]>
Commit | Line | Data |
---|---|---|
1 | from .common import InfoExtractor | |
2 | from ..compat import compat_urlparse | |
3 | from ..utils import ( | |
4 | determine_ext, | |
5 | int_or_none, | |
6 | join_nonempty, | |
7 | parse_duration, | |
8 | parse_iso8601, | |
9 | url_or_none, | |
10 | xpath_text, | |
11 | ) | |
12 | ||
13 | ||
14 | class MDRIE(InfoExtractor): | |
15 | IE_DESC = 'MDR.DE and KiKA' | |
16 | _VALID_URL = r'https?://(?:www\.)?(?:mdr|kika)\.de/(?:.*)/[a-z-]+-?(?P<id>\d+)(?:_.+?)?\.html' | |
17 | ||
18 | _GEO_COUNTRIES = ['DE'] | |
19 | ||
20 | _TESTS = [{ | |
21 | # MDR regularly deletes its videos | |
22 | 'url': 'http://www.mdr.de/fakt/video189002.html', | |
23 | 'only_matching': True, | |
24 | }, { | |
25 | # audio | |
26 | 'url': 'http://www.mdr.de/kultur/audio1312272_zc-15948bad_zs-86171fdd.html', | |
27 | 'md5': '64c4ee50f0a791deb9479cd7bbe9d2fa', | |
28 | 'info_dict': { | |
29 | 'id': '1312272', | |
30 | 'ext': 'mp3', | |
31 | 'title': 'Feuilleton vom 30. Oktober 2015', | |
32 | 'duration': 250, | |
33 | 'uploader': 'MITTELDEUTSCHER RUNDFUNK', | |
34 | }, | |
35 | 'skip': '404 not found', | |
36 | }, { | |
37 | 'url': 'http://www.kika.de/baumhaus/videos/video19636.html', | |
38 | 'md5': '4930515e36b06c111213e80d1e4aad0e', | |
39 | 'info_dict': { | |
40 | 'id': '19636', | |
41 | 'ext': 'mp4', | |
42 | 'title': 'Baumhaus vom 30. Oktober 2015', | |
43 | 'duration': 134, | |
44 | 'uploader': 'KIKA', | |
45 | }, | |
46 | 'skip': '404 not found', | |
47 | }, { | |
48 | 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/videos/video8182.html', | |
49 | 'md5': '5fe9c4dd7d71e3b238f04b8fdd588357', | |
50 | 'info_dict': { | |
51 | 'id': '8182', | |
52 | 'ext': 'mp4', | |
53 | 'title': 'Beutolomäus und der geheime Weihnachtswunsch', | |
54 | 'description': 'md5:b69d32d7b2c55cbe86945ab309d39bbd', | |
55 | 'timestamp': 1482541200, | |
56 | 'upload_date': '20161224', | |
57 | 'duration': 4628, | |
58 | 'uploader': 'KIKA', | |
59 | }, | |
60 | }, { | |
61 | # audio with alternative playerURL pattern | |
62 | 'url': 'http://www.mdr.de/kultur/videos-und-audios/audio-radio/operation-mindfuck-robert-wilson100.html', | |
63 | 'info_dict': { | |
64 | 'id': '100', | |
65 | 'ext': 'mp4', | |
66 | 'title': 'Feature: Operation Mindfuck - Robert Anton Wilson', | |
67 | 'duration': 3239, | |
68 | 'uploader': 'MITTELDEUTSCHER RUNDFUNK', | |
69 | }, | |
70 | }, { | |
71 | # empty bitrateVideo and bitrateAudio | |
72 | 'url': 'https://www.kika.de/filme/sendung128372_zc-572e3f45_zs-1d9fb70e.html', | |
73 | 'info_dict': { | |
74 | 'id': '128372', | |
75 | 'ext': 'mp4', | |
76 | 'title': 'Der kleine Wichtel kehrt zurück', | |
77 | 'description': 'md5:f77fafdff90f7aa1e9dca14f662c052a', | |
78 | 'duration': 4876, | |
79 | 'timestamp': 1607823300, | |
80 | 'upload_date': '20201213', | |
81 | 'uploader': 'ZDF', | |
82 | }, | |
83 | 'params': { | |
84 | 'skip_download': True, | |
85 | }, | |
86 | }, { | |
87 | 'url': 'http://www.kika.de/baumhaus/sendungen/video19636_zc-fea7f8a0_zs-4bf89c60.html', | |
88 | 'only_matching': True, | |
89 | }, { | |
90 | 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/einzelsendung2534.html', | |
91 | 'only_matching': True, | |
92 | }, { | |
93 | 'url': 'http://www.mdr.de/mediathek/mdr-videos/a/video-1334.html', | |
