xml.etree.ElementTree.XMLID(out)
frames = t[1]['tractor1'].get('out')
frames = int(frames)
# fade in/out the audio of the main track
# in is already set: 0-30 frames.
# out needs to be set to last 30 frames
fadeout = tree[1]['fadeout']
# fadeout.set("in",str(frames-30))
# fadeout.set("out",str(frames))
fadeout.set("in", "-120")
fadeout.set("out", "-90")
# add volume tweeks
"""
{'mlt_service':'volume',
'max_gain':'20',
'limiter':'20',
'normalise':self.options.normalise} )
"""
normalise = episode.normalise or '-12db'
if self.options.upload_formats == 'flac':
normalise = ''
if normalise and normalise != '0':
if self.options.verbose:
print "normalise:", normalise
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'volume',
'normalise': normalise})
playlist.insert(pos, new)
# default defined here:
# 01 is copy Left to Right
channelcopy = episode.channelcopy or \
episode.location.channelcopy or \
"01"
if channelcopy:
if self.options.verbose:
print 'channelcopy:', channelcopy
# channelcopy should be 01 or 10.
# or m/'mono' to kick in this hack
if channelcopy == 'm':
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'mono', 'channels': '2'})
else:
fro, to = list(channelcopy)
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'channelcopy',
'from': fro, 'to': to})
playlist.insert(pos, new)
if self.options.upload_formats == 'flac':
# mix channels to mono
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'mono', 'channels': '2'})
# this should be 1, but
# "service=mono channels=1" lowers pitch
# https://sourceforge.net/tracker/?func=detail&aid=2972735
playlist.insert(pos, new)
# super hack: remove a bunch of stuff that messes up flac
# like the title and transistion from title to cut
tree[0].remove(title)
x = tree[1]['playlist0']
# print x
tree[0].remove(x)
x = tree[1]['tractor0']
tree[0].remove(x)
if self.options.verbose:
xml.etree.ElementTree.dump(tree[0])
mlt_xml = xml.etree.ElementTree.tostring(tree[0])
open(mlt_pathname, 'w').write(mlt_xml)
if self.options.debug_log:
mlt_xml = mlt_xml.replace('<', '<').replace('>', '>')
mlt_xml = mlt_xml.replace('&', '&')
episode.description += "\n%s\n" % (mlt_xml,)
episode.save()
return mlt_pathname
def get_params(self, episode, rfs, cls):
"""
assemble a dict of params to send to mk_mlt
mlt template, title screen image,
filter parameters (currently just audio)
and cutlist+raw filenames
"""
def get_title(episode):
# if we find titles/custom/(slug).svg, use that
# else make one from the tempalte
custom_png_name = os.path.join(
self.show_dir, "titles", "custom", episode.slug + ".png")
print "custom:", custom_png_name
if os.path.exists(custom_png_name):
title_img = custom_png_name
else:
title_img = self.mk_title(episode)
return title_img
def get_foot(episode):
# define credits
credits_img = episode.show.client.credits \
if episode.show.client.credits \
else 'ndv-169.png'
credits_img = os.path.join(
os.path.split(os.path.abspath(__file__))[0],
"bling",
credits_img)
if credits_img[-4:] == ".svg":
# convert to png because melt doesn't do svgs as well
png_name = credits_img[:-4] + ".png"
credits_img = self.svg2png(credits_img, png_name, episode)
return credits_img
def get_clips(rfs):
"""
return list of possible input files
this may get the files and store them localy.
start/end segments are under get_cuts.
ps. this is not used for encoding,
just shows in ShotCut for easy dragging onto the timeline.
"""
clips = []
for rf in rfs:
# if rf.filename.startswith('\\'):
# rawpathname = rf.filename
# else:
rawpathname = os.path.join(
self.episode_dir, rf.filename)
# check for missing input file
# typically due to incorrect fs mount
if not os.path.exists(rawpathname):
print "can't find rawpathname", rawpathname
return False
clips.append({
'id': rf.id,
'filename': rawpathname,
})
return clips
def get_cuts(cls):
"""
gets the list of cuts.
input file, start, end, filters
ps, does not reference the clips above.
