kumi/yt-dlp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[ign] Modernize

Browse source
This commit is contained in:
Jaime Marquínez Ferrándiz 2014-03-29 14:34:03 +01:00
parent 79bfd01001
commit 40c716d2a2
1 changed files with 50 additions and 52 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

102
youtube_dl/extractor/ign.py
View file

@ -1,10 +1,8 @@
from __future__ import unicode_literals
import re import re
import json
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import (
determine_ext,
)
class IGNIE(InfoExtractor): class IGNIE(InfoExtractor):
@ -14,52 +12,57 @@ class IGNIE(InfoExtractor):
""" """
_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))(/.+)?/(?P<name_or_id>.+)'
IE_NAME = u'ign.com' IE_NAME = 'ign.com'
_CONFIG_URL_TEMPLATE = 'http://www.ign.com/videos/configs/id/%s.config' _CONFIG_URL_TEMPLATE = 'http://www.ign.com/videos/configs/id/%s.config'
_DESCRIPTION_RE = [r'<span class="page-object-description">(.+?)</span>', _DESCRIPTION_RE = [
r'id="my_show_video">.*?<p>(.*?)</p>', r'<span class="page-object-description">(.+?)</span>',
] r'id="my_show_video">.*?<p>(.*?)</p>',
]
_TESTS = [ _TESTS = [
{ {
u'url': u'http://www.ign.com/videos/2013/06/05/the-last-of-us-review', 'url': 'http://www.ign.com/videos/2013/06/05/the-last-of-us-review',
u'file': u'8f862beef863986b2785559b9e1aa599.mp4', 'md5': 'eac8bdc1890980122c3b66f14bdd02e9',
u'md5': u'eac8bdc1890980122c3b66f14bdd02e9', 'info_dict': {
u'info_dict': { 'id': '8f862beef863986b2785559b9e1aa599',
u'title': u'The Last of Us Review', 'ext': 'mp4',
u'description': u'md5:c8946d4260a4d43a00d5ae8ed998870c', 'title': 'The Last of Us Review',
'description': 'md5:c8946d4260a4d43a00d5ae8ed998870c',
} }
}, },
{ {
u'url': u'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind', 'url': 'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind',
u'playlist': [ 'playlist': [
{ {
u'file': u'5ebbd138523268b93c9141af17bec937.mp4', 'info_dict': {
u'info_dict': { 'id': '5ebbd138523268b93c9141af17bec937',
u'title': u'GTA 5 Video Review', 'ext': 'mp4',
u'description': u'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.', '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.',
}, },
}, },
{ {
u'file': u'638672ee848ae4ff108df2a296418ee2.mp4', 'info_dict': {
u'info_dict': { 'id': '638672ee848ae4ff108df2a296418ee2',
u'title': u'26 Twisted Moments from GTA 5 in Slow Motion', 'ext': 'mp4',
u'description': u'The twisted beauty of GTA 5 in stunning slow motion.', 'title': '26 Twisted Moments from GTA 5 in Slow Motion',
'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
}, },
}, },
], ],
u'params': { 'params': {
u'skip_download': True, 'skip_download': True,
}, },
}, },
] ]
def _find_video_id(self, webpage): def _find_video_id(self, webpage):
res_id = [r'data-video-id="(.+?)"', res_id = [
r'<object id="vid_(.+?)"', r'data-video-id="(.+?)"',
r'<meta name="og:image" content=".*/(.+?)-(.+?)/.+.jpg"', 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')
def _real_extract(self, url): def _real_extract(self, url):
@ -68,7 +71,7 @@ class IGNIE(InfoExtractor):
page_type = mobj.group('type') page_type = mobj.group('type')
webpage = self._download_webpage(url, name_or_id) webpage = self._download_webpage(url, name_or_id)
if page_type == 'articles': if page_type == 'articles':
video_url = self._search_regex(r'var videoUrl = "(.+?)"', webpage, u'video url') video_url = self._search_regex(r'var videoUrl = "(.+?)"', webpage, 'video url')
return self.url_result(video_url, ie='IGN') return self.url_result(video_url, ie='IGN')
elif page_type != 'video': elif page_type != 'video':
multiple_urls = re.findall( multiple_urls = re.findall(
@ -80,41 +83,37 @@ class IGNIE(InfoExtractor):
video_id = self._find_video_id(webpage) video_id = self._find_video_id(webpage)
result = self._get_video_info(video_id) result = self._get_video_info(video_id)
description = self._html_search_regex(self._DESCRIPTION_RE, description = self._html_search_regex(self._DESCRIPTION_RE,
webpage, 'video description', webpage, 'video description', flags=re.DOTALL)
flags=re.DOTALL)
result['description'] = description result['description'] = description
return result return result
def _get_video_info(self, video_id): def _get_video_info(self, video_id):
config_url = self._CONFIG_URL_TEMPLATE % video_id config_url = self._CONFIG_URL_TEMPLATE % video_id
config = json.loads(self._download_webpage(config_url, video_id, config = self._download_json(config_url, video_id)
u'Downloading video info'))
media = config['playlist']['media'] media = config['playlist']['media']
video_url = media['url']
return {'id': media['metadata']['videoId'], return {
'url': video_url, 'id': media['metadata']['videoId'],
'ext': determine_ext(video_url), 'url': media['url'],
'title': media['metadata']['title'], 'title': media['metadata']['title'],
'thumbnail': media['poster'][0]['url'].replace('{size}', 'grande'), 'thumbnail': media['poster'][0]['url'].replace('{size}', 'grande'),
} }
class OneUPIE(IGNIE): class OneUPIE(IGNIE):
"""Extractor for 1up.com, it uses the ign videos system."""
_VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)' _VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)'
IE_NAME = '1up.com' IE_NAME = '1up.com'
_DESCRIPTION_RE = r'<div id="vid_summary">(.+?)</div>' _DESCRIPTION_RE = r'<div id="vid_summary">(.+?)</div>'
_TEST = { _TEST = {
u'url': u'http://gamevideos.1up.com/video/id/34976', 'url': 'http://gamevideos.1up.com/video/id/34976',
u'file': u'34976.mp4', 'md5': '68a54ce4ebc772e4b71e3123d413163d',
u'md5': u'68a54ce4ebc772e4b71e3123d413163d', 'info_dict': {
u'info_dict': { 'id': '34976',
u'title': u'Sniper Elite V2 - Trailer', 'ext': 'mp4',
u'description': u'md5:5d289b722f5a6d940ca3136e9dae89cf', 'title': 'Sniper Elite V2 - Trailer',
'description': 'md5:5d289b722f5a6d940ca3136e9dae89cf',
} }
} }
@ -123,7 +122,6 @@ class OneUPIE(IGNIE):
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
id = mobj.group('name_or_id')
result = super(OneUPIE, self)._real_extract(url) result = super(OneUPIE, self)._real_extract(url)
result['id'] = id result['id'] = mobj.group('name_or_id')
return result return result