]>
Commit | Line | Data |
---|---|---|
dfa6661e | 1 | from .common import InfoExtractor |
2 | from ..utils import ( | |
c6e07cf1 | 3 | ExtractorError, |
94389b22 | 4 | int_or_none, |
c6e07cf1 | 5 | parse_duration, |
6 | traverse_obj, | |
7 | unified_timestamp, | |
8 | url_or_none, | |
dfa6661e | 9 | ) |
10 | ||
11 | ||
12 | class RTVSLOIE(InfoExtractor): | |
13 | IE_NAME = 'rtvslo.si' | |
14 | _VALID_URL = r'''(?x) | |
15 | https?://(?: | |
16 | (?:365|4d)\.rtvslo.si/arhiv/[^/?#&;]+| | |
17 | (?:www\.)?rtvslo\.si/rtv365/arhiv | |
18 | )/(?P<id>\d+)''' | |
19 | _GEO_COUNTRIES = ['SI'] | |
20 | ||
21 | _API_BASE = 'https://api.rtvslo.si/ava/{}/{}?client_id=82013fb3a531d5414f478747c1aca622' | |
22 | SUB_LANGS_MAP = {'Slovenski': 'sl'} | |
23 | ||
24 | _TESTS = [ | |
25 | { | |
26 | 'url': 'https://www.rtvslo.si/rtv365/arhiv/174842550?s=tv', | |
27 | 'info_dict': { | |
28 | 'id': '174842550', | |
94389b22 | 29 | 'ext': 'mp4', |
dfa6661e | 30 | 'release_timestamp': 1643140032, |
31 | 'upload_date': '20220125', | |
32 | 'series': 'Dnevnik', | |
33 | 'thumbnail': 'https://img.rtvcdn.si/_up/ava/ava_misc/show_logos/92/dnevnik_3_wide2.jpg', | |
34 | 'description': 'md5:76a18692757aeb8f0f51221106277dd2', | |
35 | 'timestamp': 1643137046, | |
36 | 'title': 'Dnevnik', | |
37 | 'series_id': '92', | |
38 | 'release_date': '20220125', | |
39 | 'duration': 1789, | |
40 | }, | |
41 | }, { | |
42 | 'url': 'https://365.rtvslo.si/arhiv/utrip/174843754', | |
43 | 'info_dict': { | |
44 | 'id': '174843754', | |
45 | 'ext': 'mp4', | |
46 | 'series_id': '94', | |
47 | 'release_date': '20220129', | |
48 | 'timestamp': 1643484455, | |
49 | 'title': 'Utrip', | |
50 | 'duration': 813, | |
51 | 'thumbnail': 'https://img.rtvcdn.si/_up/ava/ava_misc/show_logos/94/utrip_1_wide2.jpg', | |
52 | 'description': 'md5:77f2892630c7b17bb7a5bb84319020c9', | |
53 | 'release_timestamp': 1643485825, | |
54 | 'upload_date': '20220129', | |
55 | 'series': 'Utrip', | |
56 | }, | |
57 | }, { | |
58 | 'url': 'https://365.rtvslo.si/arhiv/il-giornale-della-sera/174844609', | |
59 | 'info_dict': { | |
60 | 'id': '174844609', | |
61 | 'ext': 'mp3', | |
62 | 'series_id': '106615841', | |
63 | 'title': 'Il giornale della sera', | |
64 | 'duration': 1328, | |
65 | 'series': 'Il giornale della sera', | |
66 | 'timestamp': 1643743800, | |
67 | 'release_timestamp': 1643745424, | |
68 | 'thumbnail': 'https://img.rtvcdn.si/_up/ava/ava_misc/show_logos/il-giornale-della-sera_wide2.jpg', | |
69 | 'upload_date': '20220201', | |
70 | 'tbr': 128000, | |
71 | 'release_date': '20220201', | |
72 | }, | |
94389b22 | 73 | }, { |
74 | 'url': 'https://365.rtvslo.si/arhiv/razred-zase/148350750', | |
75 | 'info_dict': { | |
76 | 'id': '148350750', | |
77 | 'ext': 'mp4', | |
78 | 'title': 'Prvi šolski dan, mozaična oddaja za mlade', | |
79 | 'series': 'Razred zase', | |
80 | 'series_id': '148185730', | |
81 | 'duration': 1481, | |
82 | 'upload_date': '20121019', | |
83 | 'timestamp': 1350672122, | |
84 | 'release_date': '20121019', | |
85 | 'release_timestamp': 1350672122, | |
86 | 'thumbnail': 'https://img.rtvcdn.si/_up/ava/ava_misc/show_logos/148185730/razred_zase_2014_logo_4d_wide2.jpg', | |
87 | }, | |
dfa6661e | 88 | }, { |
89 | 'url': 'https://4d.rtvslo.si/arhiv/dnevnik/174842550', | |
90 | 'only_matching': True | |
91 | } | |
92 | ] | |
93 | ||
94 | def _real_extract(self, url): | |
95 | v_id = self._match_id(url) | |
96 | meta = self._download_json(self._API_BASE.format('getRecordingDrm', v_id), v_id)['response'] | |
97 | ||
