youtube-dl/youtube_dl/extractor/brightcove.py

# coding: utf-8
from __future__ import unicode_literals

import base64
import json
import re
import struct

from .common import InfoExtractor
from .adobepass import AdobePassIE
from ..compat import (
    compat_etree_fromstring,
    compat_parse_qs,
    compat_str,
    compat_urllib_parse_urlparse,
    compat_urlparse,
    compat_xml_parse_error,
    compat_HTTPError,
)
from ..utils import (
    determine_ext,
    ExtractorError,
    extract_attributes,
    find_xpath_attr,
    fix_xml_ampersands,
    float_or_none,
    js_to_json,
    int_or_none,
    parse_iso8601,
    unescapeHTML,
    unsmuggle_url,
    update_url_query,
    clean_html,
    mimetype2ext,
)


class BrightcoveLegacyIE(InfoExtractor):
    IE_NAME = 'brightcove:legacy'
    _VALID_URL = r'(?:https?://.*brightcove\.com/(services|viewer).*?\?|brightcove:)(?P<query>.*)'
    _FEDERATED_URL = 'http://c.brightcove.com/services/viewer/htmlFederated'

    _TESTS = [
        {
            # From http://www.8tv.cat/8aldia/videos/xavier-sala-i-martin-aquesta-tarda-a-8-al-dia/
            'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1654948606001&flashID=myExperience&%40videoPlayer=2371591881001',
            'md5': '5423e113865d26e40624dce2e4b45d95',
            'note': 'Test Brightcove downloads and detection in GenericIE',
            'info_dict': {
                'id': '2371591881001',
                'ext': 'mp4',
                'title': 'Xavier Sala i Martín: “Un banc que no presta és un banc zombi que no serveix per a res”',
                'uploader': '8TV',
                'description': 'md5:a950cc4285c43e44d763d036710cd9cd',
                'timestamp': 1368213670,
                'upload_date': '20130510',
                'uploader_id': '1589608506001',
            }
        },
        {
            # From http://medianetwork.oracle.com/video/player/1785452137001
            'url': 'http://c.brightcove.com/services/viewer/htmlFederated?playerID=1217746023001&flashID=myPlayer&%40videoPlayer=1785452137001',
            'info_dict': {
                'id': '1785452137001',
                'ext': 'flv',
                'title': 'JVMLS 2012: Arrays 2.0 - Opportunities and Challenges',
                'description': 'John Rose speaks at the JVM Language Summit, August 1, 2012.',
                'uploader': 'Oracle',
                'timestamp': 1344975024,
                'upload_date': '20120814',
                'uploader_id': '1460825906',
            },
        },
        {
            # From http://mashable.com/2013/10/26/thermoelectric-bracelet-lets-you-control-your-body-temperature/
            'url': 'http://c.brightcove.com/services/viewer/federated_f9?&playerID=1265504713001&publisherID=AQ%7E%7E%2CAAABBzUwv1E%7E%2CxP-xFHVUstiMFlNYfvF4G9yFnNaqCw_9&videoID=2750934548001',
            'info_dict': {
                'id': '2750934548001',
                'ext': 'mp4',
                'title': 'This Bracelet Acts as a Personal Thermostat',
                'description': 'md5:547b78c64f4112766ccf4e151c20b6a0',
                'uploader': 'Mashable',
                'timestamp': 1382041798,
                'upload_date': '20131017',
                'uploader_id': '1130468786001',
            },
        },
        {
            # test that the default referer works
            # from http://national.ballet.ca/interact/video/Lost_in_Motion_II/
            'url': 'http://link.brightcove.com/services/player/bcpid756015033001?bckey=AQ~~,AAAApYJi_Ck~,GxhXCegT1Dp39ilhXuxMJxasUhVNZiil&bctid=2878862109001',
            'info_dict': {
                'id': '2878862109001',
                'ext': 'mp4',
                'title': 'Lost in Motion II',
                'description': 'md5:363109c02998fee92ec02211bd8000df',
                'uploader': 'National Ballet of Canada',
            },
            'skip': 'Video gone',
        },
        {
            # test flv videos served by akamaihd.net
            # From http://www.redbull.com/en/bike/stories/1331655643987/replay-uci-dh-world-cup-2014-from-fort-william
            'url': 'http://c.brightcove.com/services/viewer/htmlFederated?%40videoPlayer=ref%3Aevent-stream-356&linkBaseURL=http%3A%2F%2Fwww.redbull.com%2Fen%2Fbike%2Fvideos%2F1331655630249%2Freplay-uci-fort-william-2014-dh&playerKey=AQ%7E%7E%2CAAAApYJ7UqE%7E%2Cxqr_zXk0I-zzNndy8NlHogrCb5QdyZRf&playerID=1398061561001#__youtubedl_smuggle=%7B%22Referer%22%3A+%22http%3A%2F%2Fwww.redbull.com%2Fen%2Fbike%2Fstories%2F1331655643987%2Freplay-uci-dh-world-cup-2014-from-fort-william%22%7D',
            # The md5 checksum changes on each download
            'info_dict': {
                'id': '3750436379001',
                'ext': 'flv',
                'title': 'UCI MTB World Cup 2014: Fort William, UK - Downhill Finals',
                'uploader': 'RBTV Old (do not use)',
                'description': 'UCI MTB World Cup 2014: Fort William, UK - Downhill Finals',
                'timestamp': 1409122195,
                'upload_date': '20140827',
                'uploader_id': '710858724001',
            },
            'skip': 'Video gone',
        },
        {
            # playlist with 'videoList'
            # from http://support.brightcove.com/en/video-cloud/docs/playlist-support-single-video-players
            '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,
        },
        {
            # playlist with 'playlistTab' (https://github.com/ytdl-org/youtube-dl/issues/9965)
            'url': 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=AQ%7E%7E,AAABXlLMdok%7E,NJ4EoMlZ4rZdx9eU1rkMVd8EaYPBBUlg',
            'info_dict': {
                'id': '1522758701001',
                'title': 'Lesson 08',
            },
            'playlist_mincount': 10,
        },
        {
            # playerID inferred from bcpid
            # from http://www.un.org/chinese/News/story.asp?NewsID=27724
            'url': 'https://link.brightcove.com/services/player/bcpid1722935254001/?bctid=5360463607001&autoStart=false&secureConnections=true&width=650&height=350',
            'only_matching': True,  # Tested in GenericIE
        }
    ]
    FLV_VCODECS = {
        1: 'SORENSON',
        2: 'ON2',
        3: 'H264',
        4: 'VP8',
    }

