[yt-dlp.git] / yt_dlp / extractor / tmz.py

# coding: utf-8
from __future__ import unicode_literals

import re

from .common import InfoExtractor
from ..utils import (
    ExtractorError,
    get_element_by_attribute,
)


class TMZIE(InfoExtractor):
    _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 Comey’s 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",
            },
        },
        {
            "url": "https://www.tmz.com/2020/11/05/gervonta-davis-car-crash-hit-and-run-police/",
            "info_dict": {
                "id": "Dddb6IGe-ws",
                "ext": "mp4",
                "title": "SICK LAMBO GERVONTA DAVIS IN HIS NEW RIDE RIGHT AFTER KO AFTER LEO  EsNews Boxing",
                "uploader": "ESNEWS",
                "description": "md5:49675bc58883ccf80474b8aa701e1064",
                "upload_date": "20201101",
                "uploader_id": "ESNEWS",
            },
        },
        {
            "url": "https://www.tmz.com/2020/11/19/conor-mcgregor-dustin-poirier-contract-fight-ufc-257-fight-island/",
            "info_dict": {
                "id": "1329450007125225473",
                "ext": "mp4",
                "title": "TheMacLife - BREAKING: Conor McGregor (@thenotoriousmma) has signed his bout agreement for his rematch with Dustin Poirier for January 23.",
                "uploader": "TheMacLife",
                "description": "md5:56e6009bbc3d12498e10d08a8e1f1c69",
                "upload_date": "20201119",
                "uploader_id": "Maclifeofficial",
                "timestamp": 1605800556,
            },
        },
    ]

