[yt-dlp.git] / yt_dlp / extractor / crunchyroll.py

import base64
import json
import re
import urllib.request
import xml.etree.ElementTree
import zlib
from hashlib import sha1
from math import floor, pow, sqrt

from .common import InfoExtractor
from .vrv import VRVBaseIE
from ..aes import aes_cbc_decrypt
from ..compat import (
    compat_b64decode,
    compat_etree_fromstring,
    compat_str,
    compat_urllib_parse_urlencode,
    compat_urlparse,
)
from ..utils import (
    ExtractorError,
    bytes_to_intlist,
    extract_attributes,
    float_or_none,
    format_field,
    int_or_none,
    intlist_to_bytes,
    join_nonempty,
    lowercase_escape,
    merge_dicts,
    parse_iso8601,
    qualities,
    remove_end,
    sanitized_Request,
    traverse_obj,
    try_get,
    xpath_text,
)


class CrunchyrollBaseIE(InfoExtractor):
    _LOGIN_URL = 'https://www.crunchyroll.com/welcome/login'
    _API_BASE = 'https://api.crunchyroll.com'
    _NETRC_MACHINE = 'crunchyroll'

    def _call_rpc_api(self, method, video_id, note=None, data=None):
        data = data or {}
        data['req'] = 'RpcApi' + method
        data = compat_urllib_parse_urlencode(data).encode('utf-8')
        return self._download_xml(
            'https://www.crunchyroll.com/xml/',
            video_id, note, fatal=False, data=data, headers={
                'Content-Type': 'application/x-www-form-urlencoded',
            })

    def _perform_login(self, username, password):
        if self._get_cookies(self._LOGIN_URL).get('etp_rt'):
            return

        upsell_response = self._download_json(
            f'{self._API_BASE}/get_upsell_data.0.json', None, 'Getting session id',
            query={
                'sess_id': 1,
                'device_id': 'whatvalueshouldbeforweb',
                'device_type': 'com.crunchyroll.static',
                'access_token': 'giKq5eY27ny3cqz',
                'referer': self._LOGIN_URL
            })
        if upsell_response['code'] != 'ok':
            raise ExtractorError('Could not get session id')
        session_id = upsell_response['data']['session_id']

        login_response = self._download_json(
            f'{self._API_BASE}/login.1.json', None, 'Logging in',
            data=compat_urllib_parse_urlencode({
                'account': username,
                'password': password,
                'session_id': session_id
            }).encode('ascii'))
        if login_response['code'] != 'ok':
            raise ExtractorError('Login failed. Server message: %s' % login_response['message'], expected=True)
        if not self._get_cookies(self._LOGIN_URL).get('etp_rt'):
            raise ExtractorError('Login succeeded but did not set etp_rt cookie')

    # Beta-specific, but needed for redirects
    def _get_beta_embedded_json(self, webpage, display_id):
        initial_state = self._parse_json(self._search_regex(
            r'__INITIAL_STATE__\s*=\s*({.+?})\s*;', webpage, 'initial state'), display_id)
        app_config = self._parse_json(self._search_regex(
            r'__APP_CONFIG__\s*=\s*({.+?})\s*;', webpage, 'app config'), display_id)
        return initial_state, app_config

    def _redirect_to_beta(self, webpage, iekey, video_id):
        if not self._get_cookies(self._LOGIN_URL).get('etp_rt'):
            raise ExtractorError('Received a beta page from non-beta url when not logged in.')
        initial_state, app_config = self._get_beta_embedded_json(webpage, video_id)
        url = app_config['baseSiteUrl'] + initial_state['router']['locations']['current']['pathname']
        self.to_screen(f'{video_id}: Redirected to beta site - {url}')
        return self.url_result(f'{url}', iekey, video_id)

