]> jfr.im git - yt-dlp.git/blame - youtube_dl/extractor/ccc.py
[ccc] Extract creator
[yt-dlp.git] / youtube_dl / extractor / ccc.py
CommitLineData
24510bdc 1# coding: utf-8
8f84f571
PH
2from __future__ import unicode_literals
3
8f84f571
PH
4from .common import InfoExtractor
5from ..utils import (
6 int_or_none,
9cf79e8f 7 parse_iso8601,
24510bdc 8 try_get,
8f84f571
PH
9)
10
11
12class CCCIE(InfoExtractor):
13 IE_NAME = 'media.ccc.de'
0d5095fc 14 _VALID_URL = r'https?://(?:www\.)?media\.ccc\.de/v/(?P<id>[^/?#&]+)'
8f84f571 15
0d5095fc
S
16 _TESTS = [{
17 'url': 'https://media.ccc.de/v/30C3_-_5443_-_en_-_saal_g_-_201312281830_-_introduction_to_processor_design_-_byterazor#video',
15da7ce7 18 'md5': '3a1eda8f3a29515d27f5adb967d7e740',
8f84f571 19 'info_dict': {
9cf79e8f 20 'id': '1839',
8f84f571
PH
21 'ext': 'mp4',
22 'title': 'Introduction to Processor Design',
24510bdc 23 'creator': 'byterazor',
9cf79e8f 24 'description': 'md5:df55f6d073d4ceae55aae6f2fd98a0ac',
ec85ded8 25 'thumbnail': r're:^https?://.*\.jpg$',
8499d211 26 'upload_date': '20131228',
9cf79e8f 27 'timestamp': 1388188800,
28 'duration': 3710,
24510bdc 29 'tags': list,
8f84f571 30 }
0d5095fc
S
31 }, {
32 'url': 'https://media.ccc.de/v/32c3-7368-shopshifting#download',
24510bdc
TG
33 'info_dict': {
34 'id': '2835',
35 'ext': 'mp4',
36 'title': 'Shopshifting',
37 'creator': 'Karsten Nohl, Fabian Bräunlein, dexter',
38 'description': 'md5:0fade0535e9dc3076d0cbda4958a18eb',
39 'upload_date': '20151227',
40 'timestamp': 1451249100,
41 'tags': list,
42 }
0d5095fc 43 }]
8f84f571
PH
44
45 def _real_extract(self, url):
9cf79e8f 46 display_id = self._match_id(url)
47 webpage = self._download_webpage(url, display_id)
ec85ded8 48 event_id = self._search_regex(r"data-id='(\d+)'", webpage, 'event id')
9cf79e8f 49 event_data = self._download_json('https://media.ccc.de/public/events/%s' % event_id, event_id)
8f84f571 50
8f84f571 51 formats = []
9cf79e8f 52 for recording in event_data.get('recordings', []):
53 recording_url = recording.get('recording_url')
54 if not recording_url:
55 continue
56 language = recording.get('language')
57 folder = recording.get('folder')
58 format_id = None
59 if language:
60 format_id = language
61 if folder:
62 if language:
63 format_id += '-' + folder
64 else:
65 format_id = folder
66 vcodec = 'h264' if 'h264' in folder else (
67 'none' if folder in ('mp3', 'opus') else None
8f84f571
PH
68 )
69 formats.append({
70 'format_id': format_id,
9cf79e8f 71 'url': recording_url,
72 'width': int_or_none(recording.get('width')),
73 'height': int_or_none(recording.get('height')),
74 'filesize': int_or_none(recording.get('size'), invscale=1024 * 1024),
75 'language': language,
8f84f571 76 'vcodec': vcodec,
8f84f571 77 })
8f84f571
PH
78 self._sort_formats(formats)
79
8f84f571 80 return {
9cf79e8f 81 'id': event_id,
82 'display_id': display_id,
83 'title': event_data['title'],
24510bdc 84 'creator': try_get(event_data, lambda x: ', '.join(x['persons'])),
9cf79e8f 85 'description': event_data.get('description'),
86 'thumbnail': event_data.get('thumb_url'),
87 'timestamp': parse_iso8601(event_data.get('date')),
88 'duration': int_or_none(event_data.get('length')),
89 'tags': event_data.get('tags'),
8f84f571
PH
90 'formats': formats,
91 }
ae8c1356
TK
92
93
94class CCCPlaylistIE(InfoExtractor):
95 IE_NAME = 'media.ccc.de:lists'
96 _VALID_URL = r'https?://(?:www\.)?media\.ccc\.de/c/(?P<id>[^/?#&]+)'
97 _TESTS = [{
98 'url': 'https://media.ccc.de/c/30c3',
99 'info_dict': {
100 'title': '30C3',
101 'id': '30c3',
102 },
103 'playlist_count': 135,
104 }]
105
106 def _real_extract(self, url):
107 acronym = self._match_id(url).lower()
108
109 conf = self._download_json('https://media.ccc.de/public/conferences/' + acronym, acronym)
110
111 return self.playlist_result(
112 [self.url_result(event['frontend_link']) for event in conf['events']],
113 acronym,
114 conf['title'],
115 )