    def _real_extract(self, url):
        webpage = self._download_webpage(url, url)
        jsonld = self._search_json_ld(webpage, url)
        if not jsonld or "url" not in jsonld:
            # try to extract from YouTube Player API
            # see https://developers.google.com/youtube/iframe_api_reference#Video_Queueing_Functions
            match_obj = re.search(r'\.cueVideoById\(\s*(?P<quote>[\'"])(?P<id>.*?)(?P=quote)', webpage)
            if match_obj:
                res = self.url_result(match_obj.group("id"))
                return res
            # try to extract from twitter
            blockquote_el = get_element_by_attribute("class", "twitter-tweet", webpage)
            if blockquote_el:
                matches = re.findall(
                    r'<a[^>]+href=\s*(?P<quote>[\'"])(?P<link>.*?)(?P=quote)',
                    blockquote_el)
                if matches:
                    for _, match in matches:
                        if "/status/" in match:
                            res = self.url_result(match)
                            return res
            raise ExtractorError("No video found!")
        if id not in jsonld:
            jsonld["id"] = url
        return jsonld
Commit	Line	Data
f0c3d729 MR	1	# coding: utf-8
	2	from __future__ import unicode_literals
	3
d71eb83b DFRV	4	import re
d71eb83b DFRV	5
f0c3d729	6	from .common import InfoExtractor
d71eb83b DFRV	7	from ..utils import (
	8	ExtractorError,
	9	get_element_by_attribute,
	10	)
f0c3d729 MR	11
	12
	13	class TMZIE(InfoExtractor):
fff50711 DFRV	14	_VALID_URL = r"https?://(?:www\.)?tmz\.com/.*"
	15	_TESTS = [
	16	{
	17	"url": "http://www.tmz.com/videos/0-cegprt2p/",
	18	"info_dict": {
	19	"id": "http://www.tmz.com/videos/0-cegprt2p/",
	20	"ext": "mp4",
	21	"title": "No Charges Against Hillary Clinton? Harvey Says It Ain't Over Yet",
	22	"description": "Harvey talks about Director Comey’s decision not to prosecute Hillary Clinton.",
	23	"timestamp": 1467831837,
	24	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
	25	"upload_date": "20160706",
	26	},
	27	},
	28	{
	29	"url": "https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/",
	30	"info_dict": {
	31	"id": "https://www.tmz.com/videos/071119-chris-morgan-women-4590005-0-zcsejvcr/",
	32	"ext": "mp4",
	33	"title": "Angry Bagel Shop Guy Says He Doesn't Trust Women",
	34	"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.",
	35	"timestamp": 1562889485,
	36	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
	37	"upload_date": "20190711",
	38	},
	39	},
	40	{
	41	"url": "http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert",
	42	"md5": "5429c85db8bde39a473a56ca8c4c5602",
	43	"info_dict": {
	44	"id": "http://www.tmz.com/2015/04/19/bobby-brown-bobbi-kristina-awake-video-concert",
	45	"ext": "mp4",
	46	"title": "Bobby Brown Tells Crowd ... Bobbi Kristina is Awake",
	47	"description": 'Bobby Brown stunned his audience during a concert Saturday night, when he told the crowd, "Bobbi is awake. She\'s watching me."',
	48	"timestamp": 1429467813,
	49	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
	50	"upload_date": "20150419",
	51	},
	52	},
	53	{
	54	"url": "http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/",
	55	"info_dict": {
	56	"id": "http://www.tmz.com/2015/09/19/patti-labelle-concert-fan-stripping-kicked-out-nicki-minaj/",
	57	"ext": "mp4",
	58	"title": "Patti LaBelle -- Goes Nuclear On Stripping Fan",
	59	"description": "Patti LaBelle made it known loud and clear last night ... NO "
	60	"ONE gets on her stage and strips down.",
	61	"timestamp": 1442683746,
	62	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
	63	"upload_date": "20150919",
	64	},
	65	},
	66	{
	67	"url": "http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/",
	68	"info_dict": {
	69	"id": "http://www.tmz.com/2016/01/28/adam-silver-sting-drake-blake-griffin/",
	70	"ext": "mp4",
	71	"title": "NBA's Adam Silver -- Blake Griffin's a Great Guy ... He'll Learn from This",
	72	"description": "Two pretty parts of this video with NBA Commish Adam Silver.",
	73	"timestamp": 1454010989,
	74	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
	75	"upload_date": "20160128",
	76	},
	77	},
78	{
79	"url": "http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/",
80	"info_dict": {
81	"id": "http://www.tmz.com/2016/10/27/donald-trump-star-vandal-arrested-james-otis/",
82	"ext": "mp4",
83	"title": "Trump Star Vandal -- I'm Not Afraid of Donald or the Cops!",
84	"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.",
85	"timestamp": 1477500095,
86	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
87	"upload_date": "20161026",
88	},
89	},
90	{
91	"url": "https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/",
92	"info_dict": {
93	"id": "https://www.tmz.com/videos/2020-10-31-103120-beverly-hills-protest-4878209/",
94	"ext": "mp4",
95	"title": "Cops Use Billy Clubs Against Pro-Trump and Anti-Fascist "
96	"Demonstrators",
97	"description": "Beverly Hills may be an omen of what's coming next week, "
98	"because things got crazy on the streets and cops started "
99	"swinging their billy clubs at both Anti-Fascist and Pro-Trump "
100	"demonstrators.",
101	"timestamp": 1604182772,
102	"uploader": "{'@type': 'Person', 'name': 'TMZ Staff'}",
103	"upload_date": "20201031",
104	},
105	},
d71eb83b DFRV	106	{
	107	"url": "https://www.tmz.com/2020/11/05/gervonta-davis-car-crash-hit-and-run-police/",
	108	"info_dict": {
	109	"id": "Dddb6IGe-ws",
	110	"ext": "mp4",
	111	"title": "SICK LAMBO GERVONTA DAVIS IN HIS NEW RIDE RIGHT AFTER KO AFTER LEO EsNews Boxing",
	112	"uploader": "ESNEWS",
	113	"description": "md5:49675bc58883ccf80474b8aa701e1064",
	114	"upload_date": "20201101",
	115	"uploader_id": "ESNEWS",
	116	},
	117	},
	118	{
	119	"url": "https://www.tmz.com/2020/11/19/conor-mcgregor-dustin-poirier-contract-fight-ufc-257-fight-island/",
	120	"info_dict": {
	121	"id": "1329450007125225473",
	122	"ext": "mp4",
	123	"title": "TheMacLife - BREAKING: Conor McGregor (@thenotoriousmma) has signed his bout agreement for his rematch with Dustin Poirier for January 23.",
	124	"uploader": "TheMacLife",
	125	"description": "md5:56e6009bbc3d12498e10d08a8e1f1c69",
	126	"upload_date": "20201119",
	127	"uploader_id": "Maclifeofficial",
	128	"timestamp": 1605800556,
	129	},
	130	},
fff50711	131	]
f0c3d729 MR	132
f0c3d729 MR	133	def _real_extract(self, url):
fff50711 DFRV	134	webpage = self._download_webpage(url, url)
fff50711 DFRV	135	jsonld = self._search_json_ld(webpage, url)
d71eb83b DFRV	136	if not jsonld or "url" not in jsonld:
	137	# try to extract from YouTube Player API
	138	# see https://developers.google.com/youtube/iframe_api_reference#Video_Queueing_Functions
	139	match_obj = re.search(r'\.cueVideoById\(\s(?P<quote>[\'"])(?P<id>.?)(?P=quote)', webpage)
	140	if match_obj:
	141	res = self.url_result(match_obj.group("id"))
	142	return res
	143	# try to extract from twitter
	144	blockquote_el = get_element_by_attribute("class", "twitter-tweet", webpage)
	145	if blockquote_el:
	146	matches = re.findall(
	147	r'<a[^>]+href=\s(?P<quote>[\'"])(?P<link>.?)(?P=quote)',
	148	blockquote_el)
	149	if matches:
	150	for _, match in matches:
	151	if "/status/" in match:
	152	res = self.url_result(match)
	153	return res
	154	raise ExtractorError("No video found!")
fff50711 DFRV	155	if id not in jsonld:
	156	jsonld["id"] = url
	157	return jsonld