]>
Commit | Line | Data |
---|---|---|
2b1b2d83 | 1 | # coding: utf-8 |
e8f2025e PH |
2 | from __future__ import unicode_literals |
3 | ||
df1d7da2 | 4 | from .common import InfoExtractor |
2b1b2d83 S |
5 | from ..compat import compat_urlparse |
6 | from ..utils import ( | |
7 | determine_ext, | |
8 | int_or_none, | |
9 | parse_duration, | |
10 | parse_iso8601, | |
11 | xpath_text, | |
12 | ) | |
df1d7da2 | 13 | |
09dacfa5 | 14 | |
df1d7da2 | 15 | class MDRIE(InfoExtractor): |
2b1b2d83 S |
16 | IE_DESC = 'MDR.DE and KiKA' |
17 | _VALID_URL = r'https?://(?:www\.)?(?:mdr|kika)\.de/(?:.*)/[a-z]+(?P<id>\d+)(?:_.+?)?\.html' | |
5f6a1245 | 18 | |
2b1b2d83 | 19 | _TESTS = [{ |
dfb1b146 | 20 | # MDR regularly deletes its videos |
e8f2025e PH |
21 | 'url': 'http://www.mdr.de/fakt/video189002.html', |
22 | 'only_matching': True, | |
82b69a5c | 23 | }, { |
8cdb5c84 S |
24 | # audio |
25 | 'url': 'http://www.mdr.de/kultur/audio1312272_zc-15948bad_zs-86171fdd.html', | |
26 | 'md5': '64c4ee50f0a791deb9479cd7bbe9d2fa', | |
27 | 'info_dict': { | |
28 | 'id': '1312272', | |
29 | 'ext': 'mp3', | |
30 | 'title': 'Feuilleton vom 30. Oktober 2015', | |
31 | 'duration': 250, | |
32 | 'uploader': 'MITTELDEUTSCHER RUNDFUNK', | |
33 | }, | |
2b1b2d83 S |
34 | }, { |
35 | 'url': 'http://www.kika.de/baumhaus/videos/video19636.html', | |
36 | 'md5': '4930515e36b06c111213e80d1e4aad0e', | |
37 | 'info_dict': { | |
38 | 'id': '19636', | |
39 | 'ext': 'mp4', | |
40 | 'title': 'Baumhaus vom 30. Oktober 2015', | |
41 | 'duration': 134, | |
42 | 'uploader': 'KIKA', | |
43 | }, | |
44 | }, { | |
45 | 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/videos/video8182.html', | |
46 | 'md5': '5fe9c4dd7d71e3b238f04b8fdd588357', | |
47 | 'info_dict': { | |
48 | 'id': '8182', | |
49 | 'ext': 'mp4', | |
50 | 'title': 'Beutolomäus und der geheime Weihnachtswunsch', | |
51 | 'description': 'md5:b69d32d7b2c55cbe86945ab309d39bbd', | |
52 | 'timestamp': 1419047100, | |
53 | 'upload_date': '20141220', | |
54 | 'duration': 4628, | |
55 | 'uploader': 'KIKA', | |
56 | }, | |
57 | }, { | |
58 | 'url': 'http://www.kika.de/baumhaus/sendungen/video19636_zc-fea7f8a0_zs-4bf89c60.html', | |
59 | 'only_matching': True, | |
60 | }, { | |
61 | 'url': 'http://www.kika.de/sendungen/einzelsendungen/weihnachtsprogramm/einzelsendung2534.html', | |
62 | 'only_matching': True, | |
63 | }] | |
df1d7da2 | 64 | |
65 | def _real_extract(self, url): | |
2b1b2d83 S |
66 | video_id = self._match_id(url) |
67 | ||
68 | webpage = self._download_webpage(url, video_id) | |
df1d7da2 | 69 | |
2b1b2d83 S |
70 | data_url = self._search_regex( |
71 | r'dataURL\s*:\s*(["\'])(?P<url>/.+/(?:video|audio)[0-9]+-avCustom\.xml)\1', | |
72 | webpage, 'data url', group='url') | |
df1d7da2 | 73 | |
2b1b2d83 S |
74 | doc = self._download_xml( |
75 | compat_urlparse.urljoin(url, data_url), video_id) | |
76 | ||
11465da7 | 77 | title = xpath_text(doc, ['./title', './broadcast/broadcastName'], 'title', fatal=True) |
09dacfa5 | 78 | |
09dacfa5 | 79 | formats = [] |
2b1b2d83 S |
80 | processed_urls = [] |
81 | for asset in doc.findall('./assets/asset'): | |