"""
def hms_to_clock(hms):
"""
Converts what media players show h:m:s
to the mlt time format h:m:s.s
for more on this:
http://mltframework.blogspot.com/2012/04/time-properties.html
"""
if not hms:
return None
if ":" not in hms:
hms = "0:" + hms
if "." not in hms:
hms = hms + ".0"
return hms
cuts = []
for cl in cls:
cut = {}
cut['id'] = cl.id
rawpathname = os.path.join(
self.episode_dir, cl.raw_file.filename)
cut['filename'] = rawpathname
# set start/end on the clips if they are set in the db
# else None
cut['in']=hms_to_clock(cl.start)
cut['out']=hms_to_clock(cl.end)
cut['length'] = cl.duration()
# cut['length'] = cl.duration()
if cl.episode.channelcopy:
cut['channelcopy'] = cl.episode.channelcopy
else:
cut['channelcopy']='01'
if cl.episode.normalise:
cut['normalize'] = cl.episode.normalise
else:
cut['normalize']='-12.0'
cuts.append(cut)
return cuts
params = {}
params['title_img'] = get_title(episode)
params['foot_img'] = get_foot(episode)
params['clips'] = get_clips(rfs)
params['cuts'] = get_cuts(cls)
return params
# then the trailer gets added to the xml
# and a final version gets written out. whacky.
mlt_pathname = os.path.join(self.work_dir, "%s.mlt" % episode.slug)
# parse the xml into a tree of nodes
# mlt_template_name = os.path.join(self.show_dir,"bling/chiweb/chiweb.mlt")
mlt_template_name = os.path.join(self.show_dir,
os.path.split(
os.path.abspath(__file__))[0],
"bling",
"chiweb/chiweb.mlt")
# print mlt_template_name
# mlt = open(mlt_template_name).read()
# print mlt
tree = xml.etree.ElementTree.XMLID(mlt)
# set the title to the title slide we just made
# import code; code.interact(local=locals())
title = tree[1]['title']
title.attrib['resource'] = title_img
# get the dvfile placeholder and remove it from the tree
dvfile = tree[1]['producer0']
tree[0].remove(dvfile)
# add in the dv files
pos = 1
for rf in rfs:
if self.options.load_temp:
src_pathname = os.path.join(self.episode_dir, rf.filename)
dst_path = os.path.join(
self.tmp_dir, episode.slug, os.path.dirname(rf.filename))
rawpathname = os.path.join(
self.tmp_dir, episode.slug, rf.filename)
# self.tmp_dir,episode.slug,rf.filename.replace(':','_'))
cmds = [['mkdir', '-p', dst_path],
['rsync', '--progress', '--size-only',
src_pathname, rawpathname]]
self.run_cmds(episode, cmds)
else:
if rf.filename.startswith('\\'):
rawpathname = rf.filename
else:
rawpathname = os.path.join(
self.episode_dir, rf.filename)
# check for missing input file
# typically due to incorrect fs mount
if not os.path.exists(rawpathname):
print "can't find rawpathname", rawpathname
return False
dvfile.attrib['id'] = "producer%s" % rf.id
dvfile.attrib['resource'] = rawpathname
new = xml.etree.ElementTree.Element('producer', dvfile.attrib)
tree[0].insert(pos, new)
pos += 1
# get the dv clip placeholder, remove it from the playlist
clip = tree[1]['clip']
playlist = tree[1]['playlist1']
playlist.remove(clip)
# add in the clips
pos = 1
for cl in cls:
print cl
if cl.raw_file.duration:
print "duration:", cl.raw_file.duration
clip.attrib['id'] = "clip%s" % cl.id
clip.attrib['producer'] = "producer%s" % cl.raw_file.id
# set start/end on the clips if they are set in the db
# else remove them,
# ignoroe the error if they are not there to remove
if cl.start:
in_frame = time2f(cl.start, self.fps)
clip.attrib['in'] = str(in_frame)
else:
try:
del(clip.attrib['in'])
except KeyError:
pass
if self.options.test:
out_frame = "300"
clip.attrib['out'] = str(out_frame)
elif cl.end:
out_frame = time2f(cl.end, self.fps)
clip.attrib['out'] = str(out_frame)
else:
try:
del(clip.attrib['out'])
except KeyError:
pass
# add the new clip to the tree
print clip.attrib
new = xml.etree.ElementTree.Element('entry', clip.attrib)
playlist.insert(pos, new)
pos += 1
# write out the xml we have so far
# then pass it to melt to calc total frames
mlt_xml = xml.etree.ElementTree.tostring(tree[0])
open(mlt_pathname, 'w').write(mlt_xml)
p = subprocess.Popen(
['melt', mlt_pathname, '-consumer', 'xml'],
stdout=subprocess.PIPE, stderr=subprocess.PIPE)
out, err = p.communicate()
