2014-01-07 10:04:48 +01:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
2013-08-28 02:18:44 +02:00
|
|
|
import re
|
2013-09-29 20:49:58 +02:00
|
|
|
import json
|
2013-08-28 02:18:44 +02:00
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
from ..utils import (
|
2013-09-29 20:49:58 +02:00
|
|
|
compat_urlparse,
|
2013-08-28 02:18:44 +02:00
|
|
|
determine_ext,
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class AppleTrailersIE(InfoExtractor):
|
2013-12-04 14:34:47 +01:00
|
|
|
_VALID_URL = r'https?://(?:www\.)?trailers\.apple\.com/trailers/(?P<company>[^/]+)/(?P<movie>[^/]+)'
|
2013-08-28 02:18:44 +02:00
|
|
|
_TEST = {
|
2014-01-07 10:04:48 +01:00
|
|
|
"url": "http://trailers.apple.com/trailers/wb/manofsteel/",
|
|
|
|
|
"playlist": [
|
2013-08-28 02:18:44 +02:00
|
|
|
{
|
2014-01-07 10:04:48 +01:00
|
|
|
"file": "manofsteel-trailer4.mov",
|
|
|
|
|
"md5": "d97a8e575432dbcb81b7c3acb741f8a8",
|
|
|
|
|
"info_dict": {
|
|
|
|
|
"duration": 111,
|
|
|
|
|
"title": "Trailer 4",
|
|
|
|
|
"upload_date": "20130523",
|
|
|
|
|
"uploader_id": "wb",
|
2013-08-28 02:18:44 +02:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
2014-01-07 10:04:48 +01:00
|
|
|
"file": "manofsteel-trailer3.mov",
|
|
|
|
|
"md5": "b8017b7131b721fb4e8d6f49e1df908c",
|
|
|
|
|
"info_dict": {
|
|
|
|
|
"duration": 182,
|
|
|
|
|
"title": "Trailer 3",
|
|
|
|
|
"upload_date": "20130417",
|
|
|
|
|
"uploader_id": "wb",
|
2013-08-28 02:18:44 +02:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
2014-01-07 10:04:48 +01:00
|
|
|
"file": "manofsteel-trailer.mov",
|
|
|
|
|
"md5": "d0f1e1150989b9924679b441f3404d48",
|
|
|
|
|
"info_dict": {
|
|
|
|
|
"duration": 148,
|
|
|
|
|
"title": "Trailer",
|
|
|
|
|
"upload_date": "20121212",
|
|
|
|
|
"uploader_id": "wb",
|
2013-08-28 02:18:44 +02:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
2014-01-07 10:04:48 +01:00
|
|
|
"file": "manofsteel-teaser.mov",
|
|
|
|
|
"md5": "5fe08795b943eb2e757fa95cb6def1cb",
|
|
|
|
|
"info_dict": {
|
|
|
|
|
"duration": 93,
|
|
|
|
|
"title": "Teaser",
|
|
|
|
|
"upload_date": "20120721",
|
|
|
|
|
"uploader_id": "wb",
|
2013-08-28 02:18:44 +02:00
|
|
|
},
|
|
|
|
|
}
|
|
|
|
|
]
|
|
|
|
|
}
|
|
|
|
|
|
2013-09-29 20:49:58 +02:00
|
|
|
_JSON_RE = r'iTunes.playURL\((.*?)\);'
|
|
|
|
|
|
2013-08-28 02:18:44 +02:00
|
|
|
def _real_extract(self, url):
|
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
|
movie = mobj.group('movie')
|
|
|
|
|
uploader_id = mobj.group('company')
|
|
|
|
|
|
2013-09-29 20:49:58 +02:00
|
|
|
playlist_url = compat_urlparse.urljoin(url, u'includes/playlists/itunes.inc')
|
2013-12-10 21:03:53 +01:00
|
|
|
def fix_html(s):
|
|
|
|
|
s = re.sub(r'(?s)<script[^<]*?>.*?</script>', u'', s)
|
|
|
|
|
s = re.sub(r'<img ([^<]*?)>', r'<img \1/>', s)
|
|
|
|
|
# The ' in the onClick attributes are not escaped, it couldn't be parsed
|
|
|
|
|
# like: http://trailers.apple.com/trailers/wb/gravity/
|
