]>
Commit | Line | Data |
---|---|---|
6f8cb242 | 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 |
6e3c2047 RA |
7 | from ..utils import ( |
8 | determine_ext, | |
9 | clean_html, | |
10 | get_element_by_attribute, | |
11 | ExtractorError, | |
12 | ) | |
c3a3028f | 13 | |
5137ebac | 14 | |
6f8cb242 S |
15 | class TVPIE(InfoExtractor): |
16 | IE_NAME = 'tvp' | |
17 | IE_DESC = 'Telewizja Polska' | |
4e599194 | 18 | _VALID_URL = r'https?://[^/]+\.tvp\.(?:pl|info)/(?:video/(?:[^,\s]*,)*|(?:(?!\d+/)[^/]+/)*)(?P<id>\d+)' |
fb4b030a PH |
19 | |
20 | _TESTS = [{ | |
4e599194 | 21 | 'url': 'https://vod.tvp.pl/video/czas-honoru,i-seria-odc-13,194536', |
3c964737 | 22 | 'md5': 'a21eb0aa862f25414430f15fdfb9e76c', |
fb4b030a PH |
23 | 'info_dict': { |
24 | 'id': '194536', | |
25 | 'ext': 'mp4', | |
3c964737 S |
26 | 'title': 'Czas honoru, odc. 13 – Władek', |
27 | 'description': 'md5:437f48b93558370b031740546b696e24', | |
fb4b030a PH |
28 | }, |
29 | }, { | |
30 | 'url': 'http://www.tvp.pl/there-can-be-anything-so-i-shortened-it/17916176', | |
6e3c2047 | 31 | 'md5': 'b0005b542e5b4de643a9690326ab1257', |
fb4b030a PH |
32 | 'info_dict': { |
33 | 'id': '17916176', | |
34 | 'ext': 'mp4', | |
35 | 'title': 'TVP Gorzów pokaże filmy studentów z podroży dookoła świata', | |
fdd0b8f8 RA |
36 | 'description': 'TVP Gorzów pokaże filmy studentów z podroży dookoła świata', |
37 | }, | |
38 | }, { | |
39 | # page id is not the same as video id(#7799) | |
4e599194 RB |
40 | 'url': 'https://wiadomosci.tvp.pl/33908820/28092017-1930', |
41 | 'md5': '84cd3c8aec4840046e5ab712416b73d0', | |
fdd0b8f8 | 42 | 'info_dict': { |
4e599194 | 43 | 'id': '33908820', |
fdd0b8f8 | 44 | 'ext': 'mp4', |
4e599194 RB |
45 | 'title': 'Wiadomości, 28.09.2017, 19:30', |
46 | 'description': 'Wydanie główne codziennego serwisu informacyjnego.' | |
fb4b030a | 47 | }, |
3c964737 | 48 | 'skip': 'HTTP Error 404: Not Found', |
fb4b030a PH |
49 | }, { |
50 | 'url': 'http://vod.tvp.pl/seriale/obyczajowe/na-sygnale/sezon-2-27-/odc-39/17834272', | |
6f8cb242 S |
51 | 'only_matching': True, |
52 | }, { | |
53 | 'url': 'http://wiadomosci.tvp.pl/25169746/24052016-1200', | |
54 | 'only_matching': True, | |
55 | }, { | |
56 | 'url': 'http://krakow.tvp.pl/25511623/25lecie-mck-wyjatkowe-miejsce-na-mapie-krakowa', | |
57 | 'only_matching': True, | |
58 | }, { | |
59 | 'url': 'http://teleexpress.tvp.pl/25522307/wierni-wzieli-udzial-w-procesjach', | |
60 | 'only_matching': True, | |
61 | }, { | |
62 | 'url': 'http://sport.tvp.pl/25522165/krychowiak-uspokaja-w-sprawie-kontuzji-dwa-tygodnie-to-maksimum', | |
63 | 'only_matching': True, | |
64 | }, { | |
65 | 'url': 'http://www.tvp.info/25511919/trwa-rewolucja-wladza-zdecydowala-sie-na-pogwalcenie-konstytucji', | |
66 | 'only_matching': True, | |
fb4b030a | 67 | }] |
5137ebac | 68 | |
fdd0b8f8 RA |
69 | def _real_extract(self, url): |
70 | page_id = self._match_id(url) | |
71 | webpage = self._download_webpage(url, page_id) | |
72 | video_id = self._search_regex([ | |
73 | r'<iframe[^>]+src="[^"]*?object_id=(\d+)', | |
3d8d44c7 RA |
74 | r"object_id\s*:\s*'(\d+)'", |
75 | r'data-video-id="(\d+)"'], webpage, 'video id', default=page_id) | |
fdd0b8f8 RA |
76 | return { |
77 | '_type': 'url_transparent', | |
78 | 'url': 'tvp:' + video_id, | |
3c964737 S |
79 | 'description': self._og_search_description( |
80 | webpage, default=None) or self._html_search_meta( | |
81 | 'description', webpage, default=None), | |
