]>
Commit | Line | Data |
---|---|---|
29f400b9 | 1 | # -*- coding: utf-8 -*- |
24144e3b | 2 | from __future__ import unicode_literals |
5137ebac | 3 | |
29f400b9 TF |
4 | import re |
5 | ||
5137ebac | 6 | from .common import InfoExtractor |
c3a3028f | 7 | |
5137ebac MC |
8 | |
9 | class TvpIE(InfoExtractor): | |
24144e3b | 10 | IE_NAME = 'tvp.pl' |
fb4b030a PH |
11 | _VALID_URL = r'https?://(?:vod|www)\.tvp\.pl/.*/(?P<id>\d+)$' |
12 | ||
13 | _TESTS = [{ | |
14 | 'url': 'http://vod.tvp.pl/filmy-fabularne/filmy-za-darmo/ogniem-i-mieczem/wideo/odc-2/4278035', | |
030aa5d9 | 15 | 'md5': 'cdd98303338b8a7f7abab5cd14092bf2', |
fb4b030a PH |
16 | 'info_dict': { |
17 | 'id': '4278035', | |
18 | 'ext': 'wmv', | |
19 | 'title': 'Ogniem i mieczem, odc. 2', | |
fb4b030a PH |
20 | }, |
21 | }, { | |
22 | 'url': 'http://vod.tvp.pl/seriale/obyczajowe/czas-honoru/sezon-1-1-13/i-seria-odc-13/194536', | |
030aa5d9 | 23 | 'md5': '8aa518c15e5cc32dfe8db400dc921fbb', |
fb4b030a PH |
24 | 'info_dict': { |
25 | 'id': '194536', | |
26 | 'ext': 'mp4', | |
27 | 'title': 'Czas honoru, I seria – odc. 13', | |
fb4b030a PH |
28 | }, |
29 | }, { | |
30 | 'url': 'http://www.tvp.pl/there-can-be-anything-so-i-shortened-it/17916176', | |
030aa5d9 | 31 | 'md5': 'c3b15ed1af288131115ff17a17c19dda', |
fb4b030a PH |
32 | 'info_dict': { |
33 | 'id': '17916176', | |
34 | 'ext': 'mp4', | |
35 | 'title': 'TVP Gorzów pokaże filmy studentów z podroży dookoła świata', | |
36 | }, | |
fb4b030a PH |
37 | }, { |
38 | 'url': 'http://vod.tvp.pl/seriale/obyczajowe/na-sygnale/sezon-2-27-/odc-39/17834272', | |
030aa5d9 | 39 | 'md5': 'c3b15ed1af288131115ff17a17c19dda', |
fb4b030a PH |
40 | 'info_dict': { |
41 | 'id': '17834272', | |
42 | 'ext': 'mp4', | |
43 | 'title': 'Na sygnale, odc. 39', | |
fb4b030a PH |
44 | }, |
45 | }] | |
5137ebac MC |
46 | |
47 | def _real_extract(self, url): | |
fb4b030a | 48 | video_id = self._match_id(url) |
030aa5d9 | 49 | |
29f400b9 TF |
50 | webpage = self._download_webpage( |
51 | 'http://www.tvp.pl/sess/tvplayer.php?object_id=%s' % video_id, video_id) | |
fb4b030a | 52 | |
030aa5d9 S |
53 | title = self._search_regex( |
54 | r'name\s*:\s*([\'"])Title\1\s*,\s*value\s*:\s*\1(?P<title>.+?)\1', | |
55 | webpage, 'title', group='title') | |
56 | series_title = self._search_regex( | |
57 | r'name\s*:\s*([\'"])SeriesTitle\1\s*,\s*value\s*:\s*\1(?P<series>.+?)\1', | |
29f400b9 | 58 | webpage, 'series', group='series', default=None) |
030aa5d9 S |
59 | if series_title: |
60 | title = '%s, %s' % (series_title, title) | |
61 | ||
62 | thumbnail = self._search_regex( | |
63 | r"poster\s*:\s*'([^']+)'", webpage, 'thumbnail', default=None) | |
fb4b030a | 64 | |
29f400b9 TF |
65 | video_url = self._search_regex( |
66 | r'0:{src:([\'"])(?P<url>.*?)\1', webpage, 'formats', group='url', default=None) | |