# hack to remove "Plugin 1046 exists in both ...."
# out = '\n'.join(
# l for l in out.split('\n') if not l.startswith('Plugin') )
print out
print "err", err
t = xml.etree.ElementTree.XMLID(out)
frames = t[1]['tractor1'].get('out')
frames = int(frames)
# fade in/out the audio of the main track
# in is already set: 0-30 frames.
# out needs to be set to last 30 frames
fadeout = tree[1]['fadeout']
# fadeout.set("in",str(frames-30))
# fadeout.set("out",str(frames))
fadeout.set("in", "-120")
fadeout.set("out", "-90")
# add volume tweeks
"""
{'mlt_service':'volume',
'max_gain':'20',
'limiter':'20',
'normalise':self.options.normalise} )
"""
normalise = episode.normalise or '-12db'
if self.options.upload_formats == 'flac':
normalise = ''
if normalise and normalise != '0':
if self.options.verbose:
print "normalise:", normalise
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'volume',
'normalise': normalise})
playlist.insert(pos, new)
# default defined here:
# 01 is copy Left to Right
channelcopy = episode.channelcopy or \
episode.location.channelcopy or \
"01"
if channelcopy:
if self.options.verbose:
print 'channelcopy:', channelcopy
# channelcopy should be 01 or 10.
# or m/'mono' to kick in this hack
if channelcopy == 'm':
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'mono', 'channels': '2'})
else:
fro, to = list(channelcopy)
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'channelcopy',
'from': fro, 'to': to})
playlist.insert(pos, new)
if self.options.upload_formats == 'flac':
# mix channels to mono
new = xml.etree.ElementTree.Element('filter',
{'mlt_service': 'mono', 'channels': '2'})
# this should be 1, but
# "service=mono channels=1" lowers pitch
# https://sourceforge.net/tracker/?func=detail&aid=2972735
playlist.insert(pos, new)
# super hack: remove a bunch of stuff that messes up flac
# like the title and transistion from title to cut
tree[0].remove(title)
x = tree[1]['playlist0']
# print x
tree[0].remove(x)
x = tree[1]['tractor0']
tree[0].remove(x)
if self.options.verbose:
xml.etree.ElementTree.dump(tree[0])
mlt_xml = xml.etree.ElementTree.tostring(tree[0])
open(mlt_pathname, 'w').write(mlt_xml)
if self.options.debug_log:
mlt_xml = mlt_xml.replace('<', '<').replace('>', '>')
mlt_xml = mlt_xml.replace('&', '&')
episode.description += "\n%s\n" % (mlt_xml,)
episode.save()
return mlt_pathname
def enc_all(self, mlt_pathname, episode):
def enc_one(ext):
out_pathname = os.path.join(
self.show_dir, ext, "%s.%s" % (episode.slug, ext))
work_pathname = os.path.join(
self.show_dir, "work", "%s.dv" % (episode.slug))
stats_pathname = os.path.join(
self.show_dir, "tmp", "%s.dat" % (episode.slug))
wav_pathname = os.path.join(
self.show_dir, "tmp", "%s.wav" % (episode.slug))
norm_pathname = os.path.join(
self.show_dir, "tmp", "%s-normal.wav" % (episode.slug))
if ext == 'webm':
parms = {
'dv_format': self.options.dv_format,
'mlt': mlt_pathname,
'norm': norm_pathname,
'out': out_pathname,
'threads': self.options.threads,
'statsfile': stats_pathname,
'wav': wav_pathname,
'work': work_pathname,
'test': '',
}
# cmds=["melt %s -profile dv_ntsc -consumer avformat:%s progress=1 acodec=libvorbis ab=128k ar=44100 vcodec=libvpx minrate=0 b=600k aspect=@4/3 maxrate=1800k g=120 qmax=42 qmin=10"% (mlt_pathname,out_pathname,)]
#cmds = [
# "melt -profile %(dv_format)s %(mlt)s force_aspect_ratio=@64/45 -consumer avformat:%(out)s progress=1 threads=0 ab=256k vb=2000k quality=good deadline=good deinterlace=1 deinterlace_method=yadif" % parms]
cmds = [
