cablegates/pandora/archive/extract.py

424 lines
13 KiB
Python
Raw Normal View History

2010-09-04 12:59:09 +00:00
# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
from __future__ import division, with_statement
import os
2011-01-01 11:44:42 +00:00
from os.path import exists
2010-09-04 12:59:09 +00:00
import fractions
import subprocess
import tempfile
import time
import math
from glob import glob
import numpy as np
import Image
2011-01-25 15:14:25 +00:00
import ox
2010-09-04 12:59:09 +00:00
img_extension='jpg'
FFMPEG2THEORA = 'ffmpeg2theora'
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
class AspectRatio(fractions.Fraction):
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def __new__(cls, numerator, denominator=None):
if not denominator:
ratio = map(int, numerator.split(':'))
2011-01-01 11:44:42 +00:00
if len(ratio) == 1:
ratio.append(1)
2010-09-04 12:59:09 +00:00
numerator = ratio[0]
denominator = ratio[1]
#if its close enough to the common aspect ratios rather use that
if abs(numerator/denominator - 4/3) < 0.03:
numerator = 4
denominator = 3
elif abs(numerator/denominator - 16/9) < 0.02:
numerator = 16
denominator = 9
return super(AspectRatio, cls).__new__(cls, numerator, denominator)
@property
def ratio(self):
return "%d:%d" % (self.numerator, self.denominator)
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def stream(video, target, profile, info):
if not os.path.exists(target):
fdir = os.path.dirname(target)
if not os.path.exists(fdir):
os.makedirs(fdir)
dar = AspectRatio(info['video'][0]['display_aspect_ratio'])
'''
WebM look into
lag
mb_static_threshold
qmax/qmin
rc_buf_aggressivity=0.95
token_partitions=4
level / speedlevel
bt?
H264, should bitrates be a bit lower? other stuff possible?
'''
profile, format = profile.split('.')
if profile == '1080p':
height = 1080
audiorate = 48000
audioquality = 6
audiobitrate = None
audiochannels = None
if profile == '720p':
height = 720
audiorate = 48000
audioquality = 5
audiobitrate = None
audiochannels = None
if profile == '480p':
height = 480
audiorate = 44100
audioquality = 2
audiobitrate = None
audiochannels = 2
elif profile == '360p':
height = 360
audiorate = 44100
audioquality = 1
audiobitrate = None
audiochannels = 1
2010-12-28 10:15:45 +00:00
elif profile == '240p':
height = 240
audiorate = 44100
2010-09-04 12:59:09 +00:00
audioquality = 0
audiobitrate = None
audiochannels = 1
else:
height = 96
audiorate = 22050
audioquality = -1
audiobitrate = '22k'
audiochannels = 1
bpp = 0.17
fps = AspectRatio(info['video'][0]['framerate'])
2011-01-01 11:44:42 +00:00
width = int(dar * height)
width += width % 2
2010-09-04 12:59:09 +00:00
bitrate = height*width*fps*bpp/1000
aspect = dar.ratio
#use 1:1 pixel aspect ratio if dar is close to that
if abs(width/height - dar) < 0.02:
aspect = '%s:%s' % (width, height)
if info['audio']:
audio_settings = ['-ar', str(audiorate), '-aq', str(audioquality)]
if audiochannels and 'channels' in info['audio'][0] and info['audio'][0]['channels'] > audiochannels:
audio_settings += ['-ac', str(audiochannels)]
if audiobitrate:
audio_settings += ['-ab', audiobitrate]
if format == 'mp4':
audio_settings += ['-acodec', 'libfaac']
else:
audio_settings += ['-acodec', 'libvorbis']
else:
audio_settings = ['-an']
if info['video']:
video_settings = [
'-vb', '%dk'%bitrate, '-g', '%d' % int(fps*2),
'-s', '%dx%d'%(width, height),
'-aspect', aspect,
]
if format == 'mp4':
#quicktime does not support bpyramid
'''
2010-09-04 12:59:09 +00:00
video_settings += [
'-vcodec', 'libx264',
'-flags', '+loop+mv4',
'-cmp', '256',
'-partitions', '+parti4x4+parti8x8+partp4x4+partp8x8+partb8x8',
'-me_method', 'hex',
'-subq', '7',
'-trellis', '1',
'-refs', '5',
'-bf', '3',
'-flags2', '+bpyramid+wpred+mixed_refs+dct8x8',
'-coder', '1',
'-me_range', '16',
'-keyint_min', '25', #FIXME: should this be related to fps?
'-sc_threshold','40',
'-i_qfactor', '0.71',
'-qmin', '10', '-qmax', '51',
'-qdiff', '4'
]
'''
video_settings += [
'-vcodec', 'libx264',
'-flags', '+loop+mv4',
'-cmp', '256',
'-partitions', '+parti4x4+parti8x8+partp4x4+partp8x8+partb8x8',
'-me_method', 'hex',
'-subq', '7',
'-trellis', '1',
'-refs', '5',
'-bf', '0',
'-flags2', '+mixed_refs',
'-coder', '0',
'-me_range', '16',
'-g', '250', #FIXME: should this be related to fps?