    @staticmethod
    def _add_skip_wall(url):
        parsed_url = compat_urlparse.urlparse(url)
        qs = compat_urlparse.parse_qs(parsed_url.query)
        # Always force skip_wall to bypass maturity wall, namely 18+ confirmation message:
        # > This content may be inappropriate for some people.
        # > Are you sure you want to continue?
        # since it's not disabled by default in crunchyroll account's settings.
        # See https://github.com/ytdl-org/youtube-dl/issues/7202.
        qs['skip_wall'] = ['1']
        return compat_urlparse.urlunparse(
            parsed_url._replace(query=compat_urllib_parse_urlencode(qs, True)))


class CrunchyrollIE(CrunchyrollBaseIE, VRVBaseIE):
    IE_NAME = 'crunchyroll'
    _VALID_URL = r'https?://(?:(?P<prefix>www|m)\.)?(?P<url>crunchyroll\.(?:com|fr)/(?:media(?:-|/\?id=)|(?!series/|watch/)(?:[^/]+/){1,2}[^/?&]*?)(?P<id>[0-9]+))(?:[/?&]|$)'
    _TESTS = [{
        'url': 'http://www.crunchyroll.com/wanna-be-the-strongest-in-the-world/episode-1-an-idol-wrestler-is-born-645513',
        'info_dict': {
            'id': '645513',
            'ext': 'mp4',
            'title': 'Wanna be the Strongest in the World Episode 1 – An Idol-Wrestler is Born!',
            'description': 'md5:2d17137920c64f2f49981a7797d275ef',
            'thumbnail': r're:^https?://.*\.jpg$',
            'uploader': 'Yomiuri Telecasting Corporation (YTV)',
            'upload_date': '20131013',
            'url': 're:(?!.*&amp)',
        },
        'params': {
            # rtmp
            'skip_download': True,
        },
        'skip': 'Video gone',
    }, {
        'url': 'http://www.crunchyroll.com/media-589804/culture-japan-1',
        'info_dict': {
            'id': '589804',
            'ext': 'flv',
            'title': 'Culture Japan Episode 1 – Rebuilding Japan after the 3.11',
            'description': 'md5:2fbc01f90b87e8e9137296f37b461c12',
            'thumbnail': r're:^https?://.*\.jpg$',
            'uploader': 'Danny Choo Network',
            'upload_date': '20120213',
        },
        'params': {
            # rtmp
            'skip_download': True,
        },
        'skip': 'Video gone',
    }, {
        'url': 'http://www.crunchyroll.com/rezero-starting-life-in-another-world-/episode-5-the-morning-of-our-promise-is-still-distant-702409',
        'info_dict': {
            'id': '702409',
            'ext': 'mp4',
            'title': compat_str,
            'description': compat_str,
            'thumbnail': r're:^https?://.*\.jpg$',
            'uploader': 'Re:Zero Partners',
            'timestamp': 1462098900,
            'upload_date': '20160501',
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }, {
        'url': 'http://www.crunchyroll.com/konosuba-gods-blessing-on-this-wonderful-world/episode-1-give-me-deliverance-from-this-judicial-injustice-727589',
        'info_dict': {
            'id': '727589',
            'ext': 'mp4',
            'title': compat_str,
            'description': compat_str,
            'thumbnail': r're:^https?://.*\.jpg$',
            'uploader': 'Kadokawa Pictures Inc.',
            'timestamp': 1484130900,
            'upload_date': '20170111',
            'series': compat_str,
            'season': "KONOSUBA -God's blessing on this wonderful world! 2",
            'season_number': 2,
            'episode': 'Give Me Deliverance From This Judicial Injustice!',
            'episode_number': 1,
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }, {
        'url': 'http://www.crunchyroll.fr/girl-friend-beta/episode-11-goodbye-la-mode-661697',
        'only_matching': True,
    }, {
        # geo-restricted (US), 18+ maturity wall, non-premium available
        'url': 'http://www.crunchyroll.com/cosplay-complex-ova/episode-1-the-birth-of-the-cosplay-club-565617',
        'only_matching': True,
    }, {
        # A description with double quotes
        'url': 'http://www.crunchyroll.com/11eyes/episode-1-piros-jszaka-red-night-535080',
        'info_dict': {
            'id': '535080',
            'ext': 'mp4',
            'title': compat_str,
            'description': compat_str,
            'uploader': 'Marvelous AQL Inc.',
            'timestamp': 1255512600,
            'upload_date': '20091014',
        },
        'params': {
            # Just test metadata extraction
            'skip_download': True,
        },
    }, {
        # make sure we can extract an uploader name that's not a link
        'url': 'http://www.crunchyroll.com/hakuoki-reimeiroku/episode-1-dawn-of-the-divine-warriors-606899',
        'info_dict': {
            'id': '606899',
            'ext': 'mp4',
            'title': 'Hakuoki Reimeiroku Episode 1 – Dawn of the Divine Warriors',
            'description': 'Ryunosuke was left to die, but Serizawa-san asked him a simple question "Do you want to live?"',
            'uploader': 'Geneon Entertainment',
            'upload_date': '20120717',
        },
        'params': {
            # just test metadata extraction
            'skip_download': True,
        },
        'skip': 'Video gone',
    }, {
        # A video with a vastly different season name compared to the series name
        'url': 'http://www.crunchyroll.com/nyarko-san-another-crawling-chaos/episode-1-test-590532',
        'info_dict': {
            'id': '590532',
            'ext': 'mp4',
            'title': compat_str,
            'description': compat_str,
            'uploader': 'TV TOKYO',
            'timestamp': 1330956000,
            'upload_date': '20120305',
            'series': 'Nyarko-san: Another Crawling Chaos',
            'season': 'Haiyoru! Nyaruani (ONA)',
        },
        'params': {
            # Just test metadata extraction
            'skip_download': True,
        },
    }, {
        'url': 'http://www.crunchyroll.com/media-723735',
        'only_matching': True,
    }, {
        'url': 'https://www.crunchyroll.com/en-gb/mob-psycho-100/episode-2-urban-legends-encountering-rumors-780921',
        'only_matching': True,
    }]

