]>
Commit | Line | Data |
---|---|---|
ecfef3e5 PH |
1 | from __future__ import unicode_literals |
2 | ||
5d9f6cbc S |
3 | import base64 |
4 | import json | |
d8d61486 | 5 | import re |
d8d61486 JMF |
6 | |
7 | from .common import InfoExtractor | |
dd67702a | 8 | from ..utils import ( |
7dd6ab4a | 9 | determine_ext, |
96c2e3e9 | 10 | mimetype2ext, |
0167f0db | 11 | parse_duration, |
dd67702a | 12 | qualities, |
5d9f6cbc | 13 | try_get, |
3052a30d | 14 | url_or_none, |
d8d61486 JMF |
15 | ) |
16 | ||
17 | ||
18 | class ImdbIE(InfoExtractor): | |
ecfef3e5 PH |
19 | IE_NAME = 'imdb' |
20 | IE_DESC = 'Internet Movie Database trailers' | |
5d9f6cbc | 21 | _VALID_URL = r'https?://(?:www|m)\.imdb\.com/(?:video|title|list).*?[/-]vi(?P<id>\d+)' |
d8d61486 | 22 | |
f196508f | 23 | _TESTS = [{ |
ecfef3e5 | 24 | 'url': 'http://www.imdb.com/video/imdb/vi2524815897', |
ecfef3e5 PH |
25 | 'info_dict': { |
26 | 'id': '2524815897', | |
27 | 'ext': 'mp4', | |
5d9f6cbc | 28 | 'title': 'No. 2', |
0167f0db | 29 | 'description': 'md5:87bd0bdc61e351f21f20d2d7441cb4e7', |
5d9f6cbc | 30 | 'duration': 152, |
d8d61486 | 31 | } |
f196508f S |
32 | }, { |
33 | 'url': 'http://www.imdb.com/video/_/vi2524815897', | |
34 | 'only_matching': True, | |
4c93ee8d S |
35 | }, { |
36 | 'url': 'http://www.imdb.com/title/tt1667889/?ref_=ext_shr_eml_vi#lb-vi2524815897', | |
37 | 'only_matching': True, | |
38 | }, { | |
39 | 'url': 'http://www.imdb.com/title/tt1667889/#lb-vi2524815897', | |
40 | 'only_matching': True, | |
13607896 S |
41 | }, { |
42 | 'url': 'http://www.imdb.com/videoplayer/vi1562949145', | |
43 | 'only_matching': True, | |
b8457665 S |
44 | }, { |
45 | 'url': 'http://www.imdb.com/title/tt4218696/videoplayer/vi2608641561', | |
46 | 'only_matching': True, | |
0167f0db RA |
47 | }, { |
48 | 'url': 'https://www.imdb.com/list/ls009921623/videoplayer/vi260482329', | |
49 | 'only_matching': True, | |
f196508f | 50 | }] |
d8d61486 JMF |
51 | |
52 | def _real_extract(self, url): | |
11fba175 | 53 | video_id = self._match_id(url) |
5d9f6cbc S |
54 | |
55 | data = self._download_json( | |
56 | 'https://www.imdb.com/ve/data/VIDEO_PLAYBACK_DATA', video_id, | |
57 | query={ | |
58 | 'key': base64.b64encode(json.dumps({ | |
59 | 'type': 'VIDEO_PLAYER', | |
60 | 'subType': 'FORCE_LEGACY', | |
61 | 'id': 'vi%s' % video_id, | |
62 | }).encode()).decode(), | |
63 | })[0] | |
dd67702a | 64 | |
f7f2e53a | 65 | quality = qualities(('SD', '480p', '720p', '1080p')) |
d8d61486 | 66 | formats = [] |
5d9f6cbc | 67 | for encoding in data['videoLegacyEncodings']: |
0167f0db | 68 | if not encoding or not isinstance(encoding, dict): |
96c2e3e9 | 69 | continue |
5d9f6cbc | 70 | video_url = url_or_none(encoding.get('url')) |
3052a30d | 71 | if not video_url: |
96c2e3e9 | 72 | continue |
a62460aa S |
73 | ext = mimetype2ext(encoding.get( |
74 | 'mimeType')) or determine_ext(video_url) | |
0167f0db RA |
75 | if ext == 'm3u8': |
76 | formats.extend(self._extract_m3u8_formats( | |
77 | video_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
5d9f6cbc | 78 | preference=1, m3u8_id='hls', fatal=False)) |
