]>
Commit | Line | Data |
---|---|---|
592882aa JMF |
1 | # encoding: utf-8 |
2 | ||
fbaaad49 JMF |
3 | import re |
4 | import json | |
cfe50f04 | 5 | import xml.etree.ElementTree |
fbaaad49 JMF |
6 | |
7 | from .common import InfoExtractor | |
cfe50f04 JMF |
8 | from ..utils import ( |
9 | compat_urllib_parse, | |
45ff2d51 | 10 | find_xpath_attr, |
6543f0dc | 11 | compat_urlparse, |
eeb165e6 | 12 | compat_str, |
dd5bcdc4 | 13 | compat_urllib_request, |
592882aa JMF |
14 | |
15 | ExtractorError, | |
cfe50f04 | 16 | ) |
fbaaad49 | 17 | |
dd5bcdc4 | 18 | |
fbaaad49 | 19 | class BrightcoveIE(InfoExtractor): |
abb285fb | 20 | _VALID_URL = r'https?://.*brightcove\.com/(services|viewer).*\?(?P<query>.*)' |
cfe50f04 | 21 | _FEDERATED_URL_TEMPLATE = 'http://c.brightcove.com/services/viewer/htmlFederated?%s' |
abb285fb | 22 | _PLAYLIST_URL_TEMPLATE = 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=%s' |
592882aa JMF |
23 | |
24 | _TESTS = [ | |
25 | { | |
4de1994b JMF |
26 | # From http://www.8tv.cat/8aldia/videos/xavier-sala-i-martin-aquesta-tarda-a-8-al-dia/ |
27 | u'url': u'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1654948606001&flashID=myExperience&%40videoPlayer=2371591881001', | |
592882aa | 28 | u'file': u'2371591881001.mp4', |
f52f01b5 | 29 | u'md5': u'8eccab865181d29ec2958f32a6a754f5', |
592882aa JMF |
30 | u'note': u'Test Brightcove downloads and detection in GenericIE', |
31 | u'info_dict': { | |
32 | u'title': u'Xavier Sala i Martín: “Un banc que no presta és un banc zombi que no serveix per a res”', | |
33 | u'uploader': u'8TV', | |
34 | u'description': u'md5:a950cc4285c43e44d763d036710cd9cd', | |
35 | } | |
36 | }, | |
37 | { | |
4de1994b JMF |
38 | # From http://medianetwork.oracle.com/video/player/1785452137001 |
39 | u'url': u'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1217746023001&flashID=myPlayer&%40videoPlayer=1785452137001', | |
592882aa JMF |
40 | u'file': u'1785452137001.flv', |
41 | u'info_dict': { | |
42 | u'title': u'JVMLS 2012: Arrays 2.0 - Opportunities and Challenges', | |
43 | u'description': u'John Rose speaks at the JVM Language Summit, August 1, 2012.', | |
44 | u'uploader': u'Oracle', | |
45 | }, | |
46 | }, | |
fc4a0c2a JMF |
47 | { |
48 | # From http://mashable.com/2013/10/26/thermoelectric-bracelet-lets-you-control-your-body-temperature/ | |
49 | u'url': u'http://c.brightcove.com/services/viewer/federated_f9?&playerID=1265504713001&publisherID=AQ%7E%7E%2CAAABBzUwv1E%7E%2CxP-xFHVUstiMFlNYfvF4G9yFnNaqCw_9&videoID=2750934548001', | |
50 | u'info_dict': { | |
51 | u'id': u'2750934548001', | |
52 | u'ext': u'mp4', | |
53 | u'title': u'This Bracelet Acts as a Personal Thermostat', | |
54 | u'description': u'md5:547b78c64f4112766ccf4e151c20b6a0', | |
55 | u'uploader': u'Mashable', | |
56 | }, | |
57 | }, | |
592882aa | 58 | ] |
cfe50f04 JMF |
59 | |
60 | @classmethod | |
61 | def _build_brighcove_url(cls, object_str): | |
62 | """ | |
63 | Build a Brightcove url from a xml string containing | |
64 | <object class="BrightcoveExperience">{params}</object> | |
65 | """ | |
46e28a84 PH |
66 | |
67 | # Fix up some stupid HTML, see https://github.com/rg3/youtube-dl/issues/1553 | |
68 | object_str = re.sub(r'(<param name="[^"]+" value="[^"]+")>', | |
69 | lambda m: m.group(1) + '/>', object_str) | |
2d0efe70 PH |
70 | # Fix up some stupid XML, see https://github.com/rg3/youtube-dl/issues/1608 |
71 | object_str = object_str.replace(u'<--', u'<!--') | |
46e28a84 | 72 | |
cfe50f04 | 73 | object_doc = xml.etree.ElementTree.fromstring(object_str) |
117adb0f | 74 | assert u'BrightcoveExperience' in object_doc.attrib['class'] |
cfe50f04 | 75 | params = {'flashID': object_doc.attrib['id'], |
5de3ece2 | 76 | 'playerID': find_xpath_attr(object_doc, './param', 'name', 'playerID').attrib['value'], |
cfe50f04 | 77 | } |
36de0a0e JMF |
78 | def find_param(name): |
79 | return find_xpath_attr(object_doc, './param', 'name', name) | |
80 | playerKey = find_param('playerKey') | |
cfe50f04 JMF |
81 | # Not all pages define this value |
82 | if playerKey is not None: | |
83 | params['playerKey'] = playerKey.attrib['value'] | |
36de0a0e JMF |
84 | # The three fields hold the id of the video |
85 | videoPlayer = find_param('@videoPlayer') or find_param('videoId') or find_param('videoID') | |
abb285fb JMF |