94 | 'only_matching': True, | |
95 | }] | |
96 | ||
97 | def _real_extract(self, url): | |
98 | video_id = self._match_id(url) | |
99 | ||
100 | webpage = self._download_webpage(url, video_id) | |
101 | ||
102 | data_url = self._search_regex( | |
103 | r'(?:dataURL|playerXml(?:["\'])?)\s*:\s*(["\'])(?P<url>.+?-avCustom\.xml)\1', | |
104 | webpage, 'data url', group='url').replace(r'\/', '/') | |
105 | ||
106 | doc = self._download_xml( | |
107 | compat_urlparse.urljoin(url, data_url), video_id) | |
108 | ||
109 | title = xpath_text(doc, ['./title', './broadcast/broadcastName'], 'title', fatal=True) | |
110 | ||
111 | type_ = xpath_text(doc, './type', default=None) | |
112 | ||
113 | formats = [] | |
114 | processed_urls = [] | |
115 | for asset in doc.findall('./assets/asset'): | |
116 | for source in ( | |
117 | 'download', | |
118 | 'progressiveDownload', | |
119 | 'dynamicHttpStreamingRedirector', | |
120 | 'adaptiveHttpStreamingRedirector'): | |
121 | url_el = asset.find('./%sUrl' % source) | |
122 | if url_el is None: | |
123 | continue | |
124 | ||
125 | video_url = url_or_none(url_el.text) | |
126 | if not video_url or video_url in processed_urls: | |
127 | continue | |
128 | ||
129 | processed_urls.append(video_url) | |
130 | ||
131 | ext = determine_ext(video_url) | |
132 | if ext == 'm3u8': | |
133 | formats.extend(self._extract_m3u8_formats( | |
134 | video_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
135 | quality=1, m3u8_id='HLS', fatal=False)) | |
136 | elif ext == 'f4m': | |
137 | formats.extend(self._extract_f4m_formats( | |
138 | video_url + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id, | |
139 | quality=1, f4m_id='HDS', fatal=False)) | |
140 | else: | |
141 | media_type = xpath_text(asset, './mediaType', 'media type', default='MP4') | |
142 | vbr = int_or_none(xpath_text(asset, './bitrateVideo', 'vbr'), 1000) | |
143 | abr = int_or_none(xpath_text(asset, './bitrateAudio', 'abr'), 1000) | |
144 | filesize = int_or_none(xpath_text(asset, './fileSize', 'file size')) | |
145 | ||
146 | f = { | |
147 | 'url': video_url, | |
148 | 'format_id': join_nonempty(media_type, vbr or abr), | |
149 | 'filesize': filesize, | |
150 | 'abr': abr, | |
151 | 'vbr': vbr, | |
152 | } | |
153 | ||
154 | if vbr: | |
155 | f.update({ | |
156 | 'width': int_or_none(xpath_text(asset, './frameWidth', 'width')), | |
157 | 'height': int_or_none(xpath_text(asset, './frameHeight', 'height')), | |
158 | }) | |
159 | ||
160 | if type_ == 'audio': | |
161 | f['vcodec'] = 'none' | |
162 | ||
163 | formats.append(f) | |
164 | ||
165 | description = xpath_text(doc, './broadcast/broadcastDescription', 'description') | |
166 | timestamp = parse_iso8601( | |
167 | xpath_text( | |
168 | doc, [ | |
169 | './broadcast/broadcastDate', | |
170 | './broadcast/broadcastStartDate', | |
171 | './broadcast/broadcastEndDate'], | |
172 | 'timestamp', default=None)) | |
173 | duration = parse_duration(xpath_text(doc, './duration', 'duration')) | |
174 | uploader = xpath_text(doc, './rights', 'uploader') | |
175 | ||
176 | return { | |
177 | 'id': video_id, | |
178 | 'title': title, | |
179 | 'description': description, | |
180 | 'timestamp': timestamp, | |
181 | 'duration': duration, | |
182 | 'uploader': uploader, | |
183 | 'formats': formats, | |
184 | } |