]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/crackle.py
13f425b2bf1672c66a723f22a0e6e3d73c0456c3
[youtubedl] / youtube_dl / extractor / crackle.py
1 # coding: utf-8
2 from __future__ import unicode_literals, division
3
4 from .common import InfoExtractor
5 from ..utils import int_or_none
6
7
8 class CrackleIE(InfoExtractor):
9 _GEO_COUNTRIES = ['US']
10 _VALID_URL = r'(?:crackle:|https?://(?:(?:www|m)\.)?crackle\.com/(?:playlist/\d+/|(?:[^/]+/)+))(?P<id>\d+)'
11 _TEST = {
12 'url': 'http://www.crackle.com/comedians-in-cars-getting-coffee/2498934',
13 'info_dict': {
14 'id': '2498934',
15 'ext': 'mp4',
16 'title': 'Everybody Respects A Bloody Nose',
17 'description': 'Jerry is kaffeeklatsching in L.A. with funnyman J.B. Smoove (Saturday Night Live, Real Husbands of Hollywood). They’re headed for brew at 10 Speed Coffee in a 1964 Studebaker Avanti.',
18 'thumbnail': r're:^https?://.*\.jpg',
19 'duration': 906,
20 'series': 'Comedians In Cars Getting Coffee',
21 'season_number': 8,
22 'episode_number': 4,
23 'subtitles': {
24 'en-US': [
25 {'ext': 'vtt'},
26 {'ext': 'tt'},
27 ]
28 },
29 },
30 'params': {
31 # m3u8 download
32 'skip_download': True,
33 }
34 }
35
36 _THUMBNAIL_RES = [
37 (120, 90),
38 (208, 156),
39 (220, 124),
40 (220, 220),
41 (240, 180),
42 (250, 141),
43 (315, 236),
44 (320, 180),
45 (360, 203),
46 (400, 300),
47 (421, 316),
48 (460, 330),
49 (460, 460),
50 (462, 260),
51 (480, 270),
52 (587, 330),
53 (640, 480),
54 (700, 330),
55 (700, 394),
56 (854, 480),
57 (1024, 1024),
58 (1920, 1080),
59 ]
60
61 # extracted from http://legacyweb-us.crackle.com/flash/ReferrerRedirect.ashx
62 _MEDIA_FILE_SLOTS = {
63 'c544.flv': {
64 'width': 544,
65 'height': 306,
66 },
67 '360p.mp4': {
68 'width': 640,
69 'height': 360,
70 },
71 '480p.mp4': {
72 'width': 852,
73 'height': 478,
74 },
75 '480p_1mbps.mp4': {
76 'width': 852,
77 'height': 478,
78 },
79 }
80
81 def _real_extract(self, url):
82 video_id = self._match_id(url)
83
84 config_doc = self._download_xml(
85 'http://legacyweb-us.crackle.com/flash/QueryReferrer.ashx?site=16',
86 video_id, 'Downloading config')
87
88 item = self._download_xml(
89 'http://legacyweb-us.crackle.com/app/revamp/vidwallcache.aspx?flags=-1&fm=%s' % video_id,
90 video_id, headers=self.geo_verification_headers()).find('i')
91 title = item.attrib['t']
92
93 subtitles = {}
94 formats = self._extract_m3u8_formats(
95 'http://content.uplynk.com/ext/%s/%s.m3u8' % (config_doc.attrib['strUplynkOwnerId'], video_id),
96 video_id, 'mp4', m3u8_id='hls', fatal=None)
97 thumbnails = []
98 path = item.attrib.get('p')
99 if path:
100 for width, height in self._THUMBNAIL_RES:
101 res = '%dx%d' % (width, height)
102 thumbnails.append({
103 'id': res,
104 'url': 'http://images-us-am.crackle.com/%stnl_%s.jpg' % (path, res),
105 'width': width,
106 'height': height,
107 'resolution': res,
108 })
109 http_base_url = 'http://ahttp.crackle.com/' + path
110 for mfs_path, mfs_info in self._MEDIA_FILE_SLOTS.items():
111 formats.append({
112 'url': http_base_url + mfs_path,
113 'format_id': 'http-' + mfs_path.split('.')[0],
114 'width': mfs_info['width'],
115 'height': mfs_info['height'],
116 })
117 for cc in item.findall('cc'):
118 locale = cc.attrib.get('l')
119 v = cc.attrib.get('v')
120 if locale and v:
121 if locale not in subtitles:
122 subtitles[locale] = []
123 for url_ext, ext in (('vtt', 'vtt'), ('xml', 'tt')):
124 subtitles.setdefault(locale, []).append({
125 'url': '%s/%s%s_%s.%s' % (config_doc.attrib['strSubtitleServer'], path, locale, v, url_ext),
126 'ext': ext,
127 })
128 self._sort_formats(formats, ('width', 'height', 'tbr', 'format_id'))
129
130 return {
131 'id': video_id,
132 'title': title,
133 'description': item.attrib.get('d'),
134 'duration': int(item.attrib.get('r'), 16) / 1000 if item.attrib.get('r') else None,
135 'series': item.attrib.get('sn'),
136 'season_number': int_or_none(item.attrib.get('se')),
137 'episode_number': int_or_none(item.attrib.get('ep')),
138 'thumbnails': thumbnails,
139 'subtitles': subtitles,
140 'formats': formats,
141 }