030aa5d9 | 67 | if not video_url: |
29f400b9 TF |
68 | video_url = self._download_json( |
69 | 'http://www.tvp.pl/pub/stat/videofileinfo?video_id=%s' % video_id, | |
70 | video_id)['video_url'] | |
71 | ||
72 | ext = video_url.rsplit('.', 1)[-1] | |
73 | if ext != 'ism/manifest': | |
74 | if '/' in ext: | |
75 | ext = 'mp4' | |
fb4b030a PH |
76 | formats = [{ |
77 | 'format_id': 'direct', | |
29f400b9 | 78 | 'url': video_url, |
fb4b030a PH |
79 | 'ext': ext, |
80 | }] | |
29f400b9 TF |
81 | else: |
82 | m3u8_url = re.sub('([^/]*)\.ism/manifest', r'\1.ism/\1.m3u8', video_url) | |
83 | formats = self._extract_m3u8_formats(m3u8_url, video_id, 'mp4') | |
fb4b030a PH |
84 | |
85 | self._sort_formats(formats) | |
86 | ||
87 | return { | |
88 | 'id': video_id, | |
89 | 'title': title, | |
030aa5d9 | 90 | 'thumbnail': thumbnail, |
fb4b030a PH |
91 | 'formats': formats, |
92 | } | |
6ce2c678 TF |
93 | |
94 | ||
95 | class TvpSeriesIE(InfoExtractor): | |
96 | IE_NAME = 'tvp.pl:Series' | |
97 | _VALID_URL = r'https?://vod\.tvp\.pl/(?:[^/]+/){2}(?P<id>[^/]+)/?$' | |
98 | ||
fb4b030a PH |
99 | _TESTS = [{ |
100 | 'url': 'http://vod.tvp.pl/filmy-fabularne/filmy-za-darmo/ogniem-i-mieczem', | |
101 | 'info_dict': { | |
102 | 'title': 'Ogniem i mieczem', | |
103 | 'id': '4278026', | |
104 | }, | |
105 | 'playlist_count': 4, | |
106 | }, { | |
107 | 'url': 'http://vod.tvp.pl/audycje/podroze/boso-przez-swiat', | |
108 | 'info_dict': { | |
109 | 'title': 'Boso przez świat', | |
110 | 'id': '9329207', | |
111 | }, | |
112 | 'playlist_count': 86, | |
113 | }] | |
6ce2c678 | 114 | |
6ce2c678 TF |
115 | def _real_extract(self, url): |
116 | display_id = self._match_id(url) | |
2415951e | 117 | webpage = self._download_webpage(url, display_id, tries=5) |
fb4b030a | 118 | |
6ce2c678 | 119 | title = self._html_search_regex( |
2415951e | 120 | r'(?s) id=[\'"]path[\'"]>(?:.*? / ){2}(.*?)</span>', webpage, 'series') |
6ce2c678 | 121 | playlist_id = self._search_regex(r'nodeId:\s*(\d+)', webpage, 'playlist id') |
2415951e | 122 | playlist = self._download_webpage( |
6ce2c678 | 123 | 'http://vod.tvp.pl/vod/seriesAjax?type=series&nodeId=%s&recommend' |
fb4b030a PH |
124 | 'edId=0&sort=&page=0&pageSize=10000' % playlist_id, display_id, tries=5, |
125 | note='Downloading playlist') | |
126 | ||
6ce2c678 TF |
127 | videos_paths = re.findall( |
128 | '(?s)class="shortTitle">.*?href="(/[^"]+)', playlist) | |
129 | entries = [ | |
130 | self.url_result('http://vod.tvp.pl%s' % v_path, ie=TvpIE.ie_key()) | |
131 | for v_path in videos_paths] | |
fb4b030a | 132 | |
6ce2c678 TF |
133 | return { |
134 | '_type': 'playlist', | |
135 | 'id': playlist_id, | |
136 | 'display_id': display_id, | |
137 | 'title': title, | |
138 | 'entries': entries, | |
139 | } |