86 | if videoPlayer is not None: |
87 | params['@videoPlayer'] = videoPlayer.attrib['value'] | |
36de0a0e | 88 | linkBase = find_param('linkBaseURL') |
dd5bcdc4 JMF |
89 | if linkBase is not None: |
90 | params['linkBaseURL'] = linkBase.attrib['value'] | |
cfe50f04 JMF |
91 | data = compat_urllib_parse.urlencode(params) |
92 | return cls._FEDERATED_URL_TEMPLATE % data | |
fbaaad49 | 93 | |
eeb165e6 JMF |
94 | @classmethod |
95 | def _extract_brightcove_url(cls, webpage): | |
96 | """Try to extract the brightcove url from the wepbage, returns None | |
97 | if it can't be found | |
98 | """ | |
99 | m_brightcove = re.search( | |
100 | r'<object[^>]+?class=([\'"])[^>]*?BrightcoveExperience.*?\1.+?</object>', | |
101 | webpage, re.DOTALL) | |
102 | if m_brightcove is not None: | |
103 | return cls._build_brighcove_url(m_brightcove.group()) | |
104 | else: | |
105 | return None | |
106 | ||
fbaaad49 | 107 | def _real_extract(self, url): |
51040b72 JMF |
108 | # Change the 'videoId' and others field to '@videoPlayer' |
109 | url = re.sub(r'(?<=[?&])(videoI(d|D)|bctid)', '%40videoPlayer', url) | |
110 | # Change bckey (used by bcove.me urls) to playerKey | |
111 | url = re.sub(r'(?<=[?&])bckey', 'playerKey', url) | |
fbaaad49 | 112 | mobj = re.match(self._VALID_URL, url) |
6543f0dc JMF |
113 | query_str = mobj.group('query') |
114 | query = compat_urlparse.parse_qs(query_str) | |
fbaaad49 | 115 | |
6543f0dc JMF |
116 | videoPlayer = query.get('@videoPlayer') |
117 | if videoPlayer: | |
dd5bcdc4 | 118 | return self._get_video_info(videoPlayer[0], query_str, query) |
abb285fb | 119 | else: |
6543f0dc JMF |
120 | player_key = query['playerKey'] |
121 | return self._get_playlist_info(player_key[0]) | |
abb285fb | 122 | |
dd5bcdc4 JMF |
123 | def _get_video_info(self, video_id, query_str, query): |
124 | request_url = self._FEDERATED_URL_TEMPLATE % query_str | |
125 | req = compat_urllib_request.Request(request_url) | |
126 | linkBase = query.get('linkBaseURL') | |
127 | if linkBase is not None: | |
128 | req.add_header('Referer', linkBase[0]) | |
129 | webpage = self._download_webpage(req, video_id) | |
fbaaad49 JMF |
130 | |
131 | self.report_extraction(video_id) | |
132 | info = self._search_regex(r'var experienceJSON = ({.*?});', webpage, 'json') | |
133 | info = json.loads(info)['data'] | |
134 | video_info = info['programmedContent']['videoPlayer']['mediaDTO'] | |
abb285fb JMF |
135 | |
136 | return self._extract_video_info(video_info) | |
137 | ||
138 | def _get_playlist_info(self, player_key): | |
139 | playlist_info = self._download_webpage(self._PLAYLIST_URL_TEMPLATE % player_key, | |
140 | player_key, u'Downloading playlist information') | |
141 | ||
59145479 PH |
142 | json_data = json.loads(playlist_info) |
143 | if 'videoList' not in json_data: | |
144 | raise ExtractorError(u'Empty playlist') | |
145 | playlist_info = json_data['videoList'] | |
abb285fb JMF |
146 | videos = [self._extract_video_info(video_info) for video_info in playlist_info['mediaCollectionDTO']['videoDTOs']] |
147 | ||
148 | return self.playlist_result(videos, playlist_id=playlist_info['id'], | |
149 | playlist_title=playlist_info['mediaCollectionDTO']['displayName']) | |
150 | ||
151 | def _extract_video_info(self, video_info): | |
592882aa | 152 | info = { |
eeb165e6 | 153 | 'id': compat_str(video_info['id']), |
592882aa JMF |
154 | 'title': video_info['displayName'], |
155 | 'description': video_info.get('shortDescription'), | |
156 | 'thumbnail': video_info.get('videoStillURL') or video_info.get('thumbnailURL'), | |
157 | 'uploader': video_info.get('publisherName'), | |
158 | } | |
abb285fb | 159 | |
592882aa JMF |
160 | renditions = video_info.get('renditions') |
161 | if renditions: | |
162 | renditions = sorted(renditions, key=lambda r: r['size']) | |
b0759f0c JMF |
163 | info['formats'] = [{ |
164 | 'url': rend['defaultURL'], | |
165 | 'height': rend.get('frameHeight'), | |
166 | 'width': rend.get('frameWidth'), | |
167 | } for rend in renditions] | |
592882aa JMF |
168 | elif video_info.get('FLVFullLengthURL') is not None: |
169 | info.update({ | |
170 | 'url': video_info['FLVFullLengthURL'], | |
592882aa JMF |
171 | }) |
172 | else: | |
173 | raise ExtractorError(u'Unable to extract video url for %s' % info['id']) | |
174 | return info |