98 | thumbs = [{'id': k, 'url': v, 'http_headers': {'Accept': 'image/jpeg'}} | |
99 | for k, v in (meta.get('images') or {}).items()] | |
100 | ||
101 | subs = {} | |
102 | for s in traverse_obj(meta, 'subs', 'subtitles', default=[]): | |
103 | lang = self.SUB_LANGS_MAP.get(s.get('language'), s.get('language') or 'und') | |
104 | subs.setdefault(lang, []).append({ | |
105 | 'url': s.get('file'), | |
106 | 'ext': traverse_obj(s, 'format', expected_type=str.lower), | |
107 | }) | |
108 | ||
109 | jwt = meta.get('jwt') | |
110 | if not jwt: | |
111 | raise ExtractorError('Site did not provide an authentication token, cannot proceed.') | |
112 | ||
113 | media = self._download_json(self._API_BASE.format('getMedia', v_id), v_id, query={'jwt': jwt})['response'] | |
114 | ||
115 | formats = [] | |
94389b22 | 116 | skip_protocols = ['smil', 'f4m', 'dash'] |
dfa6661e | 117 | adaptive_url = traverse_obj(media, ('addaptiveMedia', 'hls_sec'), expected_type=url_or_none) |
118 | if adaptive_url: | |
94389b22 | 119 | formats = self._extract_wowza_formats(adaptive_url, v_id, skip_protocols=skip_protocols) |
dfa6661e | 120 | |
121 | adaptive_url = traverse_obj(media, ('addaptiveMedia_sl', 'hls_sec'), expected_type=url_or_none) | |
122 | if adaptive_url: | |
94389b22 | 123 | for f in self._extract_wowza_formats(adaptive_url, v_id, skip_protocols=skip_protocols): |
dfa6661e | 124 | formats.append({ |
125 | **f, | |
126 | 'format_id': 'sign-' + f['format_id'], | |
127 | 'format_note': 'Sign language interpretation', 'preference': -10, | |
128 | 'language': ( | |
129 | 'slv' if f.get('language') == 'eng' and f.get('acodec') != 'none' | |
130 | else f.get('language')) | |
131 | }) | |
132 | ||
94389b22 | 133 | for mediafile in traverse_obj(media, ('mediaFiles', lambda _, v: url_or_none(v['streams']['https']))): |
134 | formats.append(traverse_obj(mediafile, { | |
135 | 'url': ('streams', 'https'), | |
136 | 'ext': ('mediaType', {str.lower}), | |
137 | 'width': ('width', {int_or_none}), | |
138 | 'height': ('height', {int_or_none}), | |
139 | 'tbr': ('bitrate', {int_or_none}), | |
140 | 'filesize': ('filesize', {int_or_none}), | |
141 | })) | |
142 | ||
143 | for mediafile in traverse_obj(media, ('mediaFiles', lambda _, v: url_or_none(v['streams']['hls_sec']))): | |
144 | formats.extend(self._extract_wowza_formats( | |
145 | mediafile['streams']['hls_sec'], v_id, skip_protocols=skip_protocols)) | |
dfa6661e | 146 | |
147 | if any('intermission.mp4' in x['url'] for x in formats): | |
148 | self.raise_geo_restricted(countries=self._GEO_COUNTRIES, metadata_available=True) | |
149 | if any('dummy_720p.mp4' in x.get('manifest_url', '') for x in formats) and meta.get('stub') == 'error': | |
150 | raise ExtractorError(f'{self.IE_NAME} said: Clip not available', expected=True) | |
151 | ||
dfa6661e | 152 | return { |
153 | 'id': v_id, | |
154 | 'webpage_url': ''.join(traverse_obj(meta, ('canonical', ('domain', 'path')))), | |
155 | 'title': meta.get('title'), | |
156 | 'formats': formats, | |
157 | 'subtitles': subs, | |
158 | 'thumbnails': thumbs, | |
159 | 'description': meta.get('description'), | |
160 | 'timestamp': unified_timestamp(traverse_obj(meta, 'broadcastDate', ('broadcastDates', 0))), | |
161 | 'release_timestamp': unified_timestamp(meta.get('recordingDate')), | |
162 | 'duration': meta.get('duration') or parse_duration(meta.get('length')), | |
163 | 'tags': meta.get('genre'), | |
164 | 'series': meta.get('showName'), | |
165 | 'series_id': meta.get('showId'), | |
166 | } |