    @classmethod
    def _build_brighcove_url(cls, object_str):
        """
        Build a Brightcove url from a xml string containing
        <object class="BrightcoveExperience">{params}</object>
        """

        # Fix up some stupid HTML, see https://github.com/ytdl-org/youtube-dl/issues/1553
        object_str = re.sub(r'(<param(?:\s+[a-zA-Z0-9_]+="[^"]*")*)>',
                            lambda m: m.group(1) + '/>', object_str)
        # Fix up some stupid XML, see https://github.com/ytdl-org/youtube-dl/issues/1608
        object_str = object_str.replace('<--', '<!--')
        # remove namespace to simplify extraction
        object_str = re.sub(r'(<object[^>]*)(xmlns=".*?")', r'\1', object_str)
        object_str = fix_xml_ampersands(object_str)

        try:
            object_doc = compat_etree_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:
            flashvars = dict(
                (k, v[0])
                for k, v in compat_parse_qs(fv_el.attrib['value']).items())
        else:
            flashvars = {}

        data_url = object_doc.attrib.get('data', '')
        data_url_params = compat_parse_qs(compat_urllib_parse_urlparse(data_url).query)

        def find_param(name):
            if name in flashvars:
                return flashvars[name]
            node = find_xpath_attr(object_doc, './param', 'name', name)
            if node is not None:
                return node.attrib['value']
            return data_url_params.get(name)

        params = {}

        playerID = find_param('playerID') or find_param('playerId')
        if playerID is None:
            raise ExtractorError('Cannot find player ID')
        params['playerID'] = playerID

