compat_urllib_parse_urlparse,
compat_urllib_request,
compat_urlparse,
+ compat_xml_parse_error,
)
from ..utils import (
determine_ext,
fix_xml_ampersands,
unescapeHTML,
unsmuggle_url,
+ js_to_json,
+ int_or_none,
+ parse_iso8601,
+ extract_attributes,
)
'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=3550052898001&playerKey=AQ%7E%7E%2CAAABmA9XpXk%7E%2C-Kp7jNgisre1fG5OdqpAFUTcs0lP_ZoL',
'info_dict': {
'title': 'Sealife',
+ 'id': '3550319591001',
},
'playlist_mincount': 7,
},
object_str = re.sub(r'(<object[^>]*)(xmlns=".*?")', r'\1', object_str)
object_str = fix_xml_ampersands(object_str)
- object_doc = xml.etree.ElementTree.fromstring(object_str.encode('utf-8'))
+ try:
+ object_doc = xml.etree.ElementTree.fromstring(object_str.encode('utf-8'))
+ except compat_xml_parse_error:
+ return
fv_el = find_xpath_attr(object_doc, './param', 'name', 'flashVars')
if fv_el is not None:
linkBase = find_param('linkBaseURL')
if linkBase is not None:
params['linkBaseURL'] = linkBase
+ return cls._make_brightcove_url(params)
+
+ @classmethod
+ def _build_brighcove_url_from_js(cls, object_js):
+ # The layout of JS is as follows:
+ # customBC.createVideo = function (width, height, playerID, playerKey, videoPlayer, VideoRandomID) {
+ # // build Brightcove <object /> XML
+ # }
+ m = re.search(
+ r'''(?x)customBC.\createVideo\(
+ .*? # skipping width and height
+ ["\'](?P<playerID>\d+)["\']\s*,\s* # playerID
+ ["\'](?P<playerKey>AQ[^"\']{48})[^"\']*["\']\s*,\s* # playerKey begins with AQ and is 50 characters
+ # in length, however it's appended to itself
+ # in places, so truncate
+ ["\'](?P<videoID>\d+)["\'] # @videoPlayer
+ ''', object_js)
+ if m:
+ return cls._make_brightcove_url(m.groupdict())
+
+ @classmethod
+ def _make_brightcove_url(cls, params):
data = compat_urllib_parse.urlencode(params)
return cls._FEDERATED_URL_TEMPLATE % data
"""Return a list of all Brightcove URLs from the webpage """
url_m = re.search(
- r'<meta\s+property="og:video"\s+content="(https?://(?:secure|c)\.brightcove.com/[^"]+)"',
+ r'<meta\s+property=[\'"]og:video[\'"]\s+content=[\'"](https?://(?:secure|c)\.brightcove.com/[^\'"]+)[\'"]',
webpage)
if url_m:
url = unescapeHTML(url_m.group(1))
(?:
[^>]+?class=[\'"][^>]*?BrightcoveExperience.*?[\'"] |
[^>]*?>\s*<param\s+name="movie"\s+value="https?://[^/]*brightcove\.com/
- ).+?</object>''',
+ ).+?>\s*</object>''',
webpage)
- return [cls._build_brighcove_url(m) for m in matches]
+ if matches:
+ return list(filter(None, [cls._build_brighcove_url(m) for m in matches]))
+
+ return list(filter(None, [
+ cls._build_brighcove_url_from_js(custom_bc)
+ for custom_bc in re.findall(r'(customBC\.createVideo\(.+?\);)', webpage)]))
def _real_extract(self, url):
url, smuggled_data = unsmuggle_url(url, {})
playlist_info = json_data['videoList']
videos = [self._extract_video_info(video_info) for video_info in playlist_info['mediaCollectionDTO']['videoDTOs']]
- return self.playlist_result(videos, playlist_id=playlist_info['id'],
+ return self.playlist_result(videos, playlist_id='%s' % playlist_info['id'],
playlist_title=playlist_info['mediaCollectionDTO']['displayName'])
def _extract_video_info(self, video_info):
if 'url' not in info and not info.get('formats'):
raise ExtractorError('Unable to extract video url for %s' % info['id'])
return info
+
+
+class BrightcoveInPageEmbedIE(InfoExtractor):
+ _VALID_URL = r'https?://players\.brightcove\.net/(?P<account_id>\d+)/([a-z0-9-]+)_([a-z]+)/index.html?.*videoId=(?P<video_id>\d+)'
+ TEST = {
+ 'url': 'http://players.brightcove.net/929656772001/e41d32dc-ec74-459e-a845-6c69f7b724ea_default/index.html?videoId=4463358922001',
+ 'info_dict': {
+ 'id': '4463358922001',
+ 'ext': 'flv',
+ 'title': 'Meet the man behind Popcorn Time',
+ 'description': 'md5:a950cc4285c43e44d763d036710cd9cd',
+ 'duration': 165768,
+ }
+ }
+
+ @staticmethod
+ def _extract_url(webpage):
+ video_attributes = re.search(r'(?s)<video([^>]*)>.*?</(?:video|audio)>', webpage)
+ if video_attributes:
+ video_attributes = extract_attributes(video_attributes.group(), r'(?s)\s*data-(account|video-id|playlist-id|policy-key|player|embed)\s*=\s*["\']([^"\']+)["\']')
+ account_id = video_attributes.get('account')
+ player_id = video_attributes.get('player')
+ embed = video_attributes.get('embed')
+ video_id = video_attributes.get('video-id')
+ if account_id and player_id and embed and video_id:
+ return 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s' % (account_id, player_id, embed, video_id)
+ return None
+
+ def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ account_id, player_id, embed, video_id = mobj.groups()
+
+ webpage = self._download_webpage('http://players.brightcove.net/%s/%s_%s/index.min.js' % (account_id, player_id, embed), video_id)
+
+ catalog = self._parse_json(
+ js_to_json(
+ self._search_regex(
+ r'catalog\(({[^}]+})\);',
+ webpage,
+ 'catalog'
+ )
+ ),
+ video_id
+ )
+ policy_key = catalog['policyKey']
+
+ req = compat_urllib_request.Request(
+ 'https://edge.api.brightcove.com/playback/v1/accounts/%s/videos/%s' % (account_id, video_id),
+ headers={'Accept': 'application/json;pk=%s' % policy_key})
+ json_data = self._download_json(req, video_id)
+
+ title = json_data['name']
+ description = json_data.get('description')
+ thumbnail = json_data.get('name')
+ timestamp = parse_iso8601(json_data.get('published_at'))
+ duration = int_or_none(json_data.get('duration'))
+
+ formats = []
+ for source in json_data.get('sources'):
+ source_type = source.get('type')
+ if source_type == 'application/x-mpegURL':
+ formats.extend(self._extract_m3u8_formats(source.get('src'), video_id))
+ else:
+ src = source.get('src')
+ if src:
+ formats.append({
+ 'url': src,
+ 'abr': source.get('avg_bitrate'),
+ 'width': int_or_none(source.get('width')),
+ 'height': int_or_none(source.get('height')),
+ 'filesize': source.get('size'),
+ 'container': source.get('container'),
+ 'vcodec': source.get('container'),
+ })
+ else:
+ formats.extend(self._extract_f4m_formats(source.get('streaming_src'), video_id))
+
+ self._sort_formats(formats)
+
+ return {
+ 'id': video_id,
+ 'title': title,
+ 'description': description,
+ 'thumbnail': thumbnail,
+ 'timestamp': timestamp,
+ 'duration': duration,
+ 'formats': formats,
+ }