    _FORMAT_IDS = {
        '360': ('60', '106'),
        '480': ('61', '106'),
        '720': ('62', '106'),
        '1080': ('80', '108'),
    }

    def _download_webpage(self, url_or_request, *args, **kwargs):
        request = (url_or_request if isinstance(url_or_request, urllib.request.Request)
                   else sanitized_Request(url_or_request))
        # Accept-Language must be set explicitly to accept any language to avoid issues
        # similar to https://github.com/ytdl-org/youtube-dl/issues/6797.
        # Along with IP address Crunchyroll uses Accept-Language to guess whether georestriction
        # should be imposed or not (from what I can see it just takes the first language
        # ignoring the priority and requires it to correspond the IP). By the way this causes
        # Crunchyroll to not work in georestriction cases in some browsers that don't place
        # the locale lang first in header. However allowing any language seems to workaround the issue.
        request.add_header('Accept-Language', '*')
        return super(CrunchyrollBaseIE, self)._download_webpage(request, *args, **kwargs)

    def _decrypt_subtitles(self, data, iv, id):
        data = bytes_to_intlist(compat_b64decode(data))
        iv = bytes_to_intlist(compat_b64decode(iv))
        id = int(id)

        def obfuscate_key_aux(count, modulo, start):
            output = list(start)
            for _ in range(count):
                output.append(output[-1] + output[-2])
            # cut off start values
            output = output[2:]
            output = list(map(lambda x: x % modulo + 33, output))
            return output

        def obfuscate_key(key):
            num1 = int(floor(pow(2, 25) * sqrt(6.9)))
            num2 = (num1 ^ key) << 5
            num3 = key ^ num1
            num4 = num3 ^ (num3 >> 3) ^ num2
            prefix = intlist_to_bytes(obfuscate_key_aux(20, 97, (1, 2)))
            shaHash = bytes_to_intlist(sha1(prefix + str(num4).encode('ascii')).digest())
            # Extend 160 Bit hash to 256 Bit
            return shaHash + [0] * 12

        key = obfuscate_key(id)

        decrypted_data = intlist_to_bytes(aes_cbc_decrypt(data, key, iv))
        return zlib.decompress(decrypted_data)

    def _convert_subtitles_to_srt(self, sub_root):
        output = ''

