projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| 9c2a17f2 S |
1 | from __future__ import unicode_literals |
| 2 | ||
| 9c2a17f2 S |
3 | |
| 4 | from .common import InfoExtractor | |
| 5 | from ..utils import ( | |
| 6 | extract_attributes, | |
| 7 | int_or_none, | |
| 8 | unified_timestamp, | |
| 9 | ) | |
| 10 | ||
| 11 | ||
| 12 | class PopcornTVIE(InfoExtractor): | |
| 13 | _VALID_URL = r'https?://[^/]+\.popcorntv\.it/guarda/(?P<display_id>[^/]+)/(?P<id>\d+)' | |
| 14 | _TESTS = [{ | |
| 15 | 'url': 'https://animemanga.popcorntv.it/guarda/food-wars-battaglie-culinarie-episodio-01/9183', | |
| 16 | 'md5': '47d65a48d147caf692ab8562fe630b45', | |
| 17 | 'info_dict': { | |
| 18 | 'id': '9183', | |
| 19 | 'display_id': 'food-wars-battaglie-culinarie-episodio-01', | |
| 20 | 'ext': 'mp4', | |
| 21 | 'title': 'Food Wars, Battaglie Culinarie | Episodio 01', | |
| 22 | 'description': 'md5:b8bea378faae4651d3b34c6e112463d0', | |
| 23 | 'thumbnail': r're:^https?://.*\.jpg$', | |
| 24 | 'timestamp': 1497610857, | |
| 25 | 'upload_date': '20170616', | |
| 26 | 'duration': 1440, | |
| 27 | 'view_count': int, | |
| 28 | }, | |
| 29 | }, { | |
| 30 | 'url': 'https://cinema.popcorntv.it/guarda/smash-cut/10433', | |
| 31 | 'only_matching': True, | |
| 32 | }] | |
| 33 | ||
| 34 | def _real_extract(self, url): | |
| 5ad28e7f | 35 | mobj = self._match_valid_url(url) |
| 9c2a17f2 S |
36 | display_id, video_id = mobj.group('display_id', 'id') |
| 37 | ||
| 38 | webpage = self._download_webpage(url, display_id) | |
| 39 | ||
| 40 | m3u8_url = extract_attributes( | |
| 41 | self._search_regex( | |
| 42 | r'(<link[^>]+itemprop=["\'](?:content|embed)Url[^>]*>)', | |
| 43 | webpage, 'content' | |
| 44 | ))['href'] | |
| 45 | ||
| 46 | formats = self._extract_m3u8_formats( | |
| 47 | m3u8_url, display_id, 'mp4', entry_protocol='m3u8_native', | |
| 48 | m3u8_id='hls') | |
| 49 | ||
| 50 | title = self._search_regex( | |
| 51 | r'<h1[^>]+itemprop=["\']name[^>]*>([^<]+)', webpage, | |
| 52 | 'title', default=None) or self._og_search_title(webpage) | |
| 53 | ||
| 54 | description = self._html_search_regex( | |
| 55 | r'(?s)<article[^>]+itemprop=["\']description[^>]*>(.+?)</article>', | |
| 56 | webpage, 'description', fatal=False) | |
| 57 | thumbnail = self._og_search_thumbnail(webpage) | |
| 58 | timestamp = unified_timestamp(self._html_search_meta( | |
| 59 | 'uploadDate', webpage, 'timestamp')) | |
| 9c2a17f2 S |
60 | duration = int_or_none(self._html_search_meta( |
| 61 | 'duration', webpage), invscale=60) | |
| 62 | view_count = int_or_none(self._html_search_meta( | |
| 63 | 'interactionCount', webpage, 'view count')) | |
| 64 | ||
| 65 | return { | |
| 66 | 'id': video_id, | |
| 67 | 'display_id': display_id, | |
| 68 | 'title': title, | |
| 69 | 'description': description, | |
| 70 | 'thumbnail': thumbnail, | |
| 71 | 'timestamp': timestamp, | |
| 72 | 'duration': duration, | |
| 73 | 'view_count': view_count, | |
| 74 | 'formats': formats, | |
| 75 | } |