        playerKey = find_param('playerKey')
        # Not all pages define this value
        if playerKey is not None:
            params['playerKey'] = playerKey
        # These fields hold the id of the video
        videoPlayer = find_param('@videoPlayer') or find_param('videoId') or find_param('videoID') or find_param('@videoList')
        if videoPlayer is not None:
            if isinstance(videoPlayer, list):
                videoPlayer = videoPlayer[0]
            videoPlayer = videoPlayer.strip()
            # UUID is also possible for videoPlayer (e.g.
            # http://www.popcornflix.com/hoodies-vs-hooligans/7f2d2b87-bbf2-4623-acfb-ea942b4f01dd
            # or http://www8.hp.com/cn/zh/home.html)
            if not (re.match(
                    r'^(?:\d+|[\da-fA-F]{8}-?[\da-fA-F]{4}-?[\da-fA-F]{4}-?[\da-fA-F]{4}-?[\da-fA-F]{12})$',
                    videoPlayer) or videoPlayer.startswith('ref:')):
                return None
            params['@videoPlayer'] = videoPlayer
        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):
        return update_url_query(cls._FEDERATED_URL, params)

    @classmethod
    def _extract_brightcove_url(cls, webpage):
        """Try to extract the brightcove url from the webpage, returns None
        if it can't be found
        """
        urls = cls._extract_brightcove_urls(webpage)
        return urls[0] if urls else None

    @classmethod
    def _extract_brightcove_urls(cls, webpage):
        """Return a list of all Brightcove URLs from the webpage """

        url_m = re.search(
            r'''(?x)
                <meta\s+
                    (?:property|itemprop)=([\'"])(?:og:video|embedURL)\1[^>]+
                    content=([\'"])(?P<url>https?://(?:secure|c)\.brightcove.com/(?:(?!\2).)+)\2
            ''', webpage)
        if url_m:
            url = unescapeHTML(url_m.group('url'))
            # Some sites don't add it, we can't download with this url, for example:
            # http://www.ktvu.com/videos/news/raw-video-caltrain-releases-video-of-man-almost/vCTZdY/
            if 'playerKey' in url or 'videoId' in url or 'idVideo' in url:
                return [url]

        matches = re.findall(
            r'''(?sx)<object
            (?:
                [^>]+?class=[\'"][^>]*?BrightcoveExperience.*?[\'"] |
                [^>]*?>\s*<param\s+name="movie"\s+value="https?://[^/]*brightcove\.com/
            ).+?>\s*</object>''',
            webpage)
        if matches:
            return list(filter(None, [cls._build_brighcove_url(m) for m in matches]))

        matches = re.findall(r'(customBC\.createVideo\(.+?\);)', webpage)
        if matches:
            return list(filter(None, [
                cls._build_brighcove_url_from_js(custom_bc)
                for custom_bc in matches]))
        return [src for _, src in re.findall(
            r'<iframe[^>]+src=([\'"])((?:https?:)?//link\.brightcove\.com/services/player/(?!\1).+)\1', webpage)]

    def _real_extract(self, url):
        url, smuggled_data = unsmuggle_url(url, {})

        # Change the 'videoId' and others field to '@videoPlayer'
        url = re.sub(r'(?<=[?&])(videoI(d|D)|idVideo|bctid)', '%40videoPlayer', url)
        # Change bckey (used by bcove.me urls) to playerKey
        url = re.sub(r'(?<=[?&])bckey', 'playerKey', url)
        mobj = re.match(self._VALID_URL, url)
        query_str = mobj.group('query')
        query = compat_urlparse.parse_qs(query_str)

        videoPlayer = query.get('@videoPlayer')
        if videoPlayer:
            # We set the original url as the default 'Referer' header
            referer = smuggled_data.get('Referer', url)
            if 'playerID' not in query:
                mobj = re.search(r'/bcpid(\d+)', url)
                if mobj is not None:
                    query['playerID'] = [mobj.group(1)]
            return self._get_video_info(
                videoPlayer[0], query, referer=referer)
        elif 'playerKey' in query:
            player_key = query['playerKey']
            return self._get_playlist_info(player_key[0])
        else:
            raise ExtractorError(
                'Cannot find playerKey= variable. Did you forget quotes in a shell invocation?',
                expected=True)

    def _brightcove_new_url_result(self, publisher_id, video_id):
        brightcove_new_url = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s' % (publisher_id, video_id)
        return self.url_result(brightcove_new_url, BrightcoveNewIE.ie_key(), video_id)