        for i, event in enumerate(sub_root.findall('./events/event'), 1):
            start = event.attrib['start'].replace('.', ',')
            end = event.attrib['end'].replace('.', ',')
            text = event.attrib['text'].replace('\\N', '\n')
            output += '%d\n%s --> %s\n%s\n\n' % (i, start, end, text)
        return output

    def _convert_subtitles_to_ass(self, sub_root):
        output = ''

        def ass_bool(strvalue):
            assvalue = '0'
            if strvalue == '1':
                assvalue = '-1'
            return assvalue

        output = '[Script Info]\n'
        output += 'Title: %s\n' % sub_root.attrib['title']
        output += 'ScriptType: v4.00+\n'
        output += 'WrapStyle: %s\n' % sub_root.attrib['wrap_style']
        output += 'PlayResX: %s\n' % sub_root.attrib['play_res_x']
        output += 'PlayResY: %s\n' % sub_root.attrib['play_res_y']
        output += """
[V4+ Styles]
Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
"""
        for style in sub_root.findall('./styles/style'):
            output += 'Style: ' + style.attrib['name']
            output += ',' + style.attrib['font_name']
            output += ',' + style.attrib['font_size']
            output += ',' + style.attrib['primary_colour']
            output += ',' + style.attrib['secondary_colour']
            output += ',' + style.attrib['outline_colour']
            output += ',' + style.attrib['back_colour']
            output += ',' + ass_bool(style.attrib['bold'])
            output += ',' + ass_bool(style.attrib['italic'])
            output += ',' + ass_bool(style.attrib['underline'])
            output += ',' + ass_bool(style.attrib['strikeout'])
            output += ',' + style.attrib['scale_x']
            output += ',' + style.attrib['scale_y']
            output += ',' + style.attrib['spacing']
            output += ',' + style.attrib['angle']
            output += ',' + style.attrib['border_style']
            output += ',' + style.attrib['outline']
            output += ',' + style.attrib['shadow']
            output += ',' + style.attrib['alignment']
            output += ',' + style.attrib['margin_l']
            output += ',' + style.attrib['margin_r']
            output += ',' + style.attrib['margin_v']
            output += ',' + style.attrib['encoding']
            output += '\n'

        output += """
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
"""
        for event in sub_root.findall('./events/event'):
            output += 'Dialogue: 0'
            output += ',' + event.attrib['start']
            output += ',' + event.attrib['end']
            output += ',' + event.attrib['style']
            output += ',' + event.attrib['name']
            output += ',' + event.attrib['margin_l']
            output += ',' + event.attrib['margin_r']
            output += ',' + event.attrib['margin_v']
            output += ',' + event.attrib['effect']
            output += ',' + event.attrib['text']
            output += '\n'

        return output

    def _extract_subtitles(self, subtitle):
        sub_root = compat_etree_fromstring(subtitle)
        return [{
            'ext': 'srt',
            'data': self._convert_subtitles_to_srt(sub_root),
        }, {
            'ext': 'ass',
            'data': self._convert_subtitles_to_ass(sub_root),
        }]

    def _get_subtitles(self, video_id, webpage):
        subtitles = {}
        for sub_id, sub_name in re.findall(r'\bssid=([0-9]+)"[^>]+?\btitle="([^"]+)', webpage):
            sub_doc = self._call_rpc_api(
                'Subtitle_GetXml', video_id,
                'Downloading subtitles for ' + sub_name, data={
                    'subtitle_script_id': sub_id,
                })
            if not isinstance(sub_doc, xml.etree.ElementTree.Element):
                continue
            sid = sub_doc.get('id')
            iv = xpath_text(sub_doc, 'iv', 'subtitle iv')
            data = xpath_text(sub_doc, 'data', 'subtitle data')
            if not sid or not iv or not data:
                continue
            subtitle = self._decrypt_subtitles(data, iv, sid).decode('utf-8')
            lang_code = self._search_regex(r'lang_code=["\']([^"\']+)', subtitle, 'subtitle_lang_code', fatal=False)
            if not lang_code:
                continue
            subtitles[lang_code] = self._extract_subtitles(subtitle)
        return subtitles

