[TMZ] Add support for new page structure using JSON-LD

This commit is contained in:
Diego Fernando Rodríguez Varón 2020-11-08 15:36:41 -05:00
parent 60351178a5
commit fff5071112
2 changed files with 99 additions and 94 deletions

View File

@ -1178,10 +1178,7 @@
from .threeqsdn import ThreeQSDNIE
from .tiktok import TikTokIE
from .tinypic import TinyPicIE
from .tmz import (
TMZIE,
TMZArticleIE,
)
from .tmz import TMZIE
from .tnaflix import (
TNAFlixNetworkEmbedIE,
TNAFlixIE,

View File

@ -5,95 +5,103 @@
class TMZIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tmz\.com/videos/.*(?P<id>[^/?#]{10,10})'
_TESTS = [{
'url': 'http://www.tmz.com/videos/0_okj015ty/',
'md5': '4d22a51ef205b6c06395d8394f72d560',
'info_dict': {
'id': '0_okj015ty',
'ext': 'mp4',
'title': 'Kim Kardashian\'s Boobs Unlock a Mystery!',
'timestamp': 1394747163,
'uploader_id': 'batchUser',
'upload_date': '20140313',
}
}, {
'url': 'http://www.tmz.com/videos/0-cegprt2p/',
'info_dict': {
'id': '0_cegprt2p',
'ext': 'mp4',
'title': "No Charges Against Hillary Clinton? Harvey Says It Ain't Over Yet",
'timestamp': 1467831837,
'uploader_id': 'batchUser',
'upload_date': '20160706',
}
}, {
'url': 'https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/',
'info_dict': {
'id': '0_zcsejvcr',
'ext': 'mxf',
'title': "Angry Bagel Shop Guy Says He Doesn't Trust Women",
'timestamp': 1562889485,
'uploader_id': 'batchUser',
'upload_date': '20190711',
}
}]
_VALID_URL = r"https?://(?:www\.)?tmz\.com/.*"
_TESTS = [
{
"url": "http://www.tmz.com/videos/0-cegprt2p/",
"info_dict": {
"id": "http://www.tmz.com/videos/0-cegprt2p/",
"ext": "mp4",
"title": "No Charges Against Hillary Clinton? Harvey Says It Ain't Over Yet",
"description": "Harvey talks about Director Comeys decision not to prosecute Hillary Clinton.",
"timestamp": 1467831837,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20160706",
},
},
{
"url": "https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/",
"info_dict": {
"id": "https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/",
"ext": "mp4",
"title": "Angry Bagel Shop Guy Says He Doesn't Trust Women",
"description": "The enraged man who went viral for ranting about women on dating sites before getting ragdolled in a bagel shop is defending his misogyny ... he says it's women's fault in the first place.",
"timestamp": 1562889485,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20190711",
},
},
{
"url": "http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert",
"md5": "5429c85db8bde39a473a56ca8c4c5602",
"info_dict": {
"id": "http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert",
"ext": "mp4",
"title": "Bobby Brown Tells Crowd ... Bobbi Kristina is Awake",
"description": 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake. She\'s watching me."',
"timestamp": 1429467813,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20150419",
},
},
{
"url": "http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/",
"info_dict": {
"id": "http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/",
"ext": "mp4",
"title": "Patti LaBelle -- Goes Nuclear On Stripping Fan",
"description": "Patti LaBelle made it known loud and clear last night ... NO "
"ONE gets on her stage and strips down.",
"timestamp": 1442683746,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20150919",
},
},
{
"url": "http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/",
"info_dict": {
"id": "http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/",
"ext": "mp4",
"title": "NBA's Adam Silver -- Blake Griffin's a Great Guy ... He'll Learn from This",
"description": "Two pretty parts of this video with NBA Commish Adam Silver.",
"timestamp": 1454010989,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20160128",
},
},
{
"url": "http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/",
"info_dict": {
"id": "http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/",
"ext": "mp4",
"title": "Trump Star Vandal -- I'm Not Afraid of Donald or the Cops!",
"description": "James Otis is the the guy who took a pickaxe to Donald Trump's star on the Walk of Fame, and he tells TMZ .. he's ready and willing to go to jail for the crime.",
"timestamp": 1477500095,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20161026",
},
},
{
"url": "https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/",
"info_dict": {
"id": "https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/",
"ext": "mp4",
"title": "Cops Use Billy Clubs Against Pro-Trump and Anti-Fascist "
"Demonstrators",
"description": "Beverly Hills may be an omen of what's coming next week, "
"because things got crazy on the streets and cops started "
"swinging their billy clubs at both Anti-Fascist and Pro-Trump "
"demonstrators.",
"timestamp": 1604182772,
"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
"upload_date": "20201031",
},
},
]
def _real_extract(self, url):
video_id = self._match_id(url).replace('-', '_')
return self.url_result('kaltura:591531:%s' % video_id, 'Kaltura', video_id)
class TMZArticleIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tmz\.com/\d{4}/\d{2}/\d{2}/(?P<id>[^/]+)/?'
_TESTS = [{
'url': 'http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert',
'md5': '5429c85db8bde39a473a56ca8c4c5602',
'info_dict': {
'id': '0_6snoelag',
'ext': 'mp4',
'title': 'Bobby Brown Tells Crowd ... Bobbi Kristina is Awake',
'timestamp': 1429467813,
'upload_date': '20150419',
'uploader_id': 'batchUser',
}
}, {
'url': 'http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/',
'info_dict': {
'id': '0_jerz7s3l',
'ext': 'mp4',
'title': 'Patti LaBelle -- Goes Nuclear On Stripping Fan',
'timestamp': 1442683746,
'upload_date': '20150919',
'uploader_id': 'batchUser',
}
}, {
'url': 'http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/',
'info_dict': {
'id': '0_ytz87kk7',
'ext': 'mp4',
'title': "NBA's Adam Silver -- Blake Griffin's a Great Guy ... He'll Learn from This",
'timestamp': 1454010989,
'upload_date': '20160128',
'uploader_id': 'batchUser',
}
}, {
'url': 'http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/',
'info_dict': {
'id': '0_isigfatu',
'ext': 'mp4',
'title': "Trump Star Vandal -- I'm Not Afraid of Donald or the Cops!",
'timestamp': 1477500095,
'upload_date': '20161026',
'uploader_id': 'batchUser',
}
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
params = self._html_search_regex(r'TMZ.actions.clickLink\(([\s\S]+?)\)',
webpage, 'embedded video info').split(',')
new_url = params[0].strip("'\"")
if new_url != url:
return self.url_result(new_url)
webpage = self._download_webpage(url, url)
jsonld = self._search_json_ld(webpage, url)
if id not in jsonld:
jsonld["id"] = url
return jsonld