    def _get_video_info(self, video_id, query, referer=None):
        headers = {}
        linkBase = query.get('linkBaseURL')
        if linkBase is not None:
            referer = linkBase[0]
        if referer is not None:
            headers['Referer'] = referer
        webpage = self._download_webpage(self._FEDERATED_URL, video_id, headers=headers, query=query)

        error_msg = self._html_search_regex(
            r"<h1>We're sorry.</h1>([\s\n]*<p>.*?</p>)+", webpage,
            'error message', default=None)
        if error_msg is not None:
            publisher_id = query.get('publisherId')
            if publisher_id and publisher_id[0].isdigit():
                publisher_id = publisher_id[0]
            if not publisher_id:
                player_key = query.get('playerKey')
                if player_key and ',' in player_key[0]:
                    player_key = player_key[0]
                else:
                    player_id = query.get('playerID')
                    if player_id and player_id[0].isdigit():
                        player_page = self._download_webpage(
                            'http://link.brightcove.com/services/player/bcpid' + player_id[0],
                            video_id, headers=headers, fatal=False)
                        if player_page:
                            player_key = self._search_regex(
                                r'<param\s+name="playerKey"\s+value="([\w~,-]+)"',
                                player_page, 'player key', fatal=False)
                if player_key:
                    enc_pub_id = player_key.split(',')[1].replace('~', '=')
                    publisher_id = struct.unpack('>Q', base64.urlsafe_b64decode(enc_pub_id))[0]
                if publisher_id:
                    return self._brightcove_new_url_result(publisher_id, video_id)
            raise ExtractorError(
                'brightcove said: %s' % error_msg, expected=True)

        self.report_extraction(video_id)
        info = self._search_regex(r'var experienceJSON = ({.*});', webpage, 'json')
        info = json.loads(info)['data']
        video_info = info['programmedContent']['videoPlayer']['mediaDTO']
        video_info['_youtubedl_adServerURL'] = info.get('adServerURL')

        return self._extract_video_info(video_info)

    def _get_playlist_info(self, player_key):
        info_url = 'http://c.brightcove.com/services/json/experience/runtime/?command=get_programming_for_experience&playerKey=%s' % player_key
        playlist_info = self._download_webpage(
            info_url, player_key, 'Downloading playlist information')

        json_data = json.loads(playlist_info)
        if 'videoList' in json_data:
            playlist_info = json_data['videoList']
            playlist_dto = playlist_info['mediaCollectionDTO']
        elif 'playlistTabs' in json_data:
            playlist_info = json_data['playlistTabs']
            playlist_dto = playlist_info['lineupListDTO']['playlistDTOs'][0]
        else:
            raise ExtractorError('Empty playlist')

        videos = [self._extract_video_info(video_info) for video_info in playlist_dto['videoDTOs']]

        return self.playlist_result(videos, playlist_id='%s' % playlist_info['id'],
                                    playlist_title=playlist_dto['displayName'])

    def _extract_video_info(self, video_info):
        video_id = compat_str(video_info['id'])
        publisher_id = video_info.get('publisherId')
        info = {
            'id': video_id,
            'title': video_info['displayName'].strip(),
            'description': video_info.get('shortDescription'),
            'thumbnail': video_info.get('videoStillURL') or video_info.get('thumbnailURL'),
            'uploader': video_info.get('publisherName'),
            'uploader_id': compat_str(publisher_id) if publisher_id else None,
            'duration': float_or_none(video_info.get('length'), 1000),
            'timestamp': int_or_none(video_info.get('creationDate'), 1000),
        }

        renditions = video_info.get('renditions', []) + video_info.get('IOSRenditions', [])
        if renditions:
            formats = []
            for rend in renditions:
                url = rend['defaultURL']
                if not url:
                    continue
                ext = None
                if rend['remote']:
                    url_comp = compat_urllib_parse_urlparse(url)
                    if url_comp.path.endswith('.m3u8'):
                        formats.extend(
                            self._extract_m3u8_formats(
                                url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
                        continue
                    elif 'akamaihd.net' in url_comp.netloc:
                        # This type of renditions are served through
                        # akamaihd.net, but they don't use f4m manifests
                        url = url.replace('control/', '') + '?&v=3.3.0&fp=13&r=FEEFJ&g=RTSJIMBMPFPB'
                        ext = 'flv'
                if ext is None:
                    ext = determine_ext(url)
                tbr = int_or_none(rend.get('encodingRate'), 1000)
                a_format = {
                    'format_id': 'http%s' % ('-%s' % tbr if tbr else ''),
                    'url': url,
                    'ext': ext,
                    'filesize': int_or_none(rend.get('size')) or None,
                    'tbr': tbr,
                }
                if rend.get('audioOnly'):
                    a_format.update({
                        'vcodec': 'none',
                    })
                else:
                    a_format.update({
                        'height': int_or_none(rend.get('frameHeight')),
                        'width': int_or_none(rend.get('frameWidth')),
                        'vcodec': rend.get('videoCodec'),
                    })