    def _real_extract(self, url):
        mobj = self._match_valid_url(url)
        video_id = mobj.group('id')

        if mobj.group('prefix') == 'm':
            mobile_webpage = self._download_webpage(url, video_id, 'Downloading mobile webpage')
            webpage_url = self._search_regex(r'<link rel="canonical" href="([^"]+)" />', mobile_webpage, 'webpage_url')
        else:
            webpage_url = 'http://www.' + mobj.group('url')

        webpage = self._download_webpage(
            self._add_skip_wall(webpage_url), video_id,
            headers=self.geo_verification_headers())
        if re.search(r'<div id="preload-data">', webpage):
            return self._redirect_to_beta(webpage, CrunchyrollBetaIE.ie_key(), video_id)
        note_m = self._html_search_regex(
            r'<div class="showmedia-trailer-notice">(.+?)</div>',
            webpage, 'trailer-notice', default='')
        if note_m:
            raise ExtractorError(note_m, expected=True)

        mobj = re.search(r'Page\.messaging_box_controller\.addItems\(\[(?P<msg>{.+?})\]\)', webpage)
        if mobj:
            msg = json.loads(mobj.group('msg'))
            if msg.get('type') == 'error':
                raise ExtractorError('crunchyroll returned error: %s' % msg['message_body'], expected=True)

        if 'To view this, please log in to verify you are 18 or older.' in webpage:
            self.raise_login_required()

        media = self._parse_json(self._search_regex(
            r'vilos\.config\.media\s*=\s*({.+?});',
            webpage, 'vilos media', default='{}'), video_id)
        media_metadata = media.get('metadata') or {}

        language = self._search_regex(
            r'(?:vilos\.config\.player\.language|LOCALE)\s*=\s*(["\'])(?P<lang>(?:(?!\1).)+)\1',
            webpage, 'language', default=None, group='lang')

        video_title = self._html_search_regex(
            (r'(?s)<h1[^>]*>((?:(?!<h1).)*?<(?:span[^>]+itemprop=["\']title["\']|meta[^>]+itemprop=["\']position["\'])[^>]*>(?:(?!<h1).)+?)</h1>',
             r'<title>(.+?),\s+-\s+.+? Crunchyroll'),
            webpage, 'video_title', default=None)
        if not video_title:
            video_title = re.sub(r'^Watch\s+', '', self._og_search_description(webpage))
        video_title = re.sub(r' {2,}', ' ', video_title)
        video_description = (self._parse_json(self._html_search_regex(
            r'<script[^>]*>\s*.+?\[media_id=%s\].+?({.+?"description"\s*:.+?})\);' % video_id,
            webpage, 'description', default='{}'), video_id) or media_metadata).get('description')

        thumbnails = []
        thumbnail_url = (self._parse_json(self._html_search_regex(
            r'<script type="application\/ld\+json">\n\s*(.+?)<\/script>',
            webpage, 'thumbnail_url', default='{}'), video_id)).get('image')
        if thumbnail_url:
            thumbnails.append({
                'url': thumbnail_url,
                'width': 1920,
                'height': 1080
            })

        if video_description:
            video_description = lowercase_escape(video_description.replace(r'\r\n', '\n'))
        video_uploader = self._html_search_regex(
            # try looking for both an uploader that's a link and one that's not
            [r'<a[^>]+href="/publisher/[^"]+"[^>]*>([^<]+)</a>', r'<div>\s*Publisher:\s*<span>\s*(.+?)\s*</span>\s*</div>'],
            webpage, 'video_uploader', default=False)

        requested_languages = self._configuration_arg('language')
        requested_hardsubs = [('' if val == 'none' else val) for val in self._configuration_arg('hardsub')]
        language_preference = qualities((requested_languages or [language or ''])[::-1])
        hardsub_preference = qualities((requested_hardsubs or ['', language or ''])[::-1])