82 | 'thumbnail': self._og_search_thumbnail(webpage, default=None), | |
fdd0b8f8 RA |
83 | 'ie_key': 'TVPEmbed', |
84 | } | |
85 | ||
86 | ||
87 | class TVPEmbedIE(InfoExtractor): | |
88 | IE_NAME = 'tvp:embed' | |
89 | IE_DESC = 'Telewizja Polska' | |
90 | _VALID_URL = r'(?:tvp:|https?://[^/]+\.tvp\.(?:pl|info)/sess/tvplayer\.php\?.*?object_id=)(?P<id>\d+)' | |
91 | ||
92 | _TESTS = [{ | |
3c964737 S |
93 | 'url': 'tvp:194536', |
94 | 'md5': 'a21eb0aa862f25414430f15fdfb9e76c', | |
95 | 'info_dict': { | |
96 | 'id': '194536', | |
97 | 'ext': 'mp4', | |
98 | 'title': 'Czas honoru, odc. 13 – Władek', | |
99 | }, | |
100 | }, { | |
34568dc2 | 101 | # not available |
fdd0b8f8 RA |
102 | 'url': 'http://www.tvp.pl/sess/tvplayer.php?object_id=22670268', |
103 | 'md5': '8c9cd59d16edabf39331f93bf8a766c7', | |
104 | 'info_dict': { | |
105 | 'id': '22670268', | |
106 | 'ext': 'mp4', | |
107 | 'title': 'Panorama, 07.12.2015, 15:40', | |
108 | }, | |
34568dc2 | 109 | 'skip': 'Transmisja została zakończona lub materiał niedostępny', |
fdd0b8f8 RA |
110 | }, { |
111 | 'url': 'tvp:22670268', | |
112 | 'only_matching': True, | |
113 | }] | |
114 | ||
5137ebac | 115 | def _real_extract(self, url): |
fb4b030a | 116 | video_id = self._match_id(url) |
030aa5d9 | 117 | |
29f400b9 TF |
118 | webpage = self._download_webpage( |
119 | 'http://www.tvp.pl/sess/tvplayer.php?object_id=%s' % video_id, video_id) | |
fb4b030a | 120 | |
34568dc2 S |
121 | error = self._html_search_regex( |
122 | r'(?s)<p[^>]+\bclass=["\']notAvailable__text["\'][^>]*>(.+?)</p>', | |
123 | webpage, 'error', default=None) or clean_html( | |
124 | get_element_by_attribute('class', 'msg error', webpage)) | |
125 | if error: | |
6e3c2047 | 126 | raise ExtractorError('%s said: %s' % ( |
34568dc2 | 127 | self.IE_NAME, clean_html(error)), expected=True) |
6e3c2047 | 128 | |
030aa5d9 S |
129 | title = self._search_regex( |
130 | r'name\s*:\s*([\'"])Title\1\s*,\s*value\s*:\s*\1(?P<title>.+?)\1', | |
131 | webpage, 'title', group='title') | |
132 | series_title = self._search_regex( | |
133 | r'name\s*:\s*([\'"])SeriesTitle\1\s*,\s*value\s*:\s*\1(?P<series>.+?)\1', | |
29f400b9 | 134 | webpage, 'series', group='series', default=None) |
030aa5d9 S |
135 | if series_title: |
136 | title = '%s, %s' % (series_title, title) | |
137 | ||
138 | thumbnail = self._search_regex( | |
139 | r"poster\s*:\s*'([^']+)'", webpage, 'thumbnail', default=None) | |
fb4b030a | 140 | |
29f400b9 | 141 | video_url = self._search_regex( |
6e3c2047 RA |
142 | r'0:{src:([\'"])(?P<url>.*?)\1', webpage, |
143 | 'formats', group='url', default=None) | |
144 | if not video_url or 'material_niedostepny.mp4' in video_url: | |
29f400b9 TF |
145 | video_url = self._download_json( |
146 | 'http://www.tvp.pl/pub/stat/videofileinfo?video_id=%s' % video_id, | |
147 | video_id)['video_url'] | |
148 | ||
6e3c2047 RA |
149 | formats = [] |
150 | video_url_base = self._search_regex( | |
151 | r'(https?://.+?/video)(?:\.(?:ism|f4m|m3u8)|-\d+\.mp4)', | |
152 | video_url, 'video base url', default=None) | |
153 | if video_url_base: | |
9513c1eb S |
154 | # TODO: <Group> found instead of <AdaptationSet> in MPD manifest. |
155 | # It's not mentioned in MPEG-DASH standard. Figure that out. | |
6e3c2047 RA |
156 | # formats.extend(self._extract_mpd_formats( |
157 | # video_url_base + '.ism/video.mpd', | |
158 | # video_id, mpd_id='dash', fatal=False)) | |
