]>
Commit | Line | Data |
---|---|---|
e8f2025e PH |
1 | from __future__ import unicode_literals |
2 | ||
df1d7da2 | 3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
df1d7da2 | 6 | |
09dacfa5 | 7 | |
df1d7da2 | 8 | class MDRIE(InfoExtractor): |
e8f2025e | 9 | _VALID_URL = r'^(?P<domain>https?://(?:www\.)?mdr\.de)/(?:.*)/(?P<type>video|audio)(?P<video_id>[^/_]+)(?:_|\.html)' |
5f6a1245 | 10 | |
bd1488ae | 11 | # No tests, MDR regularily deletes its videos |
e8f2025e PH |
12 | _TEST = { |
13 | 'url': 'http://www.mdr.de/fakt/video189002.html', | |
14 | 'only_matching': True, | |
15 | } | |
df1d7da2 | 16 | |
17 | def _real_extract(self, url): | |
df1d7da2 | 18 | m = re.match(self._VALID_URL, url) |
19 | video_id = m.group('video_id') | |
20 | domain = m.group('domain') | |
df1d7da2 | 21 | |
22 | # determine title and media streams from webpage | |
23 | html = self._download_webpage(url, video_id) | |
df1d7da2 | 24 | |
e8f2025e | 25 | title = self._html_search_regex(r'<h[12]>(.*?)</h[12]>', html, 'title') |
09dacfa5 | 26 | xmlurl = self._search_regex( |
e8f2025e | 27 | r'dataURL:\'(/(?:.+)/(?:video|audio)[0-9]+-avCustom.xml)', html, 'XML URL') |
09dacfa5 PH |
28 | |
29 | doc = self._download_xml(domain + xmlurl, video_id) | |
30 | formats = [] | |
31 | for a in doc.findall('./assets/asset'): | |
32 | url_el = a.find('.//progressiveDownloadUrl') | |
33 | if url_el is None: | |
34 | continue | |
35 | abr = int(a.find('bitrateAudio').text) // 1000 | |
36 | media_type = a.find('mediaType').text | |
37 | format = { | |
38 | 'abr': abr, | |
39 | 'filesize': int(a.find('fileSize').text), | |
40 | 'url': url_el.text, | |
41 | } | |
42 | ||
43 | vbr_el = a.find('bitrateVideo') | |
44 | if vbr_el is None: | |
45 | format.update({ | |
46 | 'vcodec': 'none', | |
e8f2025e | 47 | 'format_id': '%s-%d' % (media_type, abr), |
09dacfa5 PH |
48 | }) |
49 | else: | |
50 | vbr = int(vbr_el.text) // 1000 | |
51 | format.update({ | |
52 | 'vbr': vbr, | |
53 | 'width': int(a.find('frameWidth').text), | |
54 | 'height': int(a.find('frameHeight').text), | |
e8f2025e | 55 | 'format_id': '%s-%d' % (media_type, vbr), |
09dacfa5 PH |
56 | }) |
57 | formats.append(format) | |
b874fe2d PH |
58 | self._sort_formats(formats) |
59 | ||
09dacfa5 PH |
60 | return { |
61 | 'id': video_id, | |
62 | 'title': title, | |
63 | 'formats': formats, | |
64 | } |