                # m3u8 manifests with remote == false are media playlists
                # Not calling _extract_m3u8_formats here to save network traffic
                if ext == 'm3u8':
                    a_format.update({
                        'format_id': 'hls%s' % ('-%s' % tbr if tbr else ''),
                        'ext': 'mp4',
                        'protocol': 'm3u8_native',
                    })

                formats.append(a_format)
            self._sort_formats(formats)
            info['formats'] = formats
        elif video_info.get('FLVFullLengthURL') is not None:
            info.update({
                'url': video_info['FLVFullLengthURL'],
                'vcodec': self.FLV_VCODECS.get(video_info.get('FLVFullCodec')),
                'filesize': int_or_none(video_info.get('FLVFullSize')),
            })

        if self._downloader.params.get('include_ads', False):
            adServerURL = video_info.get('_youtubedl_adServerURL')
            if adServerURL:
                ad_info = {
                    '_type': 'url',
                    'url': adServerURL,
                }
                if 'url' in info:
                    return {
                        '_type': 'playlist',
                        'title': info['title'],
                        'entries': [ad_info, info],
                    }
                else:
                    return ad_info

        if not info.get('url') and not info.get('formats'):
            uploader_id = info.get('uploader_id')
            if uploader_id:
                info.update(self._brightcove_new_url_result(uploader_id, video_id))
            else:
                raise ExtractorError('Unable to extract video url for %s' % video_id)
        return info


class BrightcoveNewIE(AdobePassIE):
    IE_NAME = 'brightcove:new'
    _VALID_URL = r'https?://players\.brightcove\.net/(?P<account_id>\d+)/(?P<player_id>[^/]+)_(?P<embed>[^/]+)/index\.html\?.*videoId=(?P<video_id>\d+|ref:[^&]+)'
    _TESTS = [{
        'url': 'http://players.brightcove.net/929656772001/e41d32dc-ec74-459e-a845-6c69f7b724ea_default/index.html?videoId=4463358922001',
        'md5': 'c8100925723840d4b0d243f7025703be',
        'info_dict': {
            'id': '4463358922001',
            'ext': 'mp4',
            'title': 'Meet the man behind Popcorn Time',
            'description': 'md5:eac376a4fe366edc70279bfb681aea16',
            'duration': 165.768,
            'timestamp': 1441391203,
            'upload_date': '20150904',
            'uploader_id': '929656772001',
            'formats': 'mincount:20',
        },
    }, {
        # with rtmp streams
        'url': 'http://players.brightcove.net/4036320279001/5d112ed9-283f-485f-a7f9-33f42e8bc042_default/index.html?videoId=4279049078001',
        'info_dict': {
            'id': '4279049078001',
            'ext': 'mp4',
            'title': 'Titansgrave: Chapter 0',
            'description': 'Titansgrave: Chapter 0',
            'duration': 1242.058,
            'timestamp': 1433556729,
            'upload_date': '20150606',
            'uploader_id': '4036320279001',
            'formats': 'mincount:39',
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        }
    }, {
        # ref: prefixed video id
        'url': 'http://players.brightcove.net/3910869709001/21519b5c-4b3b-4363-accb-bdc8f358f823_default/index.html?videoId=ref:7069442',
        'only_matching': True,
    }, {
        # non numeric ref: prefixed video id
        'url': 'http://players.brightcove.net/710858724001/default_default/index.html?videoId=ref:event-stream-356',
        'only_matching': True,
    }, {
        # unavailable video without message but with error_code
        'url': 'http://players.brightcove.net/1305187701/c832abfb-641b-44eb-9da0-2fe76786505f_default/index.html?videoId=4377407326001',
        'only_matching': True,
    }]

