w/youtube-dl
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Merge pull request #7045 from remitamine/ign

Browse source 
[ign] add support for pcmag and extract all formats and more metadata(fixes #5335)(fixes #7006)
This commit is contained in:
remitamine 2015-12-25 20:06:27 +01:00
commit 9f0ee2a388
2 changed files with 115 additions and 32 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
youtube_dl/extractor/__init__.py
View file

@ -259,7 +259,11 @@ from .howstuffworks import HowStuffWorksIE
from .huffpost import HuffPostIE
from .hypem import HypemIE
from .iconosquare import IconosquareIE
from .ign import IGNIE, OneUPIE
from .ign import (
IGNIE,
OneUPIE,
PCMagIE,
)
from .imdb import (
ImdbIE,
ImdbListIE

139
youtube_dl/extractor/ign.py
View file

@ -3,6 +3,10 @@ from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import (
int_or_none,
parse_iso8601,
)
class IGNIE(InfoExtractor):
@ -11,25 +15,24 @@ class IGNIE(InfoExtractor):
Some videos of it.ign.com are also supported
"""
_VALID_URL = r'https?://.+?\.ign\.com/(?P<type>videos|show_videos|articles|(?:[^/]*/feature))(/.+)?/(?P<name_or_id>.+)'
_VALID_URL = r'https?://.+?\.ign\.com/(?:[^/]+/)?(?P<type>videos|show_videos|articles|feature|(?:[^/]+/\d+/video))(/.+)?/(?P<name_or_id>.+)'
IE_NAME = 'ign.com'
_CONFIG_URL_TEMPLATE = 'http://www.ign.com/videos/configs/id/%s.config'
_DESCRIPTION_RE = [
r'<span class="page-object-description">(.+?)</span>',
r'id="my_show_video">.*?<p>(.*?)</p>',
r'<meta name="description" content="(.*?)"',
]
_API_URL_TEMPLATE = 'http://apis.ign.com/video/v3/videos/%s'
_EMBED_RE = r'<iframe[^>]+?["\']((?:https?:)?//.+?\.ign\.com.+?/embed.+?)["\']'
_TESTS = [
{
'url': 'http://www.ign.com/videos/2013/06/05/the-last-of-us-review',
'md5': 'eac8bdc1890980122c3b66f14bdd02e9',
'md5': 'febda82c4bafecd2d44b6e1a18a595f8',
'info_dict': {
'id': '8f862beef863986b2785559b9e1aa599',
'ext': 'mp4',
'title': 'The Last of Us Review',
'description': 'md5:c8946d4260a4d43a00d5ae8ed998870c',
'timestamp': 1370440800,
'upload_date': '20130605',
'uploader_id': 'cberidon@ign.com',
}
},
{
@ -44,6 +47,9 @@ class IGNIE(InfoExtractor):
'ext': 'mp4',
'title': 'GTA 5 Video Review',
'description': 'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.',
'timestamp': 1379339880,
'upload_date': '20130916',
'uploader_id': 'danieljkrupa@gmail.com',
},
},
{
@ -52,6 +58,9 @@ class IGNIE(InfoExtractor):
'ext': 'mp4',
'title': '26 Twisted Moments from GTA 5 in Slow Motion',
'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
'timestamp': 1386878820,
'upload_date': '20131212',
'uploader_id': 'togilvie@ign.com',
},
},
],
@ -66,12 +75,20 @@ class IGNIE(InfoExtractor):
'id': '078fdd005f6d3c02f63d795faa1b984f',
'ext': 'mp4',
'title': 'Rewind Theater - Wild Trailer Gamescom 2014',
'description': (
'Giant skeletons, bloody hunts, and captivating'
' natural beauty take our breath away.'
),
'description': 'Brian and Jared explore Michel Ancel\'s captivating new preview.',
'timestamp': 1408047180,
'upload_date': '20140814',
'uploader_id': 'jamesduggan1990@gmail.com',
},
},
{
'url': 'http://me.ign.com/en/videos/112203/video/how-hitman-aims-to-be-different-than-every-other-s',
'only_matching': True,
},
{
'url': 'http://me.ign.com/ar/angry-birds-2/106533/video/lrd-ldyy-lwl-lfylm-angry-birds',
'only_matching': True,
},
]
def _find_video_id(self, webpage):
@ -82,7 +99,7 @@ class IGNIE(InfoExtractor):
r'<object id="vid_(.+?)"',
r'<meta name="og:image" content=".*/(.+?)-(.+?)/.+.jpg"',
]
return self._search_regex(res_id, webpage, 'video id')
return self._search_regex(res_id, webpage, 'video id', default=None)
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
@ -91,7 +108,7 @@ class IGNIE(InfoExtractor):
webpage = self._download_webpage(url, name_or_id)
if page_type != 'video':
multiple_urls = re.findall(
'<param name="flashvars"[^>]*value="[^"]*?url=(https?://www\.ign\.com/videos/.*?)["&]',
r'<param name="flashvars"[^>]*value="[^"]*?url=(https?://www\.ign\.com/videos/.*?)["&]',
