]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/svt.py
Update changelog.
[youtubedl] / youtube_dl / extractor / svt.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..utils import (
8 determine_ext,
9 )
10
11
12 class SVTBaseIE(InfoExtractor):
13 def _extract_video(self, url, video_id):
14 info = self._download_json(url, video_id)
15
16 title = info['context']['title']
17 thumbnail = info['context'].get('thumbnailImage')
18
19 video_info = info['video']
20 formats = []
21 for vr in video_info['videoReferences']:
22 vurl = vr['url']
23 ext = determine_ext(vurl)
24 if ext == 'm3u8':
25 formats.extend(self._extract_m3u8_formats(
26 vurl, video_id,
27 ext='mp4', entry_protocol='m3u8_native',
28 m3u8_id=vr.get('playerType')))
29 elif ext == 'f4m':
30 formats.extend(self._extract_f4m_formats(
31 vurl + '?hdcore=3.3.0', video_id,
32 f4m_id=vr.get('playerType')))
33 else:
34 formats.append({
35 'format_id': vr.get('playerType'),
36 'url': vurl,
37 })
38 self._sort_formats(formats)
39
40 duration = video_info.get('materialLength')
41 age_limit = 18 if video_info.get('inappropriateForChildren') else 0
42
43 return {
44 'id': video_id,
45 'title': title,
46 'formats': formats,
47 'thumbnail': thumbnail,
48 'duration': duration,
49 'age_limit': age_limit,
50 }
51
52
53 class SVTIE(SVTBaseIE):
54 _VALID_URL = r'https?://(?:www\.)?svt\.se/wd\?(?:.*?&)?widgetId=(?P<widget_id>\d+)&.*?\barticleId=(?P<id>\d+)'
55 _TEST = {
56 'url': 'http://www.svt.se/wd?widgetId=23991&sectionId=541&articleId=2900353&type=embed&contextSectionId=123&autostart=false',
57 'md5': '9648197555fc1b49e3dc22db4af51d46',
58 'info_dict': {
59 'id': '2900353',
60 'ext': 'flv',
61 'title': 'Här trycker Jagr till Giroux (under SVT-intervjun)',
62 'duration': 27,
63 'age_limit': 0,
64 },
65 }
66
67 @staticmethod
68 def _extract_url(webpage):
69 mobj = re.search(
70 r'(?:<iframe src|href)="(?P<url>%s[^"]*)"' % SVTIE._VALID_URL, webpage)
71 if mobj:
72 return mobj.group('url')
73
74 def _real_extract(self, url):
75 mobj = re.match(self._VALID_URL, url)
76 widget_id = mobj.group('widget_id')
77 article_id = mobj.group('id')
78 return self._extract_video(
79 'http://www.svt.se/wd?widgetId=%s&articleId=%s&format=json&type=embed&output=json' % (widget_id, article_id),
80 article_id)
81
82
83 class SVTPlayIE(SVTBaseIE):
84 IE_DESC = 'SVT Play and Öppet arkiv'
85 _VALID_URL = r'https?://(?:www\.)?(?P<host>svtplay|oppetarkiv)\.se/video/(?P<id>[0-9]+)'
86 _TESTS = [{
87 'url': 'http://www.svtplay.se/video/2609989/sm-veckan/sm-veckan-rally-final-sasong-1-sm-veckan-rally-final',
88 'md5': 'ade3def0643fa1c40587a422f98edfd9',
89 'info_dict': {
90 'id': '2609989',
91 'ext': 'flv',
92 'title': 'SM veckan vinter, Örebro - Rally, final',
93 'duration': 4500,
94 'thumbnail': 're:^https?://.*[\.-]jpg$',
95 'age_limit': 0,
96 },
97 }, {
98 'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318',
99 'md5': 'c3101a17ce9634f4c1f9800f0746c187',
100 'info_dict': {
101 'id': '1058509',
102 'ext': 'flv',
103 'title': 'Farlig kryssning',
104 'duration': 2566,
105 'thumbnail': 're:^https?://.*[\.-]jpg$',
106 'age_limit': 0,
107 },
108 'skip': 'Only works from Sweden',
109 }]
110
111 def _real_extract(self, url):
112 mobj = re.match(self._VALID_URL, url)
113 video_id = mobj.group('id')
114 host = mobj.group('host')
115 return self._extract_video(
116 'http://www.%s.se/video/%s?output=json' % (host, video_id),
117 video_id)