'-keyint_min', '25', #FIXME: should this be related to fps?
2011-01-01 11:44:42 +00:00
'-sc_threshold', '40',
'-i_qfactor', '0.71',
'-qmin', '10', '-qmax', '51',
'-qdiff', '4'
]
2010-09-04 12:59:09 +00:00
else:
video_settings = ['-vn']
ffmpeg = FFMPEG2THEORA.replace('2theora', '')
cmd = [ffmpeg, '-y', '-threads', '2', '-i', video] \
+ audio_settings \
+ video_settings
2011-02-06 12:40:28 +00:00
if format == 'webm':
cmd += ['-f', 'webm', target]
2010-09-04 12:59:09 +00:00
if format == 'mp4':
2011-02-06 12:40:28 +00:00
#mp4 needs postprocessing(qt-faststart), write to temp file
2010-09-04 12:59:09 +00:00
cmd += ["%s.mp4"%target]
2011-02-06 12:40:28 +00:00
else :
cmd += [target]
2010-09-04 12:59:09 +00:00
print cmd
2010-09-15 13:03:00 +00:00
p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=open('/dev/null', 'w'), stderr=subprocess.STDOUT)
p.communicate()
2010-09-04 12:59:09 +00:00
if format == 'mp4':
2011-01-01 11:44:42 +00:00
cmd = ['qt-faststart', "%s.mp4"%target, target]
2010-09-04 12:59:09 +00:00
print cmd
2010-09-15 13:03:00 +00:00
p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=open('/dev/null', 'w'), stderr=subprocess.STDOUT)
p.communicate()
2010-09-04 12:59:09 +00:00
os.unlink("%s.mp4"%target)
2010-09-15 13:03:00 +00:00
return True
2010-09-04 12:59:09 +00:00
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def run_command(cmd, timeout=10):
#print cmd
2010-09-15 13:03:00 +00:00
p = subprocess.Popen(cmd, stdout=open('/dev/null', 'w'), stderr=subprocess.STDOUT)
2010-09-04 12:59:09 +00:00
while timeout > 0:
time.sleep(0.2)
timeout -= 0.2
if p.poll() != None:
return p.returncode
if p.poll() == None:
os.kill(p.pid, 9)
killedpid, stat = os.waitpid(p.pid, os.WNOHANG)
return p.returncode
2011-01-01 11:44:42 +00:00
def frame(videoFile, frame, position, width=128, redo=False):
2010-09-04 12:59:09 +00:00
'''
params:
videoFile input
frame output
2010-09-04 12:59:09 +00:00
position as float in seconds
width of frame
redo boolean to extract file even if it exists
'''
if exists(videoFile):
frameFolder = os.path.dirname(frame)
2010-09-04 12:59:09 +00:00
if redo or not exists(frame):
if not exists(frameFolder):
os.makedirs(frameFolder)
cmd = ['oxframe', '-i', videoFile, '-o', frame, '-p', str(position), '-x', str(width)]
2010-09-04 12:59:09 +00:00
run_command(cmd)
2011-01-01 11:44:42 +00:00
2010-09-07 14:05:38 +00:00
def resize_image(image_source, image_output, width=None, size=None):
2010-09-04 12:59:09 +00:00
if exists(image_source):
source = Image.open(image_source).convert('RGB')
2010-09-04 12:59:09 +00:00
source_width = source.size[0]
source_height = source.size[1]
2010-09-07 14:05:38 +00:00
if size:
if source_width > source_height:
width = size
height = int(width / (float(source_width) / source_height))
height = height - height % 2
else:
height = size
width = int(height * (float(source_width) / source_height))
width = width - width % 2
2011-01-01 11:44:42 +00:00
2010-09-07 14:05:38 +00:00
else:
height = int(width / (float(source_width) / source_height))
height = height - height % 2
2010-09-04 12:59:09 +00:00
if width < source_width:
resize_method = Image.ANTIALIAS
else:
resize_method = Image.BICUBIC
output = source.resize((width, height), resize_method)
output.save(image_output)
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def timeline(video, prefix):
cmd = ['oxtimeline', '-i', video, '-o', prefix]
p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
p.wait()
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def average_color(prefix):
height = 64
width = 1500
frames = 0
pixels = []
color = np.asarray([0, 0, 0], dtype=np.float32)
for image in sorted(glob("%s.%d.*.png" % (prefix, height))):
timeline = Image.open(image)
frames += timeline.size[0]
p = np.asarray(timeline, dtype=np.float32)
p = np.sum(p, axis=0) / height #average color per frame
pixels.append(p)
for i in range(0, len(pixels)):
p = np.sum(pixels[i], axis=0) / frames
color += p
2011-01-25 15:14:25 +00:00
color = list(map(float, color))
return ox.image.getHSL(color)
2010-09-04 12:59:09 +00:00
