]>
Commit | Line | Data |
---|---|---|
c1195541 PH |
1 | from __future__ import unicode_literals |
2 | ||
bf64ff72 | 3 | import re |
c45aa560 | 4 | import json |
bf64ff72 YK |
5 | |
6 | from .common import InfoExtractor | |
1cc79574 | 7 | from ..compat import ( |
1f80e360 | 8 | compat_urllib_parse_unquote, |
c45aa560 | 9 | compat_urlparse, |
1cc79574 PH |
10 | ) |
11 | from ..utils import ( | |
c45aa560 | 12 | unescapeHTML, |
bf64ff72 YK |
13 | ) |
14 | ||
c45aa560 | 15 | |
bf64ff72 | 16 | class GameSpotIE(InfoExtractor): |
eeb23eb7 | 17 | _VALID_URL = r'http://(?:www\.)?gamespot\.com/.*-(?P<id>\d+)/?' |
34fe5a94 | 18 | _TESTS = [{ |
a2d5a4ee S |
19 | 'url': 'http://www.gamespot.com/videos/arma-3-community-guide-sitrep-i/2300-6410818/', |
20 | 'md5': 'b2a30deaa8654fcccd43713a6b6a4825', | |
21 | 'info_dict': { | |
22 | 'id': 'gs-2300-6410818', | |
23 | 'ext': 'mp4', | |
24 | 'title': 'Arma 3 - Community Guide: SITREP I', | |
c1195541 | 25 | 'description': 'Check out this video where some of the basics of Arma 3 is explained.', |
34fe5a94 JMF |
26 | }, |
27 | }, { | |
28 | 'url': 'http://www.gamespot.com/videos/the-witcher-3-wild-hunt-xbox-one-now-playing/2300-6424837/', | |
29 | 'info_dict': { | |
30 | 'id': 'gs-2300-6424837', | |
31 | 'ext': 'flv', | |
32 | 'title': 'The Witcher 3: Wild Hunt [Xbox ONE] - Now Playing', | |
33 | 'description': 'Join us as we take a look at the early hours of The Witcher 3: Wild Hunt and more.', | |
34 | }, | |
35 | }] | |
bf64ff72 | 36 | |
bf64ff72 | 37 | def _real_extract(self, url): |
a32f2531 | 38 | page_id = self._match_id(url) |
ebdf2af7 | 39 | webpage = self._download_webpage(url, page_id) |
a32f2531 PH |
40 | data_video_json = self._search_regex( |
41 | r'data-video=["\'](.*?)["\']', webpage, 'data video') | |
c45aa560 | 42 | data_video = json.loads(unescapeHTML(data_video_json)) |
34fe5a94 | 43 | streams = data_video['videoStreams'] |
bf64ff72 | 44 | |
c45aa560 | 45 | formats = [] |
34fe5a94 JMF |
46 | f4m_url = streams.get('f4m_stream') |
47 | if f4m_url is not None: | |
48 | # Transform the manifest url to a link to the mp4 files | |
49 | # they are used in mobile devices. | |
50 | f4m_path = compat_urlparse.urlparse(f4m_url).path | |
51 | QUALITIES_RE = r'((,\d+)+,?)' | |
52 | qualities = self._search_regex(QUALITIES_RE, f4m_path, 'qualities').strip(',').split(',') | |
53 | http_path = f4m_path[1:].split('/', 1)[1] | |
54 | http_template = re.sub(QUALITIES_RE, r'%s', http_path) | |
55 | http_template = http_template.replace('.csmil/manifest.f4m', '') | |
56 | http_template = compat_urlparse.urljoin( | |
57 | 'http://video.gamespotcdn.com/', http_template) | |
58 | for q in qualities: | |
59 | formats.append({ | |
60 | 'url': http_template % q, | |
61 | 'ext': 'mp4', | |
62 | 'format_id': q, | |
63 | }) | |
64 | else: | |
65 | for quality in ['sd', 'hd']: | |
66 | # It's actually a link to a flv file | |
67 | flv_url = streams.get('f4m_{0}'.format(quality)) | |
68 | if flv_url is not None: | |
69 | formats.append({ | |
70 | 'url': flv_url, | |
71 | 'ext': 'flv', | |
72 | 'format_id': quality, | |
73 | }) | |
bf64ff72 | 74 | |
fb7abb31 | 75 | return { |
c45aa560 | 76 | 'id': data_video['guid'], |
a32f2531 | 77 | 'display_id': page_id, |
1f80e360 | 78 | 'title': compat_urllib_parse_unquote(data_video['title']), |
c45aa560 | 79 | 'formats': formats, |
a32f2531 | 80 | 'description': self._html_search_meta('description', webpage), |
c45aa560 JMF |
81 | 'thumbnail': self._og_search_thumbnail(webpage), |
82 | } |