]>
Commit | Line | Data |
---|---|---|
dcdb292f | 1 | # coding: utf-8 |
526b276f | 2 | from __future__ import unicode_literals |
63da13e8 JMF |
3 | |
4 | from .common import InfoExtractor | |
ecbccea7 | 5 | from ..utils import ( |
6 | xpath_element, | |
7 | xpath_text, | |
8 | int_or_none, | |
9 | ) | |
63da13e8 JMF |
10 | |
11 | ||
12 | class FazIE(InfoExtractor): | |
526b276f | 13 | IE_NAME = 'faz.net' |
255f5694 | 14 | _VALID_URL = r'https?://(?:www\.)?faz\.net/(?:[^/]+/)*.*?-(?P<id>\d+)\.html' |
63da13e8 | 15 | |
255f5694 | 16 | _TESTS = [{ |
526b276f PH |
17 | 'url': 'http://www.faz.net/multimedia/videos/stockholm-chemie-nobelpreis-fuer-drei-amerikanische-forscher-12610585.html', |
18 | 'info_dict': { | |
19 | 'id': '12610585', | |
20 | 'ext': 'mp4', | |
21 | 'title': 'Stockholm: Chemie-Nobelpreis für drei amerikanische Forscher', | |
22 | 'description': 'md5:1453fbf9a0d041d985a47306192ea253', | |
63da13e8 | 23 | }, |
255f5694 S |
24 | }, { |
25 | 'url': 'http://www.faz.net/aktuell/politik/berlin-gabriel-besteht-zerreissprobe-ueber-datenspeicherung-13659345.html', | |
26 | 'only_matching': True, | |
27 | }, { | |
28 | 'url': 'http://www.faz.net/berlin-gabriel-besteht-zerreissprobe-ueber-datenspeicherung-13659345.html', | |
29 | 'only_matching': True, | |
30 | }, { | |
31 | 'url': 'http://www.faz.net/-13659345.html', | |
32 | 'only_matching': True, | |
33 | }, { | |
34 | 'url': 'http://www.faz.net/aktuell/politik/-13659345.html', | |
35 | 'only_matching': True, | |
36 | }, { | |
37 | 'url': 'http://www.faz.net/foobarblafasel-13659345.html', | |
38 | 'only_matching': True, | |
39 | }] | |
63da13e8 JMF |
40 | |
41 | def _real_extract(self, url): | |
526b276f PH |
42 | video_id = self._match_id(url) |
43 | ||
63da13e8 | 44 | webpage = self._download_webpage(url, video_id) |
ecbccea7 | 45 | description = self._og_search_description(webpage) |
526b276f | 46 | config_xml_url = self._search_regex( |
ecbccea7 | 47 | r'videoXMLURL\s*=\s*"([^"]+)', webpage, 'config xml url') |
526b276f PH |
48 | config = self._download_xml( |
49 | config_xml_url, video_id, 'Downloading config xml') | |
63da13e8 | 50 | |
ecbccea7 | 51 | encodings = xpath_element(config, 'ENCODINGS', 'encodings', True) |
63da13e8 | 52 | formats = [] |
526b276f | 53 | for pref, code in enumerate(['LOW', 'HIGH', 'HQ']): |
ecbccea7 | 54 | encoding = xpath_element(encodings, code) |
9c86d509 | 55 | if encoding is not None: |
ecbccea7 | 56 | encoding_url = xpath_text(encoding, 'FILENAME') |
57 | if encoding_url: | |
58 | formats.append({ | |
59 | 'url': encoding_url, | |
60 | 'format_id': code.lower(), | |
61 | 'quality': pref, | |
62 | 'tbr': int_or_none(xpath_text(encoding, 'AVERAGEBITRATE')), | |
63 | }) | |
526b276f | 64 | self._sort_formats(formats) |
63da13e8 | 65 | |
fb7abb31 | 66 | return { |
63da13e8 JMF |
67 | 'id': video_id, |
68 | 'title': self._og_search_title(webpage), | |
69 | 'formats': formats, | |
ecbccea7 | 70 | 'description': description.strip() if description else None, |
71 | 'thumbnail': xpath_text(config, 'STILL/STILL_BIG'), | |
72 | 'duration': int_or_none(xpath_text(config, 'DURATION')), | |
63da13e8 | 73 | } |