Akkariin/youtube-dl
1
0
Fork 0
mirror of https://github.com/l1ving/youtube-dl synced 2024-12-22 20:37:58 +08:00
Code Issues Releases Wiki Activity

[ruutu] Fix extraction (Closes #6759)

Browse Source
This commit is contained in:
Sergey M․ 2015-09-05 00:01:40 +06:00
parent 5d2354f177
commit 22889ab175
1 changed files with 10 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
youtube_dl/extractor/ruutu.py
View File

@ -6,19 +6,19 @@ from ..compat import compat_urllib_parse_urlparse
from ..utils import ( from ..utils import (
determine_ext, determine_ext,
int_or_none, int_or_none,
xpath_attr,
xpath_text, xpath_text,
) )
class RuutuIE(InfoExtractor): class RuutuIE(InfoExtractor):
_VALID_URL = r'http://(?:www\.)?ruutu\.fi/ohjelmat/(?:[^/?#]+/)*(?P<id>[^/?#]+)' _VALID_URL = r'https?://(?:www\.)?ruutu\.fi/video/(?P<id>\d+)'
_TESTS = [ _TESTS = [
{ {
'url': 'http://www.ruutu.fi/ohjelmat/oletko-aina-halunnut-tietaa-mita-tapahtuu-vain-hetki-ennen-lahetysta-nyt-se-selvisi', 'url': 'http://www.ruutu.fi/video/2058907',
'md5': 'ab2093f39be1ca8581963451b3c0234f', 'md5': 'ab2093f39be1ca8581963451b3c0234f',
'info_dict': { 'info_dict': {
'id': '2058907', 'id': '2058907',
'display_id': 'oletko-aina-halunnut-tietaa-mita-tapahtuu-vain-hetki-ennen-lahetysta-nyt-se-selvisi',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Oletko aina halunnut tietää mitä tapahtuu vain hetki ennen lähetystä? - Nyt se selvisi!', 'title': 'Oletko aina halunnut tietää mitä tapahtuu vain hetki ennen lähetystä? - Nyt se selvisi!',
'description': 'md5:cfc6ccf0e57a814360df464a91ff67d6', 'description': 'md5:cfc6ccf0e57a814360df464a91ff67d6',
@ -32,10 +32,9 @@ class RuutuIE(InfoExtractor):
'md5': '065a10ae4d5b8cfd9d0c3d332465e3d9', 'md5': '065a10ae4d5b8cfd9d0c3d332465e3d9',
'info_dict': { 'info_dict': {
'id': '2057306', 'id': '2057306',
'display_id': 'superpesis-katso-koko-kausi-ruudussa',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Superpesis: katso koko kausi Ruudussa', 'title': 'Superpesis: katso koko kausi Ruudussa',
'description': 'md5:44c44a99fdbe5b380ab74ebd75f0af77', 'description': 'md5:da2736052fef3b2bd5e0005e63c25eac',
'thumbnail': 're:^https?://.*\.jpg$', 'thumbnail': 're:^https?://.*\.jpg$',
'duration': 40, 'duration': 40,
'age_limit': 0, 'age_limit': 0,
@ -44,29 +43,10 @@ class RuutuIE(InfoExtractor):
] ]
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, display_id) video_xml = self._download_xml(
'http://gatling.ruutu.fi/media-xml-cache?id=%s' % video_id, video_id)
video_id = self._search_regex(
r'data-media-id="(\d+)"', webpage, 'media id')
video_xml_url = None
media_data = self._search_regex(
r'jQuery\.extend\([^,]+,\s*(.+?)\);', webpage,
'media data', default=None)
if media_data:
media_json = self._parse_json(media_data, display_id, fatal=False)
if media_json:
xml_url = media_json.get('ruutuplayer', {}).get('xmlUrl')
if xml_url:
video_xml_url = xml_url.replace('{ID}', video_id)
if not video_xml_url:
video_xml_url = 'http://gatling.ruutu.fi/media-xml-cache?id=%s' % video_id
video_xml = self._download_xml(video_xml_url, video_id)
formats = [] formats = []
processed_urls = [] processed_urls = []
@ -109,10 +89,9 @@ class RuutuIE(InfoExtractor):
return { return {
'id': video_id, 'id': video_id,
'display_id': display_id, 'title': xpath_attr(video_xml, './/Behavior/Program', 'program_name', 'title', fatal=True),
'title': self._og_search_title(webpage), 'description': xpath_attr(video_xml, './/Behavior/Program', 'description', 'description'),
'description': self._og_search_description(webpage), 'thumbnail': xpath_attr(video_xml, './/Behavior/Startpicture', 'href', 'thumbnail'),
'thumbnail': self._og_search_thumbnail(webpage),
'duration': int_or_none(xpath_text(video_xml, './/Runtime', 'duration')), 'duration': int_or_none(xpath_text(video_xml, './/Runtime', 'duration')),
'age_limit': int_or_none(xpath_text(video_xml, './/AgeLimit', 'age limit')), 'age_limit': int_or_none(xpath_text(video_xml, './/AgeLimit', 'age limit')),
'formats': formats, 'formats': formats,