639e3b5c RA |
159 | formats.extend(self._extract_ism_formats( |
160 | video_url_base + '.ism/Manifest', | |
161 | video_id, 'mss', fatal=False)) | |
6e3c2047 RA |
162 | formats.extend(self._extract_f4m_formats( |
163 | video_url_base + '.ism/video.f4m', | |
164 | video_id, f4m_id='hds', fatal=False)) | |
165 | m3u8_formats = self._extract_m3u8_formats( | |
166 | video_url_base + '.ism/video.m3u8', video_id, | |
167 | 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False) | |
168 | self._sort_formats(m3u8_formats) | |
169 | m3u8_formats = list(filter( | |
ff99fe52 | 170 | lambda f: f.get('vcodec') != 'none', m3u8_formats)) |
6e3c2047 RA |
171 | formats.extend(m3u8_formats) |
172 | for i, m3u8_format in enumerate(m3u8_formats, 2): | |
173 | http_url = '%s-%d.mp4' % (video_url_base, i) | |
174 | if self._is_valid_url(http_url, video_id): | |
175 | f = m3u8_format.copy() | |
176 | f.update({ | |
177 | 'url': http_url, | |
178 | 'format_id': f['format_id'].replace('hls', 'http'), | |
179 | 'protocol': 'http', | |
180 | }) | |
181 | formats.append(f) | |
182 | else: | |
fb4b030a PH |
183 | formats = [{ |
184 | 'format_id': 'direct', | |
29f400b9 | 185 | 'url': video_url, |
6e3c2047 | 186 | 'ext': determine_ext(video_url, 'mp4'), |
fb4b030a | 187 | }] |
fb4b030a PH |
188 | |
189 | self._sort_formats(formats) | |
190 | ||
191 | return { | |
192 | 'id': video_id, | |
193 | 'title': title, | |
030aa5d9 | 194 | 'thumbnail': thumbnail, |
fb4b030a PH |
195 | 'formats': formats, |
196 | } | |
6ce2c678 TF |
197 | |
198 | ||
6f8cb242 S |
199 | class TVPSeriesIE(InfoExtractor): |
200 | IE_NAME = 'tvp:series' | |
6ce2c678 TF |
201 | _VALID_URL = r'https?://vod\.tvp\.pl/(?:[^/]+/){2}(?P<id>[^/]+)/?$' |
202 | ||
fb4b030a PH |
203 | _TESTS = [{ |
204 | 'url': 'http://vod.tvp.pl/filmy-fabularne/filmy-za-darmo/ogniem-i-mieczem', | |
205 | 'info_dict': { | |
206 | 'title': 'Ogniem i mieczem', | |
207 | 'id': '4278026', | |
208 | }, | |
209 | 'playlist_count': 4, | |
210 | }, { | |
211 | 'url': 'http://vod.tvp.pl/audycje/podroze/boso-przez-swiat', | |
212 | 'info_dict': { | |
213 | 'title': 'Boso przez świat', | |
214 | 'id': '9329207', | |
215 | }, | |
216 | 'playlist_count': 86, | |
217 | }] | |
6ce2c678 | 218 | |
6ce2c678 TF |
219 | def _real_extract(self, url): |
220 | display_id = self._match_id(url) | |
2415951e | 221 | webpage = self._download_webpage(url, display_id, tries=5) |
fb4b030a | 222 | |
6ce2c678 | 223 | title = self._html_search_regex( |
2415951e | 224 | r'(?s) id=[\'"]path[\'"]>(?:.*? / ){2}(.*?)</span>', webpage, 'series') |
6ce2c678 | 225 | playlist_id = self._search_regex(r'nodeId:\s*(\d+)', webpage, 'playlist id') |
2415951e | 226 | playlist = self._download_webpage( |
6ce2c678 | 227 | 'http://vod.tvp.pl/vod/seriesAjax?type=series&nodeId=%s&recommend' |
fb4b030a PH |
228 | 'edId=0&sort=&page=0&pageSize=10000' % playlist_id, display_id, tries=5, |
229 | note='Downloading playlist') | |
230 | ||
6ce2c678 TF |
231 | videos_paths = re.findall( |
232 | '(?s)class="shortTitle">.*?href="(/[^"]+)', playlist) | |
233 | entries = [ | |
6f8cb242 | 234 | self.url_result('http://vod.tvp.pl%s' % v_path, ie=TVPIE.ie_key()) |
6ce2c678 | 235 | for v_path in videos_paths] |
fb4b030a | 236 | |
6ce2c678 TF |
237 | return { |
238 | '_type': 'playlist', | |
239 | 'id': playlist_id, | |
240 | 'display_id': display_id, | |
241 | 'title': title, | |
242 | 'entries': entries, | |
243 | } |