projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Merge branch 'master' of github.com:rg3/youtube-dl
[youtube-dl]
/
youtube_dl
/
extractor
/
rutube.py
diff --git
a/youtube_dl/extractor/rutube.py
b/youtube_dl/extractor/rutube.py
index 7c66074602521aea74f31da096be2c86a8281de8..b72b5a5869ae3ae6044ac210d55eca9df8a5ccf8 100644
(file)
--- a/
youtube_dl/extractor/rutube.py
+++ b/
youtube_dl/extractor/rutube.py
@@
-2,14
+2,15
@@
from __future__ import unicode_literals
import re
from __future__ import unicode_literals
import re
-import json
import itertools
from .common import InfoExtractor
import itertools
from .common import InfoExtractor
-from ..
utils
import (
+from ..
compat
import (
compat_str,
compat_str,
- unified_strdate,
+)
+from ..utils import (
ExtractorError,
ExtractorError,
+ unified_strdate,
)
)
@@
-37,24
+38,22
@@
class RutubeIE(InfoExtractor):
}
def _real_extract(self, url):
}
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
-
- api_response = self._download_webpage(
+ video_id = self._match_id(url)
+ video = self._download_json(
'http://rutube.ru/api/video/%s/?format=json' % video_id,
video_id, 'Downloading video JSON')
'http://rutube.ru/api/video/%s/?format=json' % video_id,
video_id, 'Downloading video JSON')
- video = json.loads(api_response)
-
- api_response = self._download_webpage(
- 'http://rutube.ru/api/play/trackinfo/%s/?format=json' % video_id,
- video_id, 'Downloading trackinfo JSON')
- trackinfo = json.loads(api_response)
-
+
# Some videos don't have the author field
# Some videos don't have the author field
- author = trackinfo.get('author') or {}
- m3u8_url = trackinfo['video_balancer'].get('m3u8')
+ author = video.get('author') or {}
+
+ options = self._download_json(
+ 'http://rutube.ru/api/play/options/%s/?format=json' % video_id,
+ video_id, 'Downloading options JSON')
+
+ m3u8_url = options['video_balancer'].get('m3u8')
if m3u8_url is None:
raise ExtractorError('Couldn\'t find m3u8 manifest url')
if m3u8_url is None:
raise ExtractorError('Couldn\'t find m3u8 manifest url')
+ formats = self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4')
return {
'id': video['id'],
return {
'id': video['id'],
@@
-62,8
+61,7
@@
class RutubeIE(InfoExtractor):
'description': video['description'],
'duration': video['duration'],
'view_count': video['hits'],
'description': video['description'],
'duration': video['duration'],
'view_count': video['hits'],
- 'url': m3u8_url,
- 'ext': 'mp4',
+ 'formats': formats,
'thumbnail': video['thumbnail_url'],
'uploader': author.get('name'),
'uploader_id': compat_str(author['id']) if author else None,
'thumbnail': video['thumbnail_url'],
'uploader': author.get('name'),
'uploader_id': compat_str(author['id']) if author else None,
@@
-76,16
+74,22
@@
class RutubeChannelIE(InfoExtractor):
IE_NAME = 'rutube:channel'
IE_DESC = 'Rutube channels'
_VALID_URL = r'http://rutube\.ru/tags/video/(?P<id>\d+)'
IE_NAME = 'rutube:channel'
IE_DESC = 'Rutube channels'
_VALID_URL = r'http://rutube\.ru/tags/video/(?P<id>\d+)'
+ _TESTS = [{
+ 'url': 'http://rutube.ru/tags/video/1800/',
+ 'info_dict': {
+ 'id': '1800',
+ },
+ 'playlist_mincount': 68,
+ }]
_PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json'
def _extract_videos(self, channel_id, channel_title=None):
entries = []
for pagenum in itertools.count(1):
_PAGE_TEMPLATE = 'http://rutube.ru/api/tags/video/%s/?page=%s&format=json'
def _extract_videos(self, channel_id, channel_title=None):
entries = []
for pagenum in itertools.count(1):
-
api_response = self._download_webpage
(
+
page = self._download_json
(
self._PAGE_TEMPLATE % (channel_id, pagenum),
channel_id, 'Downloading page %s' % pagenum)
self._PAGE_TEMPLATE % (channel_id, pagenum),
channel_id, 'Downloading page %s' % pagenum)
- page = json.loads(api_response)
results = page['results']
if not results:
break
results = page['results']
if not results:
break
@@
-104,17
+108,16
@@
class RutubeMovieIE(RutubeChannelIE):
IE_NAME = 'rutube:movie'
IE_DESC = 'Rutube movies'
_VALID_URL = r'http://rutube\.ru/metainfo/tv/(?P<id>\d+)'
IE_NAME = 'rutube:movie'
IE_DESC = 'Rutube movies'
_VALID_URL = r'http://rutube\.ru/metainfo/tv/(?P<id>\d+)'
+ _TESTS = []
_MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json'
_PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json'
def _real_extract(self, url):
_MOVIE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/?format=json'
_PAGE_TEMPLATE = 'http://rutube.ru/api/metainfo/tv/%s/video?page=%s&format=json'
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- movie_id = mobj.group('id')
- api_response = self._download_webpage(
+ movie_id = self._match_id(url)
+ movie = self._download_json(
self._MOVIE_TEMPLATE % movie_id, movie_id,
'Downloading movie JSON')
self._MOVIE_TEMPLATE % movie_id, movie_id,
'Downloading movie JSON')
- movie = json.loads(api_response)
movie_name = movie['name']
return self._extract_videos(movie_id, movie_name)
movie_name = movie['name']
return self._extract_videos(movie_id, movie_name)
@@
-123,5
+126,12
@@
class RutubePersonIE(RutubeChannelIE):
IE_NAME = 'rutube:person'
IE_DESC = 'Rutube person videos'
_VALID_URL = r'http://rutube\.ru/video/person/(?P<id>\d+)'
IE_NAME = 'rutube:person'
IE_DESC = 'Rutube person videos'
_VALID_URL = r'http://rutube\.ru/video/person/(?P<id>\d+)'
+ _TESTS = [{
+ 'url': 'http://rutube.ru/video/person/313878/',
+ 'info_dict': {
+ 'id': '313878',
+ },
+ 'playlist_mincount': 37,
+ }]
_PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json'
_PAGE_TEMPLATE = 'http://rutube.ru/api/video/person/%s/?page=%s&format=json'