96c2e3e9 | 79 | continue |
0167f0db RA |
80 | format_id = encoding.get('definition') |
81 | formats.append({ | |
82 | 'format_id': format_id, | |
83 | 'url': video_url, | |
84 | 'ext': ext, | |
85 | 'quality': quality(format_id), | |
86 | }) | |
dd67702a | 87 | self._sort_formats(formats) |
d8d61486 | 88 | |
5d9f6cbc S |
89 | webpage = self._download_webpage( |
90 | 'https://www.imdb.com/video/vi' + video_id, video_id) | |
91 | video_metadata = self._parse_json(self._search_regex( | |
92 | r'args\.push\(\s*({.+?})\s*\)\s*;', webpage, | |
93 | 'video metadata'), video_id) | |
94 | ||
95 | video_info = video_metadata.get('VIDEO_INFO') | |
96 | if video_info and isinstance(video_info, dict): | |
97 | info = try_get( | |
98 | video_info, lambda x: x[list(video_info.keys())[0]][0], dict) | |
99 | else: | |
100 | info = {} | |
101 | ||
102 | title = self._html_search_meta( | |
103 | ['og:title', 'twitter:title'], webpage) or self._html_search_regex( | |
104 | r'<title>(.+?)</title>', webpage, 'title', | |
105 | default=None) or info['videoTitle'] | |
106 | ||
d8d61486 JMF |
107 | return { |
108 | 'id': video_id, | |
0167f0db | 109 | 'title': title, |
5d9f6cbc | 110 | 'alt_title': info.get('videoSubTitle'), |
d8d61486 | 111 | 'formats': formats, |
5d9f6cbc S |
112 | 'description': info.get('videoDescription'), |
113 | 'thumbnail': url_or_none(try_get( | |
114 | video_metadata, lambda x: x['videoSlate']['source'])), | |
115 | 'duration': parse_duration(info.get('videoRuntime')), | |
d8d61486 | 116 | } |
c645c765 | 117 | |
ecfef3e5 | 118 | |
c645c765 | 119 | class ImdbListIE(InfoExtractor): |
ecfef3e5 PH |
120 | IE_NAME = 'imdb:list' |
121 | IE_DESC = 'Internet Movie Database lists' | |
27694fe7 | 122 | _VALID_URL = r'https?://(?:www\.)?imdb\.com/list/ls(?P<id>\d{9})(?!/videoplayer/vi\d+)' |
22a6f150 | 123 | _TEST = { |
0167f0db | 124 | 'url': 'https://www.imdb.com/list/ls009921623/', |
22a6f150 | 125 | 'info_dict': { |
0167f0db RA |
126 | 'id': '009921623', |
127 | 'title': 'The Bourne Legacy', | |
128 | 'description': 'A list of trailers, clips, and more from The Bourne Legacy, starring Jeremy Renner and Rachel Weisz.', | |
22a6f150 | 129 | }, |
0167f0db | 130 | 'playlist_count': 8, |
22a6f150 | 131 | } |
5f6a1245 | 132 | |
c645c765 | 133 | def _real_extract(self, url): |
11fba175 | 134 | list_id = self._match_id(url) |
d7b51547 | 135 | webpage = self._download_webpage(url, list_id) |
d7b51547 PH |
136 | entries = [ |
137 | self.url_result('http://www.imdb.com' + m, 'Imdb') | |
0167f0db | 138 | for m in re.findall(r'href="(/list/ls%s/videoplayer/vi[^"]+)"' % list_id, webpage)] |
d7b51547 PH |
139 | |
140 | list_title = self._html_search_regex( | |
0167f0db RA |
141 | r'<h1[^>]+class="[^"]*header[^"]*"[^>]*>(.*?)</h1>', |
142 | webpage, 'list title') | |
143 | list_description = self._html_search_regex( | |
144 | r'<div[^>]+class="[^"]*list-description[^"]*"[^>]*><p>(.*?)</p>', | |
145 | webpage, 'list description') | |
d7b51547 | 146 | |
0167f0db | 147 | return self.playlist_result(entries, list_id, list_title, list_description) |