|
|
|
|
def _clean_json(m):
|
|
|
|
|
return u'iTunes.playURL(%s);' % m.group(1).replace('\'', ''')
|
|
|
|
|
s = re.sub(self._JSON_RE, _clean_json, s)
|
|
|
|
|
s = u'<html>' + s + u'</html>'
|
|
|
|
|
return s
|
|
|
|
|
doc = self._download_xml(playlist_url, movie, transform_source=fix_html)
|
2013-08-28 02:18:44 +02:00
|
|
|
|
|
|
|
|
playlist = []
|
|
|
|
|
for li in doc.findall('./div/ul/li'):
|
2013-09-29 20:49:58 +02:00
|
|
|
on_click = li.find('.//a').attrib['onClick']
|
|
|
|
|
trailer_info_json = self._search_regex(self._JSON_RE,
|
|
|
|
|
on_click, u'trailer info')
|
|
|
|
|
trailer_info = json.loads(trailer_info_json)
|
|
|
|
|
title = trailer_info['title']
|
2013-08-28 02:18:44 +02:00
|
|
|
video_id = movie + '-' + re.sub(r'[^a-zA-Z0-9]', '', title).lower()
|
|
|
|
|
thumbnail = li.find('.//img').attrib['src']
|
2013-09-29 20:49:58 +02:00
|
|
|
upload_date = trailer_info['posted'].replace('-', '')
|
2013-08-28 02:18:44 +02:00
|
|
|
|
2013-09-29 20:49:58 +02:00
|
|
|
runtime = trailer_info['runtime']
|
|
|
|
|
m = re.search(r'(?P<minutes>[0-9]+):(?P<seconds>[0-9]{1,2})', runtime)
|
2013-08-28 02:18:44 +02:00
|
|
|
duration = None
|
|
|
|
|
if m:
|
|
|
|
|
duration = 60 * int(m.group('minutes')) + int(m.group('seconds'))
|
|
|
|
|
|
2013-09-29 20:49:58 +02:00
|
|
|
first_url = trailer_info['url']
|
2013-09-29 20:59:19 +02:00
|
|
|
trailer_id = first_url.split('/')[-1].rpartition('_')[0].lower()
|
2013-09-29 20:49:58 +02:00
|
|
|
settings_json_url = compat_urlparse.urljoin(url, 'includes/settings/%s.json' % trailer_id)
|
|
|
|
|
settings_json = self._download_webpage(settings_json_url, trailer_id, u'Downloading settings json')
|
|
|
|
|
settings = json.loads(settings_json)
|
2013-08-28 02:18:44 +02:00
|
|
|
|
2013-09-29 20:49:58 +02:00
|
|
|
formats = []
|
|
|
|
|
for format in settings['metadata']['sizes']:
|
|
|
|
|
# The src is a file pointing to the real video file
|
|
|
|
|
format_url = re.sub(r'_(\d*p.mov)', r'_h\1', format['src'])
|
|
|
|
|
formats.append({
|
|
|
|
|
'url': format_url,
|
|
|
|
|
'ext': determine_ext(format_url),
|
|
|
|
|
'format': format['type'],
|
|
|
|
|
'width': format['width'],
|
|
|
|
|
'height': int(format['height']),
|
|
|
|
|
})
|
2013-12-25 15:24:41 +01:00
|
|
|
|
|
|
|
|
self._sort_formats(formats)
|
2013-08-28 02:18:44 +02:00
|
|
|
|
2013-12-03 14:21:06 +01:00
|
|
|
playlist.append({
|
2013-08-28 02:18:44 +02:00
|
|
|
'_type': 'video',
|
|
|
|
|
'id': video_id,
|
|
|
|
|
'title': title,
|
|
|
|
|
'formats': formats,
|
|
|
|
|
'title': title,
|
|
|
|
|
'duration': duration,
|
|
|
|
|
'thumbnail': thumbnail,
|
|
|
|
|
'upload_date': upload_date,
|
|
|
|
|
'uploader_id': uploader_id,
|
|
|
|
|
'user_agent': 'QuickTime compatible (youtube-dl)',
|
2013-12-03 14:21:06 +01:00
|
|
|
})
|
2013-08-28 02:18:44 +02:00
|
|
|
|
|
|
|
|
return {
|
|
|
|
|
'_type': 'playlist',
|
|
|
|
|
'id': movie,
|
|
|
|
|
'entries': playlist,
|
|
|
|
|
}
|
