mirrors/youtube-dl
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[kontrtube] Extract display_id

Browse Source 
Trailing slash in URL is mandatory now
This commit is contained in:
Sergey M․ 2015-01-02 22:28:48 +06:00
parent 46d32a12c9
commit 2522a0b7da
1 changed files with 16 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
youtube_dl/extractor/kontrtube.py
View File

@ -10,13 +10,14 @@ from ..utils import int_or_none
class KontrTubeIE(InfoExtractor): class KontrTubeIE(InfoExtractor):
IE_NAME = 'kontrtube' IE_NAME = 'kontrtube'
IE_DESC = 'KontrTube.ru - Труба зовёт' IE_DESC = 'KontrTube.ru - Труба зовёт'
_VALID_URL = r'http://(?:www\.)?kontrtube\.ru/videos/(?P<id>\d+)/.+' _VALID_URL = r'http://(?:www\.)?kontrtube\.ru/videos/(?P<id>\d+)/(?P<display_id>[^/]+)/'
_TEST = { _TEST = {
'url': 'http://www.kontrtube.ru/videos/2678/nad-olimpiyskoy-derevney-v-sochi-podnyat-rossiyskiy-flag/', 'url': 'http://www.kontrtube.ru/videos/2678/nad-olimpiyskoy-derevney-v-sochi-podnyat-rossiyskiy-flag/',
'md5': '975a991a4926c9a85f383a736a2e6b80', 'md5': '975a991a4926c9a85f383a736a2e6b80',
'info_dict': { 'info_dict': {
'id': '2678', 'id': '2678',
'display_id': 'nad-olimpiyskoy-derevney-v-sochi-podnyat-rossiyskiy-flag',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Над олимпийской деревней в Сочи поднят российский флаг', 'title': 'Над олимпийской деревней в Сочи поднят российский флаг',
'description': 'md5:80edc4c613d5887ae8ccf1d59432be41', 'description': 'md5:80edc4c613d5887ae8ccf1d59432be41',
@ -28,21 +29,28 @@ class KontrTubeIE(InfoExtractor):
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id') video_id = mobj.group('id')
display_id = mobj.group('display_id')
webpage = self._download_webpage(url, video_id, 'Downloading page') webpage = self._download_webpage(
url, display_id, 'Downloading page')
video_url = self._html_search_regex(r"video_url: '(.+?)/?',", webpage, 'video URL') video_url = self._html_search_regex(
thumbnail = self._html_search_regex(r"preview_url: '(.+?)/?',", webpage, 'video thumbnail', fatal=False) r"video_url\s*:\s*'(.+?)/?',", webpage, 'video URL')
thumbnail = self._html_search_regex(
r"preview_url\s*:\s*'(.+?)/?',", webpage, 'video thumbnail', fatal=False)
title = self._html_search_regex( title = self._html_search_regex(
r'<title>(.+?)</title>', webpage, 'video title') r'<title>(.+?)</title>', webpage, 'video title')
description = self._html_search_meta('description', webpage, 'video description') description = self._html_search_meta(
'description', webpage, 'video description')
mobj = re.search( mobj = re.search(
r'<div class="col_2">Длительность: <span>(?P<minutes>\d+)м:(?P<seconds>\d+)с</span></div>', webpage) r'<div class="col_2">Длительность: <span>(?P<minutes>\d+)м:(?P<seconds>\d+)с</span></div>',
webpage)
duration = int(mobj.group('minutes')) * 60 + int(mobj.group('seconds')) if mobj else None duration = int(mobj.group('minutes')) * 60 + int(mobj.group('seconds')) if mobj else None
view_count = self._html_search_regex( view_count = self._html_search_regex(
r'<div class="col_2">Просмотров: <span>(\d+)</span></div>', webpage, 'view count', fatal=False) r'<div class="col_2">Просмотров: <span>(\d+)</span></div>',
webpage, 'view count', fatal=False)
comment_count = None comment_count = None
comment_str = self._html_search_regex( comment_str = self._html_search_regex(
@ -56,6 +64,7 @@ class KontrTubeIE(InfoExtractor):
return { return {
'id': video_id, 'id': video_id,
'display_id': display_id,
'url': video_url, 'url': video_url,
'thumbnail': thumbnail, 'thumbnail': thumbnail,
'title': title, 'title': title,