mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-01-07 17:16:08 +00:00
Merge branch 'misterhat-pinkbike'
This commit is contained in:
commit
edcd2d665b
|
@ -401,6 +401,7 @@ from .pbs import PBSIE
|
|||
from .philharmoniedeparis import PhilharmonieDeParisIE
|
||||
from .phoenix import PhoenixIE
|
||||
from .photobucket import PhotobucketIE
|
||||
from .pinkbike import PinkbikeIE
|
||||
from .planetaplay import PlanetaPlayIE
|
||||
from .pladform import PladformIE
|
||||
from .played import PlayedIE
|
||||
|
|
96
youtube_dl/extractor/pinkbike.py
Normal file
96
youtube_dl/extractor/pinkbike.py
Normal file
|
@ -0,0 +1,96 @@
|
|||
# coding: utf-8
|
||||
from __future__ import unicode_literals
|
||||
|
||||
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': 're:^https?://.*\.jpg$',
|
||||
'duration': 100,
|
||||
'upload_date': '20150406',
|
||||
'uploader': 'revelco',
|
||||
'location': 'Victoria, British Columbia, Canada',
|
||||
'view_count': int,
|
||||
'comment_count': int,
|
||||
}
|
||||
}, {
|
||||
'url': 'http://es.pinkbike.org/i/kvid/kvid-y5.swf?id=406629',
|
||||
'only_matching': True,
|
||||
}]
|
||||
|
||||
def _real_extract(self, url):
|
||||
video_id = self._match_id(url)
|
||||
|
||||
webpage = self._download_webpage(
|
||||
'http://www.pinkbike.com/video/%s' % video_id, video_id)
|
||||
|
||||
formats = []
|
||||
for _, format_id, src in re.findall(
|
||||
r'data-quality=((?:\\)?["\'])(.+?)\1[^>]+src=\1(.+?)\1', webpage):
|
||||
height = int_or_none(self._search_regex(
|
||||
r'^(\d+)[pP]$', format_id, 'height', default=None))
|
||||
formats.append({
|
||||
'url': src,
|
||||
'format_id': format_id,
|
||||
'height': height,
|
||||
})
|
||||
self._sort_formats(formats)
|
||||
|
||||
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'un:\s*"([^"]+)"', 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(
|
||||
r'<span[^>]+class="stat-num"[^>]*>([\d,.]+)</span>\s*<span[^>]+class="stat-label"[^>]*>%s' % 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
|
||||
}
|
Loading…
Reference in a new issue