def export(outdir=None, clean=False):
# Unpack dataset
cache = Cache(cacheroot=outdir, subdir=VISET)
outfile = cache.abspath('%s.csv' % VISET)
if clean:
cache.clean()
elif os.path.isfile(outfile):
print '[bobo.viset.kthactions]: exporting "%s"' % outfile
return outfile
print '[bobo.viset.kthactions][WARNING]: downloads will not show percent progress since content length is unknown'
for (url, label, sha1) in zip(URLS, LABELS, SHA1):
cache.unpack(cache.get(url, sha1), cache.abspath(label), cleanup=False)
# Check for frame export utility
#if not isexe('ffmpeg'):
# raise IOError('[bobo.viset.kthactions]: ffmpeg not found on path')
# Video list
with open(outfile, 'wb') as csvfile:
f = csv.writer(csvfile, delimiter=' ', quotechar='|', quoting=csv.QUOTE_MINIMAL)
for (idx_category, category) in enumerate(os.listdir(cache.root())):
if os.path.isdir(os.path.join(cache.root(), category)):
for (idx_video, filename) in enumerate(os.listdir(os.path.join(cache.root(), category))):
[avibase,ext] = os.path.splitext(filename)
if ext == '.avi':
imdir = cache.abspath(os.path.join(category, avibase))
print '[bobo.viset.kthactions]: exporting "%s" to "%s"' % (filename, imdir)
f.writerow([os.path.join(category, avibase, 'im_%08d.png'), category]);
cmd = "ffmpeg -i \'%s\' %s/im_%%08d.png &> /dev/null" % (os.path.join(cache.root(), category, filename), imdir)
remkdir(imdir)
if os.system(cmd) != 0:
raise IOError('Error running ffmpeg')
return outfile
def youtube(tag, n_pages=1, outdir=None):
url = 'https://www.youtube.com/results?search_query=%s&page=%d'
vidlist = []
for k in range(0, n_pages):
user_agent = random.choice(common_user_agents)
headers = {'User-Agent':user_agent}
search_request = urllib2.Request(url % (tag.replace(' ','+'), k+1), None, headers)
search_results = urllib2.urlopen(search_request)
search_data = search_results.read()
datalist = search_data.split('href="/watch?')
vidlist.extend(['https://www.youtube.com/watch?%s' % vid.split('"')[0] for vid in datalist if 'DOCTYPE' not in vid.split('"')[0]])
vidlist = list(set(vidlist)) # unique
if outdir is not None:
download(vidlist, os.path.join(remkdir(outdir), 'youtube_'+tofilename(tag)+'_%04d.mp4'))
return(vidlist)
def ustream(tag, n_pages=1, outdir=None):
url = 'http://www.ustream.tv/search?q=%s'
vidlist = []
for k in range(0, 1):
user_agent = random.choice(common_user_agents)
headers = {'User-Agent':user_agent}
search_request = urllib2.Request(url % (tag.replace(' ','+')), None, headers)
search_results = urllib2.urlopen(search_request)
search_data = search_results.read()
datalist = search_data.split('href="/recorded/')
vidlist.extend(['http://www.ustream.tv/recorded/%s' % vid.split('"')[0] for vid in datalist if 'DOCTYPE' not in vid.split('"')[0]])
vidlist = list(set(vidlist)) # unique
print vidlist
if outdir is not None:
download(vidlist, os.path.join(remkdir(outdir), 'ustream_'+tofilename(tag)+'_%04d.mp4'))
return(vidlist)
def frames(viddir):
for v in videolist(viddir):
(outdir, ext) = os.path.splitext(v)
cmd = "ffmpeg -i \'%s\' %s/%%08d.png &> /dev/null" % (v, remkdir(outdir))
print '[bobo.viset.videodir]: exporting frames from "%s" to "%s"' % (v, outdir)
os.system(cmd)
def setroot(self, path):
self._cacheroot = path
remkdir(self._cacheroot)
def frames(viddir):
for v in videolist(viddir):
(outdir, ext) = os.path.splitext(v)
cmd = 'ffmpeg -r 25 -i \'%s\' -qscale:v 2 -vf "scale=-1:240" %s/%%08d.jpg &> /dev/null' % (v, remkdir(outdir))
print '[bobo.viset.videodir]: exporting frames from "%s" to "%s"' % (v, outdir)
os.system(cmd)