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def get_distance(rgb0, rgb1):
dst = math.sqrt(pow(rgb0[0] - rgb1[0], 2) + pow(rgb0[0] - rgb1[0], 2) + pow(rgb0[0] - rgb1[0], 2))
return dst / math.sqrt(3 * pow(255, 2))
2011-01-01 11:44:42 +00:00
2010-09-04 12:59:09 +00:00
def cuts(prefix):
cuts = []
fps = 25
frames = 0
height = 64
width = 1500
pixels = []
for image in sorted(glob("%s.%d.*.png" % (prefix, height))):
timeline = Image.open(image)
frames += timeline.size[0]
pixels.append(timeline.load())
for frame in range(0, frames):
x = frame % width
if frame > 0:
dst = 0
image0 = int((frame - 1) / width)
image1 = int(frame / width)
for y in range(0, height):
rgb0 = pixels[image0][(x - 1) % width, y]
rgb1 = pixels[image1][x, y]
dst += get_distance(rgb0, rgb1) / height
#print frame / fps, dst
if dst > 0.1:
cuts.append(frame / fps)
return cuts
2011-01-01 11:44:42 +00:00
2010-09-08 13:38:44 +00:00
def divide(num, by):
# >>> divide(100, 3)
# [33, 33, 34]
arr = []
div = int(num / by)
mod = num % by
2010-09-08 17:14:01 +00:00
for i in range(int(by)):
2010-09-08 13:38:44 +00:00
arr.append(div + (i > by - 1 - mod))
return arr
2011-01-01 11:44:42 +00:00
2010-09-23 16:01:48 +00:00
def timeline_strip(item, cuts, info, prefix):
2010-09-08 13:38:44 +00:00
_debug = False
2011-01-01 11:44:42 +00:00
duration = info['duration']
2010-09-08 13:38:44 +00:00
video_height = info['video'][0]['height']
video_width = info['video'][0]['width']
video_ratio = video_width / video_height
line_image = []
timeline_height = 64
timeline_width = 1500
fps = 25
2010-09-08 17:14:01 +00:00
frames = int(duration * fps)
2010-09-08 13:38:44 +00:00
if cuts[0] != 0:
cuts.insert(0, 0)
cuts = map(lambda x: int(round(x * fps)), cuts)
for frame in range(frames):
i = int(frame / timeline_width)
x = frame % timeline_width
if x == 0:
timeline_width = min(timeline_width, frames - frame)
timeline_image = Image.new('RGB', (timeline_width, timeline_height))
if frame in cuts:
c = cuts.index(frame)
2010-09-08 17:14:01 +00:00
if c +1 < len(cuts):
duration = cuts[c + 1] - cuts[c]
2010-09-23 16:01:48 +00:00
frames = math.ceil(duration / (video_width * timeline_height / video_height))
widths = divide(duration, frames)
frame = frame
2010-09-08 13:38:44 +00:00
if _debug:
2010-09-23 16:01:48 +00:00
print widths, duration, frames, cuts[c], cuts[c + 1]
for s in range(int(frames)):
frame_ratio = widths[s] / timeline_height
if video_ratio > frame_ratio:
width = int(round(video_height * frame_ratio))
2010-09-08 17:14:01 +00:00
left = int((video_width - width) / 2)
box = (left, 0, left + width, video_height)
else:
2010-09-23 16:01:48 +00:00
height = int(round(video_width / frame_ratio))
2010-09-08 17:14:01 +00:00
top = int((video_height - height) / 2)
box = (0, top, video_width, top + height)
if _debug:
2010-09-23 16:01:48 +00:00
print frame, 'cut', c, 'frame', s, frame, 'width', widths[s], box
#FIXME: why does this have to be frame+1?
frame_image = Image.open(item.frame((frame+1)/fps))
2010-09-08 17:14:01 +00:00
frame_image = frame_image.crop(box).resize((widths[s], timeline_height), Image.ANTIALIAS)
for x_ in range(widths[s]):
line_image.append(frame_image.crop((x_, 0, x_ + 1, timeline_height)))
2010-09-23 16:01:48 +00:00
frame += widths[s]
2010-09-08 17:14:01 +00:00
if len(line_image) > frame:
timeline_image.paste(line_image[frame], (x, 0))
2010-09-08 13:38:44 +00:00
if x == timeline_width - 1:
timeline_file = '%sstrip.64.%04d.png' % (prefix, i)
if _debug:
print 'writing', timeline_file
timeline_image.save(timeline_file)
2011-01-01 11:44:42 +00:00
def chop(video, start, end):
t = end - start
tmp = tempfile.mkdtemp()
ext = os.path.splitext(video)[1]
choped_video = '%s/tmp%s' % (tmp, ext)
cmd = [
'ffmpeg',
'-y',
'-i', video,
'-ss', '%.3f'%start,
2011-01-01 11:44:42 +00:00
'-t', '%.3f'%t,
'-vcodec', 'copy',
'-acodec', 'copy',
'-f', ext[1:],
choped_video
]
p = subprocess.Popen(cmd, stdin=subprocess.PIPE,
stdout=open('/dev/null', 'w'),
stderr=open('/dev/null', 'w'))
p.wait()
f = open(choped_video, 'r')
os.unlink(choped_video)
os.rmdir(tmp)
return f