]>
Commit | Line | Data |
---|---|---|
7eeb5bef PH |
1 | from __future__ import unicode_literals |
2 | ||
a37f27ae PH |
3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
572a89cc | 6 | from ..utils import int_or_none |
a37f27ae PH |
7 | |
8 | ||
9 | class LiveLeakIE(InfoExtractor): | |
e793f767 | 10 | _VALID_URL = r'https?://(?:\w+\.)?liveleak\.com/view\?(?:.*?)i=(?P<id>[\w_]+)(?:.*)' |
89acb969 | 11 | _TESTS = [{ |
7eeb5bef | 12 | 'url': 'http://www.liveleak.com/view?i=757_1364311680', |
116283ff | 13 | 'md5': '0813c2430bea7a46bf13acf3406992f4', |
7eeb5bef | 14 | 'info_dict': { |
572a89cc | 15 | 'id': '757_1364311680', |
116283ff | 16 | 'ext': 'mp4', |
7eeb5bef PH |
17 | 'description': 'extremely bad day for this guy..!', |
18 | 'uploader': 'ljfriel2', | |
e2ee97dc | 19 | 'title': 'Most unlucky car accident', |
ec85ded8 | 20 | 'thumbnail': r're:^https?://.*\.jpg$' |
6f5ac90c | 21 | } |
8bcc8756 | 22 | }, { |
89acb969 | 23 | 'url': 'http://www.liveleak.com/view?i=f93_1390833151', |
116283ff | 24 | 'md5': 'd3f1367d14cc3c15bf24fbfbe04b9abf', |
89acb969 | 25 | 'info_dict': { |
572a89cc PH |
26 | 'id': 'f93_1390833151', |
27 | 'ext': 'mp4', | |
89acb969 PH |
28 | 'description': 'German Television Channel NDR does an exclusive interview with Edward Snowden.\r\nUploaded on LiveLeak cause German Television thinks the rest of the world isn\'t intereseted in Edward Snowden.', |
29 | 'uploader': 'ARD_Stinkt', | |
30 | 'title': 'German Television does first Edward Snowden Interview (ENGLISH)', | |
ec85ded8 | 31 | 'thumbnail': r're:^https?://.*\.jpg$' |
89acb969 | 32 | } |
8bcc8756 | 33 | }, { |
116283ff | 34 | # Prochan embed |
572a89cc PH |
35 | 'url': 'http://www.liveleak.com/view?i=4f7_1392687779', |
36 | 'md5': '42c6d97d54f1db107958760788c5f48f', | |
37 | 'info_dict': { | |
38 | 'id': '4f7_1392687779', | |
39 | 'ext': 'mp4', | |
40 | 'description': "The guy with the cigarette seems amazingly nonchalant about the whole thing... I really hope my friends' reactions would be a bit stronger.\r\n\r\nAction-go to 0:55.", | |
41 | 'uploader': 'CapObveus', | |
42 | 'title': 'Man is Fatally Struck by Reckless Car While Packing up a Moving Truck', | |
43 | 'age_limit': 18, | |
116283ff YCH |
44 | }, |
45 | 'skip': 'Video is dead', | |
8f75761f | 46 | }, { |
b95cfa91 | 47 | # Covers https://github.com/rg3/youtube-dl/pull/5983 |
116283ff | 48 | # Multiple resolutions |
8f75761f | 49 | 'url': 'http://www.liveleak.com/view?i=801_1409392012', |
116283ff | 50 | 'md5': 'c3a449dbaca5c0d1825caecd52a57d7b', |
8f75761f | 51 | 'info_dict': { |
52 | 'id': '801_1409392012', | |
53 | 'ext': 'mp4', | |
611c1dd9 | 54 | 'description': 'Happened on 27.7.2014. \r\nAt 0:53 you can see people still swimming at near beach.', |
8f75761f | 55 | 'uploader': 'bony333', |
e2ee97dc | 56 | 'title': 'Crazy Hungarian tourist films close call waterspout in Croatia', |
ec85ded8 | 57 | 'thumbnail': r're:^https?://.*\.jpg$' |
8f75761f | 58 | } |
3779d524 V |
59 | }, { |
60 | # Covers https://github.com/rg3/youtube-dl/pull/10664#issuecomment-247439521 | |
8b0d3ee6 | 61 | 'url': 'http://m.liveleak.com/view?i=763_1473349649', |
3779d524 V |
62 | 'add_ie': ['Youtube'], |
63 | 'info_dict': { | |
64 | 'id': '763_1473349649', | |
65 | 'ext': 'mp4', | |
66 | 'title': 'Reporters and public officials ignore epidemic of black on asian violence in Sacramento | Colin Flaherty', | |
67 | 'description': 'Colin being the warrior he is and showing the injustice Asians in Sacramento are being subjected to.', | |
68 | 'uploader': 'Ziz', | |
69 | 'upload_date': '20160908', | |
70 | 'uploader_id': 'UCEbta5E_jqlZmEJsriTEtnw' | |
8b0d3ee6 S |
71 | }, |
72 | 'params': { | |
73 | 'skip_download': True, | |
74 | }, | |
381ad4f3 YCH |
75 | }, { |
76 | 'url': 'https://www.liveleak.com/view?i=677_1439397581', | |
77 | 'info_dict': { | |