webpage)
if multiple_urls:
entries = [self.url_result(u, ie='IGN') for u in multiple_urls]
@ -102,22 +119,50 @@ class IGNIE(InfoExtractor):
}
video_id = self._find_video_id(webpage)
result = self._get_video_info(video_id)
description = self._html_search_regex(self._DESCRIPTION_RE,
webpage, 'video description', flags=re.DOTALL)
result['description'] = description
return result
if not video_id:
return self.url_result(self._search_regex(self._EMBED_RE, webpage, 'embed url'))
return self._get_video_info(video_id)
def _get_video_info(self, video_id):
config_url = self._CONFIG_URL_TEMPLATE % video_id
config = self._download_json(config_url, video_id)
media = config['playlist']['media']
api_data = self._download_json(self._API_URL_TEMPLATE % video_id, video_id)
formats = []
m3u8_url = api_data['refs'].get('m3uUrl')
if m3u8_url:
m3u8_formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
if m3u8_formats:
formats.extend(m3u8_formats)
f4m_url = api_data['refs'].get('f4mUrl')
if f4m_url:
f4m_formats = self._extract_f4m_formats(f4m_url, video_id, f4m_id='hds', fatal=False)
if f4m_formats:
formats.extend(f4m_formats)
for asset in api_data['assets']:
formats.append({
'url': asset['url'],
'tbr': asset.get('actual_bitrate_kbps'),
'fps': asset.get('frame_rate'),
'height': int_or_none(asset.get('height')),
'width': int_or_none(asset.get('width')),
})
self._sort_formats(formats)
thumbnails = [{
'url': thumbnail['url']
} for thumbnail in api_data.get('thumbnails', [])]
metadata = api_data['metadata']
return {
'id': media['metadata']['videoId'],
'url': media['url'],
'title': media['metadata']['title'],
'thumbnail': media['poster'][0]['url'].replace('{size}', 'grande'),
'id': api_data.get('videoId') or video_id,
'title': metadata.get('longTitle') or metadata.get('name') or metadata.get['title'],
'description': metadata.get('description'),
'timestamp': parse_iso8601(metadata.get('publishDate')),
'duration': int_or_none(metadata.get('duration')),
'display_id': metadata.get('slug') or video_id,
'uploader_id': metadata.get('creator'),
'thumbnails': thumbnails,
'formats': formats,
}
@ -125,16 +170,17 @@ class OneUPIE(IGNIE):
_VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)\.html'
IE_NAME = '1up.com'
_DESCRIPTION_RE = r'<div id="vid_summary">(.+?)</div>'
_TESTS = [{
'url': 'http://gamevideos.1up.com/video/id/34976.html',
'md5': '68a54ce4ebc772e4b71e3123d413163d',
'md5': 'c9cc69e07acb675c31a16719f909e347',
'info_dict': {
'id': '34976',
'ext': 'mp4',
'title': 'Sniper Elite V2 - Trailer',
'description': 'md5:5d289b722f5a6d940ca3136e9dae89cf',
'description': 'md5:bf0516c5ee32a3217aa703e9b1bc7826',
'timestamp': 1313099220,
'upload_date': '20110811',
'uploader_id': 'IGN',
}
}]
@ -143,3 +189,36 @@ class OneUPIE(IGNIE):
result = super(OneUPIE, self)._real_extract(url)
result['id'] = mobj.group('name_or_id')
return result
class PCMagIE(IGNIE):
_VALID_URL = r'https?://(?:www\.)?pcmag\.com/(?P<type>videos|article2)(/.+)?/(?P<name_or_id>.+)'
IE_NAME = 'pcmag'
_EMBED_RE = r'iframe.setAttribute\("src",\s*__util.objToUrlString\("http://widgets\.ign\.com/video/embed/content.html?[^"]*url=([^"]+)["&]'
_TESTS = [{
'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
'md5': '212d6154fd0361a2781075f1febbe9ad',
'info_dict': {
'id': 'ee10d774b508c9b8ec07e763b9125b91',
'ext': 'mp4',
'title': '010615_What\'s New Now: Is GoGo Snooping on Your Data?',
'description': 'md5:a7071ae64d2f68cc821c729d4ded6bb3',
'timestamp': 1420571160,
'upload_date': '20150106',
'uploader_id': 'cozzipix@gmail.com',
}
},{
'url': 'http://www.pcmag.com/article2/0,2817,2470156,00.asp',
'md5': '94130c1ca07ba0adb6088350681f16c1',
'info_dict': {
'id': '042e560ba94823d43afcb12ddf7142ca',
'ext': 'mp4',
'title': 'HTC\'s Weird New Re Camera - What\'s New Now',
'description': 'md5:53433c45df96d2ea5d0fda18be2ca908',
'timestamp': 1412953920,
'upload_date': '20141010',
'uploader_id': 'chris_snyder@pcmag.com',
}
}]