82 | for source in ( | |
83 | 'progressiveDownload', | |
84 | 'dynamicHttpStreamingRedirector', | |
85 | 'adaptiveHttpStreamingRedirector'): | |
86 | url_el = asset.find('./%sUrl' % source) | |
87 | if url_el is None: | |
88 | continue | |
89 | ||
90 | video_url = url_el.text | |
91 | if video_url in processed_urls: | |
92 | continue | |
93 | ||
94 | processed_urls.append(video_url) | |
95 | ||
96 | vbr = int_or_none(xpath_text(asset, './bitrateVideo', 'vbr'), 1000) | |
97 | abr = int_or_none(xpath_text(asset, './bitrateAudio', 'abr'), 1000) | |
98 | ||
2b1b2d83 S |
99 | ext = determine_ext(url_el.text) |
100 | if ext == 'm3u8': | |
101 | url_formats = self._extract_m3u8_formats( | |
102 | video_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
103 | preference=0, m3u8_id='HLS', fatal=False) | |
104 | elif ext == 'f4m': | |
105 | url_formats = self._extract_f4m_formats( | |
106 | video_url + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id, | |
107 | preference=0, f4m_id='HDS', fatal=False) | |
108 | else: | |
109 | media_type = xpath_text(asset, './mediaType', 'media type', default='MP4') | |
110 | vbr = int_or_none(xpath_text(asset, './bitrateVideo', 'vbr'), 1000) | |
111 | abr = int_or_none(xpath_text(asset, './bitrateAudio', 'abr'), 1000) | |
112 | filesize = int_or_none(xpath_text(asset, './fileSize', 'file size')) | |
113 | ||
114 | f = { | |
115 | 'url': video_url, | |
116 | 'format_id': '%s-%d' % (media_type, vbr or abr), | |
117 | 'filesize': filesize, | |
118 | 'abr': abr, | |
119 | 'preference': 1, | |
120 | } | |
121 | ||
122 | if vbr: | |
123 | width = int_or_none(xpath_text(asset, './frameWidth', 'width')) | |
124 | height = int_or_none(xpath_text(asset, './frameHeight', 'height')) | |
125 | f.update({ | |
126 | 'vbr': vbr, | |
127 | 'width': width, | |
128 | 'height': height, | |
129 | }) | |
130 | ||
76f0c50d S |
131 | url_formats = [f] |
132 | ||
133 | if not url_formats: | |
134 | continue | |
2b1b2d83 S |
135 | |
136 | if not vbr: | |
137 | for f in url_formats: | |
138 | abr = f.get('tbr') or abr | |
139 | if 'tbr' in f: | |
140 | del f['tbr'] | |
141 | f.update({ | |
142 | 'abr': abr, | |
143 | 'vcodec': 'none', | |
144 | }) | |
145 | ||
76f0c50d S |
146 | formats.extend(url_formats) |
147 | ||
b874fe2d PH |
148 | self._sort_formats(formats) |
149 | ||
2b1b2d83 S |
150 | description = xpath_text(doc, './broadcast/broadcastDescription', 'description') |
151 | timestamp = parse_iso8601( | |
11465da7 S |
152 | xpath_text( |
153 | doc, [ | |
154 | './broadcast/broadcastDate', | |
155 | './broadcast/broadcastStartDate', | |
156 | './broadcast/broadcastEndDate'], | |
157 | 'timestamp', default=None)) | |
2b1b2d83 S |
158 | duration = parse_duration(xpath_text(doc, './duration', 'duration')) |
159 | uploader = xpath_text(doc, './rights', 'uploader') | |
160 | ||
09dacfa5 PH |
161 | return { |
162 | 'id': video_id, | |
163 | 'title': title, | |
2b1b2d83 S |
164 | 'description': description, |
165 | 'timestamp': timestamp, | |
166 | 'duration': duration, | |
167 | 'uploader': uploader, | |
09dacfa5 PH |
168 | 'formats': formats, |
169 | } |