78 | 'id': '677_1439397581', | |
79 | 'title': 'Fuel Depot in China Explosion caught on video', | |
80 | }, | |
81 | 'playlist_count': 3, | |
89acb969 | 82 | }] |
a37f27ae | 83 | |
b8f67449 | 84 | @staticmethod |
09747ba7 YCH |
85 | def _extract_urls(webpage): |
86 | return re.findall( | |
87 | r'<iframe[^>]+src="(https?://(?:\w+\.)?liveleak\.com/ll_embed\?[^"]*[if]=[\w_]+[^"]+)"', | |
b8f67449 | 88 | webpage) |
b8f67449 | 89 | |
a37f27ae | 90 | def _real_extract(self, url): |
e793f767 | 91 | video_id = self._match_id(url) |
a37f27ae | 92 | webpage = self._download_webpage(url, video_id) |
572a89cc PH |
93 | |
94 | video_title = self._og_search_title(webpage).replace('LiveLeak.com -', '').strip() | |
95 | video_description = self._og_search_description(webpage) | |
96 | video_uploader = self._html_search_regex( | |
97 | r'By:.*?(\w+)</a>', webpage, 'uploader', fatal=False) | |
98 | age_limit = int_or_none(self._search_regex( | |
99 | r'you confirm that you are ([0-9]+) years and over.', | |
100 | webpage, 'age limit', default=None)) | |
e2ee97dc | 101 | video_thumbnail = self._og_search_thumbnail(webpage) |
572a89cc | 102 | |
116283ff YCH |
103 | entries = self._parse_html5_media_entries(url, webpage, video_id) |
104 | if not entries: | |
105 | # Maybe an embed? | |
106 | embed_url = self._search_regex( | |
107 | r'<iframe[^>]+src="((?:https?:)?//(?:www\.)?(?:prochan|youtube)\.com/embed[^"]+)"', | |
108 | webpage, 'embed URL') | |
109 | return { | |
110 | '_type': 'url_transparent', | |
111 | 'url': embed_url, | |
112 | 'id': video_id, | |
113 | 'title': video_title, | |
114 | 'description': video_description, | |
115 | 'uploader': video_uploader, | |
116 | 'age_limit': age_limit, | |
117 | } | |
89acb969 | 118 | |
381ad4f3 YCH |
119 | for idx, info_dict in enumerate(entries): |
120 | for a_format in info_dict['formats']: | |
121 | if not a_format.get('height'): | |
122 | a_format['height'] = int_or_none(self._search_regex( | |
123 | r'([0-9]+)p\.mp4', a_format['url'], 'height label', | |
124 | default=None)) | |
975fa541 | 125 | |
381ad4f3 | 126 | self._sort_formats(info_dict['formats']) |
3779d524 | 127 | |
381ad4f3 YCH |
128 | # Don't append entry ID for one-video pages to keep backward compatibility |
129 | if len(entries) > 1: | |
130 | info_dict['id'] = '%s_%s' % (video_id, idx + 1) | |
131 | else: | |
132 | info_dict['id'] = video_id | |
a37f27ae | 133 | |
381ad4f3 YCH |
134 | info_dict.update({ |
135 | 'title': video_title, | |
136 | 'description': video_description, | |
137 | 'uploader': video_uploader, | |
138 | 'age_limit': age_limit, | |
139 | 'thumbnail': video_thumbnail, | |
140 | }) | |
116283ff | 141 | |
381ad4f3 | 142 | return self.playlist_result(entries, video_id, video_title) |
09747ba7 YCH |
143 | |
144 | ||
145 | class LiveLeakEmbedIE(InfoExtractor): | |
146 | _VALID_URL = r'https?://(?:www\.)?liveleak\.com/ll_embed\?.*?\b(?P<kind>[if])=(?P<id>[\w_]+)' | |
147 | ||
148 | # See generic.py for actual test cases | |
149 | _TESTS = [{ | |
150 | 'url': 'https://www.liveleak.com/ll_embed?i=874_1459135191', | |
151 | 'only_matching': True, | |
152 | }, { | |
153 | 'url': 'https://www.liveleak.com/ll_embed?f=ab065df993c1', | |
154 | 'only_matching': True, | |
155 | }] | |
156 | ||
157 | def _real_extract(self, url): | |
158 | mobj = re.match(self._VALID_URL, url) | |
159 | kind, video_id = mobj.group('kind', 'id') | |
160 | ||
161 | if kind == 'f': | |
162 | webpage = self._download_webpage(url, video_id) | |
163 | liveleak_url = self._search_regex( | |
164 | r'logourl\s*:\s*(?P<q1>[\'"])(?P<url>%s)(?P=q1)' % LiveLeakIE._VALID_URL, | |
165 | webpage, 'LiveLeak URL', group='url') | |
166 | elif kind == 'i': | |
167 | liveleak_url = 'http://www.liveleak.com/view?i=%s' % video_id | |
168 | ||
169 | return self.url_result(liveleak_url, ie=LiveLeakIE.ie_key()) |