"melt -profile %(dv_format)s %(mlt)s force_aspect_ratio=@64/45 -consumer avformat:%(work)s progress=1 threads=%(threads)s" % parms,
"gst-launch-1.0 webmmux name=mux ! fakesink uridecodebin uri=file://%(work)s name=demux demux. ! videoconvert ! deinterlace ! vp8enc multipass-cache-file=%(statsfile)s multipass-mode=1 threads=%(threads)s ! queue ! mux.video_0 demux. ! progressreport ! audioconvert ! audiorate ! tee name=t ! queue ! vorbisenc ! queue ! mux.audio_0 t. ! queue ! wavenc ! filesink location=%(wav)s " % parms,
"sox --norm %(wav)s %(norm)s" % parms,
"gst-launch-1.0 webmmux name=mux ! filesink location=%(out)s uridecodebin uri=file://%(work)s name=video uridecodebin uri=file://%(wav)s name=audio video. ! videoconvert ! deinterlace ! vp8enc multipass-cache-file=%(statsfile)s multipass-mode=2 threads=%(threads)s ! queue ! mux.video_0 audio. ! progressreport ! audioconvert ! audiorate ! vorbisenc ! queue ! mux.audio_0" % parms,
"rm %(wav)s %(norm)s %(statsfile)s %(work)s" % parms ]
if ext == 'flv':
cmds = [
"melt %(mlt)s -progress -profile %(dv_format)s -consumer avformat:%(out)s progressive=1 acodec=libfaac ab=96k ar=44100 vcodec=libx264 b=110k vpre=/usr/share/ffmpeg/libx264-hq.ffpreset" % parms]
if ext == 'flac':
# 16kHz/mono
cmds = ["melt -verbose -progress %s -consumer avformat:%s ar=16000" %
(mlt_pathname, out_pathname)]
if ext == 'mp3':
cmds = ["melt -verbose -progress %s -consumer avformat:%s" %
(mlt_pathname, out_pathname)]
if ext == 'mp4':
# High Quality Master 720x480 NTSC
parms = {
'dv_format': self.options.dv_format,
'mlt': mlt_pathname,
'out': out_pathname,
'threads': self.options.threads,
'test': '',
}
cmd = "melt -verbose -progress "\
"-profile %(dv_format)s %(mlt)s "\
"-consumer avformat:%(out)s "\
"threads=%(threads)s "\
"progressive=1 "\
"strict=-2 "\
"properties=x264-medium "\
"ab=256k "\
% parms
cmd = cmd.split()
# 2 pass causes no video track, so dumping this.
# need to figure out how to switch between good and fast
if False:
cmds = [cmd + ['pass=1'],
cmd + ['pass=2']]
if True: # even faster!
cmds[0].append('fastfirstpass=1')
else:
cmds = [cmd]
# cmds.append( ["qt-faststart", tmp_pathname, out_pathname] )
if self.options.rm_temp:
cmds.append(["rm", tmp_pathname])
if ext == 'm4v':
# iPhone
tmp_pathname = os.path.join(
self.tmp_dir, "%s.%s" % (episode.slug, ext))
# combine settings from 2 files
ffpreset = open(
'/usr/share/ffmpeg/libx264-default.ffpreset').read().split('\n')
ffpreset.extend(
open('/usr/share/ffmpeg/libx264-ipod640.ffpreset').read().split('\n'))
ffpreset = [i for i in ffpreset if i]
cmd = "melt %(mlt)s -progress -profile %(dv_format)s -consumer avformat:%(tmp)s s=432x320 aspect=@4/3 progressive=1 acodec=libfaac ar=44100 ab=128k vcodec=libx264 b=70k" % parms
cmd = cmd.split()
cmd.extend(ffpreset)
cmds = [cmd]
cmds.append(["qt-faststart", tmp_pathname, out_pathname])
if self.options.rm_temp:
cmds.append(["rm", tmp_pathname])
if ext == 'dv':
out_pathname = os.path.join(
self.tmp_dir, "%s.%s" % (episode.slug, ext))
cmds = ["melt -verbose -progress %s -consumer avformat:%s pix_fmt=yuv411p progressive=1" %
(mlt_pathname, out_pathname)]
if ext == 'ogv':
# melt/ffmpeg ogv encoder is loopy,
# so make a .dv and pass it to ffmpeg2theora
ret = enc_one("dv")
if ret:
dv_pathname = os.path.join(
self.tmp_dir, "%s.dv" % (episode.slug,))
cmds = [
"ffmpeg2theora --videoquality 5 -V 600 --audioquality 5 --channels 1 %s -o %s" % (dv_pathname, out_pathname)]
if self.options.rm_temp:
