]>
Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/appletrailers.py
1 from __future__
import unicode_literals
6 from .common
import InfoExtractor
7 from ..compat
import compat_urlparse
15 class AppleTrailersIE(InfoExtractor
):
16 IE_NAME
= 'appletrailers'
17 _VALID_URL
= r
'https?://(?:www\.|movie)?trailers\.apple\.com/(?:trailers|ca)/(?P<company>[^/]+)/(?P<movie>[^/]+)'
19 'url': 'http://trailers.apple.com/trailers/wb/manofsteel/',
22 'title': 'Man of Steel',
26 'md5': 'd97a8e575432dbcb81b7c3acb741f8a8',
28 'id': 'manofsteel-trailer4',
32 'upload_date': '20130523',
37 'md5': 'b8017b7131b721fb4e8d6f49e1df908c',
39 'id': 'manofsteel-trailer3',
43 'upload_date': '20130417',
48 'md5': 'd0f1e1150989b9924679b441f3404d48',
50 'id': 'manofsteel-trailer',
54 'upload_date': '20121212',
59 'md5': '5fe08795b943eb2e757fa95cb6def1cb',
61 'id': 'manofsteel-teaser',
65 'upload_date': '20120721',
71 'url': 'http://trailers.apple.com/trailers/magnolia/blackthorn/',
74 'title': 'Blackthorn',
76 'playlist_mincount': 2,
77 'expected_warnings': ['Unable to download JSON metadata'],
79 # json data only available from http://trailers.apple.com/trailers/feeds/data/15881.json
80 'url': 'http://trailers.apple.com/trailers/fox/kungfupanda3/',
83 'title': 'Kung Fu Panda 3',
85 'playlist_mincount': 4,
87 'url': 'http://trailers.apple.com/ca/metropole/autrui/',
88 'only_matching': True,
90 'url': 'http://movietrailers.apple.com/trailers/focus_features/kuboandthetwostrings/',
91 'only_matching': True,
94 _JSON_RE
= r
'iTunes.playURL\((.*?)\);'
96 def _real_extract(self
, url
):
97 mobj
= re
.match(self
._VALID
_URL
, url
)
98 movie
= mobj
.group('movie')
99 uploader_id
= mobj
.group('company')
101 webpage
= self
._download
_webpage
(url
, movie
)
102 film_id
= self
._search
_regex
(r
"FilmId\s*=\s*'(\d+)'", webpage
, 'film id')
103 film_data
= self
._download
_json
(
104 'http://trailers.apple.com/trailers/feeds/data/%s.json' % film_id
,
105 film_id
, fatal
=False)
109 for clip
in film_data
.get('clips', []):
110 clip_title
= clip
['title']
113 for version
, version_data
in clip
.get('versions', {}).items():
114 for size
, size_data
in version_data
.get('sizes', {}).items():
115 src
= size_data
.get('src')
119 'format_id': '%s-%s' % (version
, size
),
120 'url': re
.sub(r
'_(\d+p\.mov)', r
'_h\1', src
),
121 'width': int_or_none(size_data
.get('width')),
122 'height': int_or_none(size_data
.get('height')),
123 'language': version
[:2],
125 self
._sort
_formats
(formats
)
128 'id': movie
+ '-' + re
.sub(r
'[^a-zA-Z0-9]', '', clip_title
).lower(),
131 'thumbnail': clip
.get('screen') or clip
.get('thumb'),
132 'duration': parse_duration(clip
.get('runtime') or clip
.get('faded')),
133 'upload_date': unified_strdate(clip
.get('posted')),
134 'uploader_id': uploader_id
,
137 page_data
= film_data
.get('page', {})
138 return self
.playlist_result(entries
, film_id
, page_data
.get('movie_title'))
140 playlist_url
= compat_urlparse
.urljoin(url
, 'includes/playlists/itunes.inc')
143 s
= re
.sub(r
'(?s)<script[^<]*?>.*?</script>', '', s
)
144 s
= re
.sub(r
'<img ([^<]*?)/?>', r
'<img \1/>', s
)
145 # The ' in the onClick attributes are not escaped, it couldn't be parsed
