projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| c8868a9d | 1 | # coding: utf-8 |
| 2 | from __future__ import unicode_literals | |
| 3 | ||
| 4 | from .common import InfoExtractor | |
| ac88d231 YCH |
5 | from ..utils import ( |
| 6 | int_or_none, | |
| 7 | unified_strdate, | |
| 8 | ) | |
| 36bb63e0 | 9 | from ..compat import compat_urlparse |
| c8868a9d | 10 | |
| 11 | ||
| 12 | class DWIE(InfoExtractor): | |
| 36bb63e0 | 13 | IE_NAME = 'dw' |
| ac88d231 | 14 | _VALID_URL = r'https?://(?:www\.)?dw\.com/(?:[^/]+/)+(?:av|e)-(?P<id>\d+)' |
| 91d6aafb | 15 | _TESTS = [{ |
| 16 | # video | |
| c8868a9d | 17 | 'url': 'http://www.dw.com/en/intelligent-light/av-19112290', |
| 18 | 'md5': '7372046e1815c5a534b43f3c3c36e6e9', | |
| 19 | 'info_dict': { | |
| 20 | 'id': '19112290', | |
| 21 | 'ext': 'mp4', | |
| 22 | 'title': 'Intelligent light', | |
| 23 | 'description': 'md5:90e00d5881719f2a6a5827cb74985af1', | |
| 24 | 'upload_date': '20160311', | |
| 25 | } | |
| 91d6aafb | 26 | }, { |
| 27 | # audio | |
| 28 | 'url': 'http://www.dw.com/en/worldlink-my-business/av-19111941', | |
| 29 | 'md5': '2814c9a1321c3a51f8a7aeb067a360dd', | |
| 30 | 'info_dict': { | |
| 31 | 'id': '19111941', | |
| 32 | 'ext': 'mp3', | |
| 33 | 'title': 'WorldLink: My business', | |
| 34 | 'description': 'md5:bc9ca6e4e063361e21c920c53af12405', | |
| 35 | 'upload_date': '20160311', | |
| 36 | } | |
| ac88d231 | 37 | }, { |
| 6c0376fe | 38 | # DW documentaries, only last for one or two weeks |
| ac88d231 YCH |
39 | 'url': 'http://www.dw.com/en/documentaries-welcome-to-the-90s-2016-05-21/e-19220158-9798', |
| 40 | 'md5': '56b6214ef463bfb9a3b71aeb886f3cf1', | |
| 41 | 'info_dict': { | |
| 42 | 'id': '19274438', | |
| 43 | 'ext': 'mp4', | |
| 44 | 'title': 'Welcome to the 90s – Hip Hop', | |
| 45 | 'description': 'Welcome to the 90s - The Golden Decade of Hip Hop', | |
| 46 | 'upload_date': '20160521', | |
| 47 | }, | |
| 6c0376fe | 48 | 'skip': 'Video removed', |
| 91d6aafb | 49 | }] |
| c8868a9d | 50 | |
| 51 | def _real_extract(self, url): | |
| 36bb63e0 | 52 | media_id = self._match_id(url) |
| 53 | webpage = self._download_webpage(url, media_id) | |
| c8868a9d | 54 | hidden_inputs = self._hidden_inputs(webpage) |
| 55 | title = hidden_inputs['media_title'] | |
| ac88d231 | 56 | media_id = hidden_inputs.get('media_id') or media_id |
| c8868a9d | 57 | |
| 91d6aafb | 58 | if hidden_inputs.get('player_type') == 'video' and hidden_inputs.get('stream_file') == '1': |
| 59 | formats = self._extract_smil_formats( | |
| 36bb63e0 | 60 | 'http://www.dw.com/smil/v-%s' % media_id, media_id, |
| 91d6aafb | 61 | transform_source=lambda s: s.replace( |
| 62 | 'rtmp://tv-od.dw.de/flash/', | |
| 63 | 'http://tv-download.dw.de/dwtv_video/flv/')) | |
| 19dbaeec | 64 | self._sort_formats(formats) |
| 91d6aafb | 65 | else: |
| 66 | formats = [{'url': hidden_inputs['file_name']}] | |
| 67 | ||
| ac88d231 YCH |
68 | upload_date = hidden_inputs.get('display_date') |
| 69 | if not upload_date: | |
| 70 | upload_date = self._html_search_regex( | |
| 71 | r'<span[^>]+class="date">([0-9.]+)\s*\|', webpage, | |
| 72 | 'upload date', default=None) | |
| 73 | upload_date = unified_strdate(upload_date) | |
| 74 | ||
| c8868a9d | 75 | return { |
| 36bb63e0 | 76 | 'id': media_id, |
| c8868a9d | 77 | 'title': title, |
| 78 | 'description': self._og_search_description(webpage), | |
| 79 | 'thumbnail': hidden_inputs.get('preview_image'), | |
| 80 | 'duration': int_or_none(hidden_inputs.get('file_duration')), | |
| ac88d231 | 81 | 'upload_date': upload_date, |
| c8868a9d | 82 | 'formats': formats, |
| 83 | } | |
| 36bb63e0 | 84 | |
| 85 | ||
| 86 | class DWArticleIE(InfoExtractor): | |
| 87 | IE_NAME = 'dw:article' | |
| 88 | _VALID_URL = r'https?://(?:www\.)?dw\.com/(?:[^/]+/)+a-(?P<id>\d+)' | |
| 89 | _TEST = { | |
| 90 | 'url': 'http://www.dw.com/en/no-hope-limited-options-for-refugees-in-idomeni/a-19111009', | |
| 91 | 'md5': '8ca657f9d068bbef74d6fc38b97fc869', | |
| 92 | 'info_dict': { | |
| 93 | 'id': '19105868', | |
| 94 | 'ext': 'mp4', | |
| 95 | 'title': 'The harsh life of refugees in Idomeni', | |
| 96 | 'description': 'md5:196015cc7e48ebf474db9399420043c7', | |
| 97 | 'upload_date': '20160310', | |
| 98 | } | |
| 99 | } | |
| 100 | ||
| 101 | def _real_extract(self, url): | |
| 102 | article_id = self._match_id(url) | |
| 103 | webpage = self._download_webpage(url, article_id) | |
| 104 | hidden_inputs = self._hidden_inputs(webpage) | |
| 105 | media_id = hidden_inputs['media_id'] | |
| 106 | media_path = self._search_regex(r'href="([^"]+av-%s)"\s+class="overlayLink"' % media_id, webpage, 'media url') | |
| 107 | media_url = compat_urlparse.urljoin(url, media_path) | |
| 108 | return self.url_result(media_url, 'DW', media_id) |