    @staticmethod
    def _extract_url(ie, webpage):
        urls = BrightcoveNewIE._extract_urls(ie, webpage)
        return urls[0] if urls else None

    @staticmethod
    def _extract_urls(ie, webpage):
        # Reference:
        # 1. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#setvideoiniframe
        # 2. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#tag
        # 3. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/publish-video.html#setvideousingjavascript
        # 4. http://docs.brightcove.com/en/video-cloud/brightcove-player/guides/in-page-embed-player-implementation.html
        # 5. https://support.brightcove.com/en/video-cloud/docs/dynamically-assigning-videos-player

        entries = []

        # Look for iframe embeds [1]
        for _, url in re.findall(
                r'<iframe[^>]+src=(["\'])((?:https?:)?//players\.brightcove\.net/\d+/[^/]+/index\.html.+?)\1', webpage):
            entries.append(url if url.startswith('http') else 'http:' + url)

        # Look for <video> tags [2] and embed_in_page embeds [3]
        # [2] looks like:
        for video, script_tag, account_id, player_id, embed in re.findall(
                r'''(?isx)
                    (<video\s+[^>]*\bdata-video-id\s*=\s*['"]?[^>]+>)
                    (?:.*?
                        (<script[^>]+
                            src=["\'](?:https?:)?//players\.brightcove\.net/
                            (\d+)/([^/]+)_([^/]+)/index(?:\.min)?\.js
                        )
                    )?
                ''', webpage):
            attrs = extract_attributes(video)

            # According to examples from [4] it's unclear whether video id
            # may be optional and what to do when it is
            video_id = attrs.get('data-video-id')
            if not video_id:
                continue

            account_id = account_id or attrs.get('data-account')
            if not account_id:
                continue

            player_id = player_id or attrs.get('data-player') or 'default'
            embed = embed or attrs.get('data-embed') or 'default'

            bc_url = 'http://players.brightcove.net/%s/%s_%s/index.html?videoId=%s' % (
                account_id, player_id, embed, video_id)

            # Some brightcove videos may be embedded with video tag only and
            # without script tag or any mentioning of brightcove at all. Such
            # embeds are considered ambiguous since they are matched based only
            # on data-video-id and data-account attributes and in the wild may
            # not be brightcove embeds at all. Let's check reconstructed
            # brightcove URLs in case of such embeds and only process valid
            # ones. By this we ensure there is indeed a brightcove embed.
            if not script_tag and not ie._is_valid_url(
                    bc_url, video_id, 'possible brightcove video'):
                continue

            entries.append(bc_url)

        return entries

    def _parse_brightcove_metadata(self, json_data, video_id, headers={}):
        title = json_data['name'].strip()

        formats = []
        for source in json_data.get('sources', []):
            container = source.get('container')
            ext = mimetype2ext(source.get('type'))
            src = source.get('src')
            # https://support.brightcove.com/playback-api-video-fields-reference#key_systems_object
            if ext == 'ism' or container == 'WVM' or source.get('key_systems'):
                continue
            elif ext == 'm3u8' or container == 'M2TS':
                if not src:
                    continue
                formats.extend(self._extract_m3u8_formats(
                    src, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
            elif ext == 'mpd':
                if not src:
                    continue
                formats.extend(self._extract_mpd_formats(src, video_id, 'dash', fatal=False))
            else:
                streaming_src = source.get('streaming_src')
                stream_name, app_name = source.get('stream_name'), source.get('app_name')
                if not src and not streaming_src and (not stream_name or not app_name):
                    continue
                tbr = float_or_none(source.get('avg_bitrate'), 1000)
                height = int_or_none(source.get('height'))
                width = int_or_none(source.get('width'))
                f = {
                    'tbr': tbr,
                    'filesize': int_or_none(source.get('size')),
                    'container': container,
                    'ext': ext or container.lower(),
                }
                if width == 0 and height == 0:
                    f.update({
                        'vcodec': 'none',
                    })
                else:
                    f.update({
                        'width': width,
                        'height': height,
                        'vcodec': source.get('codec'),
                    })

                def build_format_id(kind):
                    format_id = kind
                    if tbr:
                        format_id += '-%dk' % int(tbr)
                    if height:
                        format_id += '-%dp' % height
                    return format_id

