]>
Commit | Line | Data |
---|---|---|
1 | # coding: utf-8 | |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import re | |
5 | ||
6 | from .common import InfoExtractor | |
7 | from ..utils import ( | |
8 | js_to_json, | |
9 | unescapeHTML, | |
10 | ExtractorError, | |
11 | ) | |
12 | ||
13 | ||
14 | class DVTVIE(InfoExtractor): | |
15 | IE_NAME = 'dvtv' | |
16 | IE_DESC = 'http://video.aktualne.cz/' | |
17 | ||
18 | _VALID_URL = r'https?://video\.aktualne\.cz/(?:[^/]+/)+r~(?P<id>[0-9a-f]{32})' | |
19 | ||
20 | _TESTS = [{ | |
21 | 'url': 'http://video.aktualne.cz/dvtv/vondra-o-ceskem-stoleti-pri-pohledu-na-havla-mi-bylo-trapne/r~e5efe9ca855511e4833a0025900fea04/', | |
22 | 'md5': '67cb83e4a955d36e1b5d31993134a0c2', | |
23 | 'info_dict': { | |
24 | 'id': 'dc0768de855511e49e4b0025900fea04', | |
25 | 'ext': 'mp4', | |
26 | 'title': 'Vondra o Českém století: Při pohledu na Havla mi bylo trapně', | |
27 | } | |
28 | }, { | |
29 | 'url': 'http://video.aktualne.cz/dvtv/stropnicky-policie-vrbetice-preventivne-nekontrolovala/r~82ed4322849211e4a10c0025900fea04/', | |
30 | 'md5': '6388f1941b48537dbd28791f712af8bf', | |
31 | 'info_dict': { | |
32 | 'id': '72c02230849211e49f60002590604f2e', | |
33 | 'ext': 'mp4', | |
34 | 'title': 'Stropnický: Policie Vrbětice preventivně nekontrolovala', | |
35 | } | |
36 | }, { | |
37 | 'url': 'http://video.aktualne.cz/dvtv/dvtv-16-12-2014-utok-talibanu-boj-o-kliniku-uprchlici/r~973eb3bc854e11e498be002590604f2e/', | |
38 | 'info_dict': { | |
39 | 'title': 'DVTV 16. 12. 2014: útok Talibanu, boj o kliniku, uprchlíci', | |
40 | 'id': '973eb3bc854e11e498be002590604f2e', | |
41 | }, | |
42 | 'playlist': [{ | |
43 | 'md5': 'da7ca6be4935532241fa9520b3ad91e4', | |
44 | 'info_dict': { | |
45 | 'id': 'b0b40906854d11e4bdad0025900fea04', | |
46 | 'ext': 'mp4', | |
47 | 'title': 'Drtinová Veselovský TV 16. 12. 2014: Témata dne' | |
48 | } | |
49 | }, { | |
50 | 'md5': '5f7652a08b05009c1292317b449ffea2', | |
51 | 'info_dict': { | |
52 | 'id': '420ad9ec854a11e4bdad0025900fea04', | |
53 | 'ext': 'mp4', | |
54 | 'title': 'Školní masakr možná změní boj s Talibanem, říká novinářka' | |
55 | } | |
56 | }, { | |
57 | 'md5': '498eb9dfa97169f409126c617e2a3d64', | |
58 | 'info_dict': { | |
59 | 'id': '95d35580846a11e4b6d20025900fea04', | |
60 | 'ext': 'mp4', | |
61 | 'title': 'Boj o kliniku: Veřejný zájem, nebo právo na majetek?' | |
62 | } | |
63 | }, { | |
64 | 'md5': 'b8dc6b744844032dab6ba3781a7274b9', | |
65 | 'info_dict': { | |
66 | 'id': '6fe14d66853511e4833a0025900fea04', | |
67 | 'ext': 'mp4', | |
68 | 'title': 'Pánek: Odmítání syrských uprchlíků je ostudou české vlády' | |
69 | } | |
70 | }], | |
71 | }, { | |
72 | 'url': 'http://video.aktualne.cz/v-cechach-poprve-zazni-zelenkova-zrestaurovana-mse/r~45b4b00483ec11e4883b002590604f2e/', | |
73 | 'only_matching': True, | |
74 | }] | |
75 | ||
76 | def _parse_video_metadata(self, js, video_id): | |
77 | metadata = self._parse_json(js, video_id, transform_source=js_to_json) | |
78 | ||
79 | formats = [] | |
80 | for video in metadata['sources']: | |
81 | ext = video['type'][6:] | |
82 | formats.append({ | |
83 | 'url': video['file'], | |
84 | 'ext': ext, | |
85 | 'format_id': '%s-%s' % (ext, video['label']), | |
86 | 'height': int(video['label'].rstrip('p')), | |
87 | 'fps': 25, | |
88 | }) | |
89 | ||
90 | self._sort_formats(formats) | |
91 | ||
92 | return { | |
93 | 'id': metadata['mediaid'], | |
94 | 'title': unescapeHTML(metadata['title']), | |
95 | 'thumbnail': self._proto_relative_url(metadata['image'], 'http:'), | |
96 | 'formats': formats | |
97 | } | |
98 | ||
99 | def _real_extract(self, url): | |
100 | video_id = self._match_id(url) | |
101 | ||
102 | webpage = self._download_webpage(url, video_id) | |
103 | ||
104 | # single video | |
105 | item = self._search_regex( | |
106 | r"(?s)embedData[0-9a-f]{32}\['asset'\]\s*=\s*(\{.+?\});", | |
107 | webpage, 'video', default=None, fatal=False) | |
108 | ||
109 | if item: | |
110 | return self._parse_video_metadata(item, video_id) | |
111 | ||
112 | # playlist | |
113 | items = re.findall( | |
114 | r"(?s)BBX\.context\.assets\['[0-9a-f]{32}'\]\.push\(({.+?})\);", | |
115 | webpage) | |
116 | ||
117 | if items: | |
118 | return { | |
119 | '_type': 'playlist', | |
120 | 'id': video_id, | |
121 | 'title': self._og_search_title(webpage), | |
122 | 'entries': [self._parse_video_metadata(i, video_id) for i in items] | |
123 | } | |
124 | ||
125 | raise ExtractorError('Could not find neither video nor playlist') |