Roxedus/ytdl
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[gamespot] Add support for videos that don't use 'f4m_stream' (fixes #5707)

Browse Source
This commit is contained in:
Jaime Marquínez Ferrándiz 2015-05-15 18:42:59 +02:00
parent 62c95fd5fc
commit 34fe5a94ba
1 changed files with 40 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

60
youtube_dl/extractor/gamespot.py
View File

@ -15,7 +15,7 @@ from ..utils import (
class GameSpotIE(InfoExtractor): class GameSpotIE(InfoExtractor):
_VALID_URL = r'(?:http://)?(?:www\.)?gamespot\.com/.*-(?P<id>\d+)/?' _VALID_URL = r'(?:http://)?(?:www\.)?gamespot\.com/.*-(?P<id>\d+)/?'
_TEST = { _TESTS = [{
'url': 'http://www.gamespot.com/videos/arma-3-community-guide-sitrep-i/2300-6410818/', 'url': 'http://www.gamespot.com/videos/arma-3-community-guide-sitrep-i/2300-6410818/',
'md5': 'b2a30deaa8654fcccd43713a6b6a4825', 'md5': 'b2a30deaa8654fcccd43713a6b6a4825',
'info_dict': { 'info_dict': {
@ -23,8 +23,16 @@ class GameSpotIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'title': 'Arma 3 - Community Guide: SITREP I', 'title': 'Arma 3 - Community Guide: SITREP I',
'description': 'Check out this video where some of the basics of Arma 3 is explained.', 'description': 'Check out this video where some of the basics of Arma 3 is explained.',
} },
} }, {
'url': 'http://www.gamespot.com/videos/the-witcher-3-wild-hunt-xbox-one-now-playing/2300-6424837/',
'info_dict': {
'id': 'gs-2300-6424837',
'ext': 'flv',
'title': 'The Witcher 3: Wild Hunt [Xbox ONE] - Now Playing',
'description': 'Join us as we take a look at the early hours of The Witcher 3: Wild Hunt and more.',
},
}]
def _real_extract(self, url): def _real_extract(self, url):
page_id = self._match_id(url) page_id = self._match_id(url)
@ -32,25 +40,37 @@ class GameSpotIE(InfoExtractor):
data_video_json = self._search_regex( data_video_json = self._search_regex(
r'data-video=["\'](.*?)["\']', webpage, 'data video') r'data-video=["\'](.*?)["\']', webpage, 'data video')
data_video = json.loads(unescapeHTML(data_video_json)) data_video = json.loads(unescapeHTML(data_video_json))
streams = data_video['videoStreams']
# Transform the manifest url to a link to the mp4 files
# they are used in mobile devices.
f4m_url = data_video['videoStreams']['f4m_stream']
f4m_path = compat_urlparse.urlparse(f4m_url).path
QUALITIES_RE = r'((,\d+)+,?)'
qualities = self._search_regex(QUALITIES_RE, f4m_path, 'qualities').strip(',').split(',')
http_path = f4m_path[1:].split('/', 1)[1]
http_template = re.sub(QUALITIES_RE, r'%s', http_path)
http_template = http_template.replace('.csmil/manifest.f4m', '')
http_template = compat_urlparse.urljoin(
'http://video.gamespotcdn.com/', http_template)
formats = [] formats = []
for q in qualities: f4m_url = streams.get('f4m_stream')
formats.append({ if f4m_url is not None:
'url': http_template % q, # Transform the manifest url to a link to the mp4 files
'ext': 'mp4', # they are used in mobile devices.
'format_id': q, f4m_path = compat_urlparse.urlparse(f4m_url).path
}) QUALITIES_RE = r'((,\d+)+,?)'
qualities = self._search_regex(QUALITIES_RE, f4m_path, 'qualities').strip(',').split(',')
http_path = f4m_path[1:].split('/', 1)[1]
http_template = re.sub(QUALITIES_RE, r'%s', http_path)
http_template = http_template.replace('.csmil/manifest.f4m', '')
http_template = compat_urlparse.urljoin(
'http://video.gamespotcdn.com/', http_template)
for q in qualities:
formats.append({
'url': http_template % q,
'ext': 'mp4',
'format_id': q,
})
else:
for quality in ['sd', 'hd']:
# It's actually a link to a flv file
flv_url = streams.get('f4m_{0}'.format(quality))
if flv_url is not None:
formats.append({
'url': flv_url,
'ext': 'flv',
'format_id': quality,
})
return { return {
'id': data_video['guid'], 'id': data_video['guid'],