123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293 |
- import re
- from .common import InfoExtractor
- from ..utils import (
- int_or_none,
- remove_end,
- remove_start,
- str_to_int,
- unified_strdate,
- )
- class PinkbikeIE(InfoExtractor):
- _VALID_URL = r'https?://(?:(?:www\.)?pinkbike\.com/video/|es\.pinkbike\.org/i/kvid/kvid-y5\.swf\?id=)(?P<id>[0-9]+)'
- _TESTS = [{
- 'url': 'http://www.pinkbike.com/video/402811/',
- 'md5': '4814b8ca7651034cd87e3361d5c2155a',
- 'info_dict': {
- 'id': '402811',
- 'ext': 'mp4',
- 'title': 'Brandon Semenuk - RAW 100',
- 'description': 'Official release: www.redbull.ca/rupertwalker',
- 'thumbnail': r're:^https?://.*\.jpg$',
- 'duration': 100,
- 'upload_date': '20150406',
- 'uploader': 'revelco',
- 'location': 'Victoria, British Columbia, Canada',
- 'view_count': int,
- 'comment_count': int,
- },
- }, {
- 'url': 'http://es.pinkbike.org/i/kvid/kvid-y5.swf?id=406629',
- 'only_matching': True,
- }]
- def _real_extract(self, url):
- video_id = self._match_id(url)
- webpage = self._download_webpage(
- f'http://www.pinkbike.com/video/{video_id}', video_id)
- formats = []
- for _, format_id, src in re.findall(
- r'data-quality=((?:\\)?["\'])(.+?)\1[^>]+src=\1(.+?)\1', webpage):
- height = int_or_none(self._search_regex(
- r'^(\d+)[pP]$', format_id, 'height', default=None))
- formats.append({
- 'url': src,
- 'format_id': format_id,
- 'height': height,
- })
- title = remove_end(self._og_search_title(webpage), ' Video - Pinkbike')
- description = self._html_search_regex(
- r'(?s)id="media-description"[^>]*>(.+?)<',
- webpage, 'description', default=None) or remove_start(
- self._og_search_description(webpage), title + '. ')
- thumbnail = self._og_search_thumbnail(webpage)
- duration = int_or_none(self._html_search_meta(
- 'video:duration', webpage, 'duration'))
- uploader = self._search_regex(
- r'<a[^>]+\brel=["\']author[^>]+>([^<]+)', webpage,
- 'uploader', fatal=False)
- upload_date = unified_strdate(self._search_regex(
- r'class="fullTime"[^>]+title="([^"]+)"',
- webpage, 'upload date', fatal=False))
- location = self._html_search_regex(
- r'(?s)<dt>Location</dt>\s*<dd>(.+?)<',
- webpage, 'location', fatal=False)
- def extract_count(webpage, label):
- return str_to_int(self._search_regex(
- rf'<span[^>]+class="stat-num"[^>]*>([\d,.]+)</span>\s*<span[^>]+class="stat-label"[^>]*>{label}',
- webpage, label, fatal=False))
- view_count = extract_count(webpage, 'Views')
- comment_count = extract_count(webpage, 'Comments')
- return {
- 'id': video_id,
- 'title': title,
- 'description': description,
- 'thumbnail': thumbnail,
- 'duration': duration,
- 'upload_date': upload_date,
- 'uploader': uploader,
- 'location': location,
- 'view_count': view_count,
- 'comment_count': comment_count,
- 'formats': formats,
- }
|