]>
Commit | Line | Data |
---|---|---|
7198063d MH |
1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
680f9744 MH |
4 | from ..utils import ( |
5 | int_or_none, | |
6 | remove_end, | |
385c3e5e S |
7 | remove_start, |
8 | str_to_int, | |
9 | unified_strdate, | |
680f9744 | 10 | ) |
7198063d MH |
11 | |
12 | ||
13 | class PinkbikeIE(InfoExtractor): | |
385c3e5e | 14 | _VALID_URL = r'https?://(?:(?:www\.)?pinkbike\.com/video/|es\.pinkbike\.org/i/kvid/kvid-y5\.swf\?id=)(?P<id>[0-9]+)' |
7198063d MH |
15 | _TESTS = [{ |
16 | 'url': 'http://www.pinkbike.com/video/402811/', | |
17 | 'md5': '4814b8ca7651034cd87e3361d5c2155a', | |
18 | 'info_dict': { | |
19 | 'id': '402811', | |
20 | 'ext': 'mp4', | |
21 | 'title': 'Brandon Semenuk - RAW 100', | |
385c3e5e | 22 | 'description': 'Official release: www.redbull.ca/rupertwalker', |
ec85ded8 | 23 | 'thumbnail': r're:^https?://.*\.jpg$', |
385c3e5e | 24 | 'duration': 100, |
7198063d | 25 | 'upload_date': '20150406', |
385c3e5e S |
26 | 'uploader': 'revelco', |
27 | 'location': 'Victoria, British Columbia, Canada', | |
28 | 'view_count': int, | |
29 | 'comment_count': int, | |
7198063d MH |
30 | } |
31 | }, { | |
385c3e5e S |
32 | 'url': 'http://es.pinkbike.org/i/kvid/kvid-y5.swf?id=406629', |
33 | 'only_matching': True, | |
7198063d MH |
34 | }] |
35 | ||
36 | def _real_extract(self, url): | |
37 | video_id = self._match_id(url) | |
7198063d | 38 | |
385c3e5e S |
39 | webpage = self._download_webpage( |
40 | 'http://www.pinkbike.com/video/%s' % video_id, video_id) | |
7198063d | 41 | |
385c3e5e S |
42 | formats = [] |
43 | for _, format_id, src in re.findall( | |
6a745c2c | 44 | r'data-quality=((?:\\)?["\'])(.+?)\1[^>]+src=\1(.+?)\1', webpage): |
385c3e5e S |
45 | height = int_or_none(self._search_regex( |
46 | r'^(\d+)[pP]$', format_id, 'height', default=None)) | |
47 | formats.append({ | |
48 | 'url': src, | |
49 | 'format_id': format_id, | |
50 | 'height': height, | |
51 | }) | |
680f9744 | 52 | |
385c3e5e S |
53 | title = remove_end(self._og_search_title(webpage), ' Video - Pinkbike') |
54 | description = self._html_search_regex( | |
55 | r'(?s)id="media-description"[^>]*>(.+?)<', | |
56 | webpage, 'description', default=None) or remove_start( | |
57 | self._og_search_description(webpage), title + '. ') | |
58 | thumbnail = self._og_search_thumbnail(webpage) | |
680f9744 MH |
59 | duration = int_or_none(self._html_search_meta( |
60 | 'video:duration', webpage, 'duration')) | |
7198063d | 61 | |
385c3e5e | 62 | uploader = self._search_regex( |
3021cf83 S |
63 | r'<a[^>]+\brel=["\']author[^>]+>([^<]+)', webpage, |
64 | 'uploader', fatal=False) | |
385c3e5e S |
65 | upload_date = unified_strdate(self._search_regex( |
66 | r'class="fullTime"[^>]+title="([^"]+)"', | |
67 | webpage, 'upload date', fatal=False)) | |
7198063d MH |
68 | |
69 | location = self._html_search_regex( | |
385c3e5e S |
70 | r'(?s)<dt>Location</dt>\s*<dd>(.+?)<', |
71 | webpage, 'location', fatal=False) | |
7198063d | 72 | |
385c3e5e S |
73 | def extract_count(webpage, label): |
74 | return str_to_int(self._search_regex( | |
75 | r'<span[^>]+class="stat-num"[^>]*>([\d,.]+)</span>\s*<span[^>]+class="stat-label"[^>]*>%s' % label, | |
76 | webpage, label, fatal=False)) | |
7198063d | 77 | |
385c3e5e S |
78 | view_count = extract_count(webpage, 'Views') |
79 | comment_count = extract_count(webpage, 'Comments') | |
7198063d MH |
80 | |
81 | return { | |
82 | 'id': video_id, | |
83 | 'title': title, | |
84 | 'description': description, | |
385c3e5e | 85 | 'thumbnail': thumbnail, |
680f9744 | 86 | 'duration': duration, |
7198063d | 87 | 'upload_date': upload_date, |
385c3e5e | 88 | 'uploader': uploader, |
7198063d | 89 | 'location': location, |
385c3e5e S |
90 | 'view_count': view_count, |
91 | 'comment_count': comment_count, | |
7198063d MH |
92 | 'formats': formats |
93 | } |