]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/cbc.py
d8aa31038bfb85f6e5123fe8e7831a2eb22c0c45
[youtubedl] / youtube_dl / extractor / cbc.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..utils import js_to_json
8
9
10 class CBCIE(InfoExtractor):
11 _VALID_URL = r'https?://(?:www\.)?cbc\.ca/(?:[^/]+/)+(?P<id>[^/?#]+)'
12 _TESTS = [{
13 # with mediaId
14 'url': 'http://www.cbc.ca/22minutes/videos/clips-season-23/don-cherry-play-offs',
15 'info_dict': {
16 'id': '2682904050',
17 'ext': 'flv',
18 'title': 'Don Cherry – All-Stars',
19 'description': 'Don Cherry has a bee in his bonnet about AHL player John Scott because that guy’s got heart.',
20 'timestamp': 1454475540,
21 'upload_date': '20160203',
22 },
23 'params': {
24 # rtmp download
25 'skip_download': True,
26 },
27 }, {
28 # with clipId
29 'url': 'http://www.cbc.ca/archives/entry/1978-robin-williams-freestyles-on-90-minutes-live',
30 'info_dict': {
31 'id': '2487345465',
32 'ext': 'flv',
33 'title': 'Robin Williams freestyles on 90 Minutes Live',
34 'description': 'Wacky American comedian Robin Williams shows off his infamous "freestyle" comedic talents while being interviewed on CBC\'s 90 Minutes Live.',
35 'upload_date': '19700101',
36 },
37 'params': {
38 # rtmp download
39 'skip_download': True,
40 },
41 }, {
42 # multiple iframes
43 'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
44 'playlist': [{
45 'info_dict': {
46 'id': '2680832926',
47 'ext': 'flv',
48 'title': 'An Eagle\'s-Eye View Off Burrard Bridge',
49 'description': 'Hercules the eagle flies from Vancouver\'s Burrard Bridge down to a nearby park with a mini-camera strapped to his back.',
50 'upload_date': '19700101',
51 },
52 }, {
53 'info_dict': {
54 'id': '2658915080',
55 'ext': 'flv',
56 'title': 'Fly like an eagle!',
57 'description': 'Eagle equipped with a mini camera flies from the world\'s tallest tower',
58 'upload_date': '19700101',
59 },
60 }],
61 'params': {
62 # rtmp download
63 'skip_download': True,
64 },
65 }]
66
67 @classmethod
68 def suitable(cls, url):
69 return False if CBCPlayerIE.suitable(url) else super(CBCIE, cls).suitable(url)
70
71 def _real_extract(self, url):
72 display_id = self._match_id(url)
73 webpage = self._download_webpage(url, display_id)
74 player_init = self._search_regex(
75 r'CBC\.APP\.Caffeine\.initInstance\(({.+?})\);', webpage, 'player init',
76 default=None)
77 if player_init:
78 player_info = self._parse_json(player_init, display_id, js_to_json)
79 media_id = player_info.get('mediaId')
80 if not media_id:
81 clip_id = player_info['clipId']
82 media_id = self._download_json(
83 'http://feed.theplatform.com/f/h9dtGB/punlNGjMlc1F?fields=id&byContent=byReleases%3DbyId%253D' + clip_id,
84 clip_id)['entries'][0]['id'].split('/')[-1]
85 return self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id)
86 else:
87 entries = [self.url_result('cbcplayer:%s' % media_id, 'CBCPlayer', media_id) for media_id in re.findall(r'<iframe[^>]+src="[^"]+?mediaId=(\d+)"', webpage)]
88 return self.playlist_result(entries)
89
90
91 class CBCPlayerIE(InfoExtractor):
92 _VALID_URL = r'(?:cbcplayer:|https?://(?:www\.)?cbc\.ca/(?:player/play/|i/caffeine/syndicate/\?mediaId=))(?P<id>\d+)'
93 _TEST = {
94 'url': 'http://www.cbc.ca/player/play/2683190193',
95 'info_dict': {
96 'id': '2683190193',
97 'ext': 'flv',
98 'title': 'Gerry Runs a Sweat Shop',
99 'description': 'md5:b457e1c01e8ff408d9d801c1c2cd29b0',
100 'timestamp': 1455067800,
101 'upload_date': '20160210',
102 },
103 'params': {
104 # rtmp download
105 'skip_download': True,
106 },
107 }
108
109 def _real_extract(self, url):
110 video_id = self._match_id(url)
111 return self.url_result(
112 'http://feed.theplatform.com/f/ExhSPC/vms_5akSXx4Ng_Zn?byGuid=%s' % video_id,
113 'ThePlatformFeed', video_id)