cmds.append(["rm", dv_pathname])
else:
return ret
# run encoder:
ret = self.run_cmds(episode, cmds)
if ret and not os.path.exists(out_pathname):
print "melt returned %ret, but no output: %s" % \
(ret, out_pathname)
ret = False
return ret
ret = True
# create all the formats for uploading
for ext in self.options.upload_formats:
print "encoding to %s" % (ext,)
ret = enc_one(ext) and ret
if self.options.enc_script:
cmd = [self.options.enc_script,
self.show_dir, episode.slug]
ret = ret and self.run_cmds(episode, [cmd])
return ret
def dv2theora(self, episode, dv_path_name, cls, rfs):
"""
transcode dv to ogv
"""
oggpathname = os.path.join(
self.show_dir, "ogv", "%s.ogv" % episode.slug)
# cmd="ffmpeg2theora --videoquality 5 -V 600 --audioquality 5 --speedlevel 0 --optimize --keyint 256 --channels 1".split()
cmd = "ffmpeg2theora --videoquality 5 -V 600 --audioquality 5 --keyint 256 --channels 1".split()
cmd += ['--output', oggpathname]
cmd += [dv_path_name]
return cmd
def process_ep(self, episode):
# print episode
print episode.name
ret = False
cls = Cut_List.objects.filter(
episode=episode, apply=True).order_by('sequence')
# print len(cls), episode.name.__repr__()
if cls:
"""
for cl in cls:
print cl.start, cl.end
# title and footer were here
title_img = self.mk_title(episode)
credits_img = episode.show.client.credits \
if episode.show.client.credits \
else 'ndv1-black.png'
credits_img = os.path.join(
os.path.split(os.path.abspath(__file__))[0],
"bling",
credits_img)
"""
# get list of raw footage for this episode
rfs = Raw_File.objects. \
filter(cut_list__episode=episode).\
exclude(trash=True).distinct()
# cut_list__apply=True).\
# get a .mlt file for this episode (mlt_pathname)
# look for custom/slug.mlt and just use it,
# else build one from client.template_mlt, else "template.mlt"
mlt_pathname = os.path.join(self.work_dir, "custom", "%s.mlt" % episode.slug)
if os.path.exists(mlt_pathname):
ret = True
else:
mlt_pathname = os.path.join(self.work_dir, "%s.mlt" % episode.slug)
if episode.show.client.template_mlt:
template_mlt = episode.show.client.template_mlt
else:
template_mlt = "template.mlt"
params = self.get_params(episode, rfs, cls )
ret = mk_mlt( template_mlt, mlt_pathname, params )
if not ret:
episode.state = 0
episode.comment += "\nenc.py mlt = self.mkmlt_1 failed.\n"
episode.save()
return False
# do the final encoding:
# using melt
ret = self.enc_all(mlt_pathname, episode)
if self.options.load_temp and self.options.rm_temp:
cmds = []
for rf in rfs:
dst_path = os.path.join(
self.tmp_dir, episode.slug, os.path.dirname(rf.filename))
rawpathname = os.path.join(
self.tmp_dir, episode.slug, rf.filename)
cmds.append(['rm', rawpathname])
cmds.append(['rmdir', dst_path])
dst_path = os.path.join(self.tmp_dir, episode.slug)
cmds.append(['rmdir', dst_path])
self.run_cmds(episode, cmds)
else:
err_msg = "No cutlist found."
episode.state = 0
episode.comment += "\nenc error: %s\n" % (err_msg,)
episode.save()
print err_msg
return False
if self.options.test:
ret = False
# save the episode so the test suite can get the slug
self.episode = episode
return ret
def add_more_options(self, parser):
parser.add_option('--enc-script',
help='encode shell script')
parser.add_option('--load-temp', action="store_true",
help='copy .dv to temp files')
parser.add_option('--rm-temp',
help='remove large temp files')
parser.add_option('--threads')
def add_more_option_defaults(self, parser):
parser.set_defaults(threads=2)
if __name__ == '__main__':
p = enc()
p.main()