                if src or streaming_src:
                    f.update({
                        'url': src or streaming_src,
                        'format_id': build_format_id('http' if src else 'http-streaming'),
                        'source_preference': 0 if src else -1,
                    })
                else:
                    f.update({
                        'url': app_name,
                        'play_path': stream_name,
                        'format_id': build_format_id('rtmp'),
                    })
                formats.append(f)
        if not formats:
            # for sonyliv.com DRM protected videos
            s3_source_url = json_data.get('custom_fields', {}).get('s3sourceurl')
            if s3_source_url:
                formats.append({
                    'url': s3_source_url,
                    'format_id': 'source',
                })

        errors = json_data.get('errors')
        if not formats and errors:
            error = errors[0]
            raise ExtractorError(
                error.get('message') or error.get('error_subcode') or error['error_code'], expected=True)

        self._sort_formats(formats)

        for f in formats:
            f.setdefault('http_headers', {}).update(headers)

        subtitles = {}
        for text_track in json_data.get('text_tracks', []):
            if text_track.get('src'):
                subtitles.setdefault(text_track.get('srclang'), []).append({
                    'url': text_track['src'],
                })

        is_live = False
        duration = float_or_none(json_data.get('duration'), 1000)
        if duration is not None and duration <= 0:
            is_live = True

        return {
            'id': video_id,
            'title': self._live_title(title) if is_live else title,
            'description': clean_html(json_data.get('description')),
            'thumbnail': json_data.get('thumbnail') or json_data.get('poster'),
            'duration': duration,
            'timestamp': parse_iso8601(json_data.get('published_at')),
            'uploader_id': json_data.get('account_id'),
            'formats': formats,
            'subtitles': subtitles,
            'tags': json_data.get('tags', []),
            'is_live': is_live,
        }

    def _real_extract(self, url):
        url, smuggled_data = unsmuggle_url(url, {})
        self._initialize_geo_bypass({
            'countries': smuggled_data.get('geo_countries'),
            'ip_blocks': smuggled_data.get('geo_ip_blocks'),
        })

        account_id, player_id, embed, video_id = re.match(self._VALID_URL, url).groups()

        webpage = self._download_webpage(
            'http://players.brightcove.net/%s/%s_%s/index.min.js'
            % (account_id, player_id, embed), video_id)

        policy_key = None

        catalog = self._search_regex(
            r'catalog\(({.+?})\);', webpage, 'catalog', default=None)
        if catalog:
            catalog = self._parse_json(
                js_to_json(catalog), video_id, fatal=False)
            if catalog:
                policy_key = catalog.get('policyKey')

        if not policy_key:
            policy_key = self._search_regex(
                r'policyKey\s*:\s*(["\'])(?P<pk>.+?)\1',
                webpage, 'policy key', group='pk')

        api_url = 'https://edge.api.brightcove.com/playback/v1/accounts/%s/videos/%s' % (account_id, video_id)
        headers = {
            'Accept': 'application/json;pk=%s' % policy_key,
        }
        referrer = smuggled_data.get('referrer')
        if referrer:
            headers.update({
                'Referer': referrer,
                'Origin': re.search(r'https?://[^/]+', referrer).group(0),
            })
        try:
            json_data = self._download_json(api_url, video_id, headers=headers)
        except ExtractorError as e:
            if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
                json_data = self._parse_json(e.cause.read().decode(), video_id)[0]
                message = json_data.get('message') or json_data['error_code']
                if json_data.get('error_subcode') == 'CLIENT_GEO':
                    self.raise_geo_restricted(msg=message)
                raise ExtractorError(message, expected=True)
            raise

        errors = json_data.get('errors')
        if errors and errors[0].get('error_subcode') == 'TVE_AUTH':
            custom_fields = json_data['custom_fields']
            tve_token = self._extract_mvpd_auth(
                smuggled_data['source_url'], video_id,
                custom_fields['bcadobepassrequestorid'],
                custom_fields['bcadobepassresourceid'])
            json_data = self._download_json(
                api_url, video_id, headers={
                    'Accept': 'application/json;pk=%s' % policy_key
                }, query={
                    'tveToken': tve_token,
                })

        return self._parse_brightcove_metadata(
            json_data, video_id, headers=headers)