146 # like: http://trailers.apple.com/trailers/wb/gravity/
149 return 'iTunes.playURL(%s);' % m
.group(1).replace('\'', ''')
150 s
= re
.sub(self
._JSON
_RE
, _clean_json
, s
)
151 s
= '<html>%s</html>' % s
153 doc
= self
._download
_xml
(playlist_url
, movie
, transform_source
=fix_html
)
156 for li
in doc
.findall('./div/ul/li'):
157 on_click
= li
.find('.//a').attrib
['onClick']
158 trailer_info_json
= self
._search
_regex
(self
._JSON
_RE
,
159 on_click
, 'trailer info')
160 trailer_info
= json
.loads(trailer_info_json
)
161 first_url
= trailer_info
.get('url')
164 title
= trailer_info
['title']
165 video_id
= movie
+ '-' + re
.sub(r
'[^a-zA-Z0-9]', '', title
).lower()
166 thumbnail
= li
.find('.//img').attrib
['src']
167 upload_date
= trailer_info
['posted'].replace('-', '')
169 runtime
= trailer_info
['runtime']
170 m
= re
.search(r
'(?P<minutes>[0-9]+):(?P<seconds>[0-9]{1,2})', runtime
)
173 duration
= 60 * int(m
.group('minutes')) + int(m
.group('seconds'))
175 trailer_id
= first_url
.split('/')[-1].rpartition('_')[0].lower()
176 settings_json_url
= compat_urlparse
.urljoin(url
, 'includes/settings/%s.json' % trailer_id
)
177 settings
= self
._download
_json
(settings_json_url
, trailer_id
, 'Downloading settings json')
180 for format
in settings
['metadata']['sizes']:
181 # The src is a file pointing to the real video file
182 format_url
= re
.sub(r
'_(\d*p\.mov)', r
'_h\1', format
['src'])
185 'format': format
['type'],
186 'width': int_or_none(format
['width']),
187 'height': int_or_none(format
['height']),
190 self
._sort
_formats
(formats
)
197 'duration': duration
,
198 'thumbnail': thumbnail
,
199 'upload_date': upload_date
,
200 'uploader_id': uploader_id
,
202 'User-Agent': 'QuickTime compatible (youtube-dl)',
213 class AppleTrailersSectionIE(InfoExtractor
):
214 IE_NAME
= 'appletrailers:section'
217 'feed_path': 'just_added',
218 'title': 'Just Added',
221 'feed_path': 'exclusive',
222 'title': 'Exclusive',
225 'feed_path': 'just_hd',
229 'feed_path': 'most_pop',
230 'title': 'Most Popular',
233 'feed_path': 'studios',
234 'title': 'Movie Studios',
237 _VALID_URL
= r
'https?://(?:www\.)?trailers\.apple\.com/#section=(?P<id>%s)' % '|'.join(_SECTIONS
)
239 'url': 'http://trailers.apple.com/#section=justadded',
241 'title': 'Just Added',
244 'playlist_mincount': 80,
246 'url': 'http://trailers.apple.com/#section=exclusive',
248 'title': 'Exclusive',
251 'playlist_mincount': 80,
253 'url': 'http://trailers.apple.com/#section=justhd',
258 'playlist_mincount': 80,
260 'url': 'http://trailers.apple.com/#section=mostpopular',
262 'title': 'Most Popular',
265 'playlist_mincount': 30,
267 'url': 'http://trailers.apple.com/#section=moviestudios',
269 'title': 'Movie Studios',
270 'id': 'moviestudios',
272 'playlist_mincount': 80,
275 def _real_extract(self
, url
):
276 section
= self
._match
_id
(url
)
277 section_data
= self
._download
_json
(
278 'http://trailers.apple.com/trailers/home/feeds/%s.json' % self
._SECTIONS
[section
]['feed_path'],
281 self
.url_result('http://trailers.apple.com' + e
['location'])
282 for e
in section_data
]
283 return self
.playlist_result(entries
, section
, self
._SECTIONS
[section
]['title'])