        formats = []
        for stream in media.get('streams', []):
            audio_lang = stream.get('audio_lang') or ''
            hardsub_lang = stream.get('hardsub_lang') or ''
            if (requested_languages and audio_lang.lower() not in requested_languages
                    or requested_hardsubs and hardsub_lang.lower() not in requested_hardsubs):
                continue
            vrv_formats = self._extract_vrv_formats(
                stream.get('url'), video_id, stream.get('format'),
                audio_lang, hardsub_lang)
            for f in vrv_formats:
                f['language_preference'] = language_preference(audio_lang)
                f['quality'] = hardsub_preference(hardsub_lang)
            formats.extend(vrv_formats)
        if not formats:
            available_fmts = []
            for a, fmt in re.findall(r'(<a[^>]+token=["\']showmedia\.([0-9]{3,4})p["\'][^>]+>)', webpage):
                attrs = extract_attributes(a)
                href = attrs.get('href')
                if href and '/freetrial' in href:
                    continue
                available_fmts.append(fmt)
            if not available_fmts:
                for p in (r'token=["\']showmedia\.([0-9]{3,4})p"', r'showmedia\.([0-9]{3,4})p'):
                    available_fmts = re.findall(p, webpage)
                    if available_fmts:
                        break
            if not available_fmts:
                available_fmts = self._FORMAT_IDS.keys()
            video_encode_ids = []

            for fmt in available_fmts:
                stream_quality, stream_format = self._FORMAT_IDS[fmt]
                video_format = fmt + 'p'
                stream_infos = []
                streamdata = self._call_rpc_api(
                    'VideoPlayer_GetStandardConfig', video_id,
                    'Downloading media info for %s' % video_format, data={
                        'media_id': video_id,
                        'video_format': stream_format,
                        'video_quality': stream_quality,
                        'current_page': url,
                    })
                if isinstance(streamdata, xml.etree.ElementTree.Element):
                    stream_info = streamdata.find('./{default}preload/stream_info')
                    if stream_info is not None:
                        stream_infos.append(stream_info)
                stream_info = self._call_rpc_api(
                    'VideoEncode_GetStreamInfo', video_id,
                    'Downloading stream info for %s' % video_format, data={
                        'media_id': video_id,
                        'video_format': stream_format,
                        'video_encode_quality': stream_quality,
                    })
                if isinstance(stream_info, xml.etree.ElementTree.Element):
                    stream_infos.append(stream_info)
                for stream_info in stream_infos:
                    video_encode_id = xpath_text(stream_info, './video_encode_id')
                    if video_encode_id in video_encode_ids:
                        continue
                    video_encode_ids.append(video_encode_id)

                    video_file = xpath_text(stream_info, './file')
                    if not video_file:
                        continue
                    if video_file.startswith('http'):
                        formats.extend(self._extract_m3u8_formats(
                            video_file, video_id, 'mp4', entry_protocol='m3u8_native',
                            m3u8_id='hls', fatal=False))
                        continue

                    video_url = xpath_text(stream_info, './host')
                    if not video_url:
                        continue
                    metadata = stream_info.find('./metadata')
                    format_info = {
                        'format': video_format,
                        'height': int_or_none(xpath_text(metadata, './height')),
                        'width': int_or_none(xpath_text(metadata, './width')),
                    }

                    if '.fplive.net/' in video_url:
                        video_url = re.sub(r'^rtmpe?://', 'http://', video_url.strip())
                        parsed_video_url = compat_urlparse.urlparse(video_url)
                        direct_video_url = compat_urlparse.urlunparse(parsed_video_url._replace(
                            netloc='v.lvlt.crcdn.net',
                            path='%s/%s' % (remove_end(parsed_video_url.path, '/'), video_file.split(':')[-1])))
                        if self._is_valid_url(direct_video_url, video_id, video_format):
                            format_info.update({
                                'format_id': 'http-' + video_format,
                                'url': direct_video_url,
                            })
                            formats.append(format_info)
                            continue

                    format_info.update({
                        'format_id': 'rtmp-' + video_format,
                        'url': video_url,
                        'play_path': video_file,
                        'ext': 'flv',
                    })
                    formats.append(format_info)
        self._sort_formats(formats)

        metadata = self._call_rpc_api(
            'VideoPlayer_GetMediaMetadata', video_id,
            note='Downloading media info', data={
                'media_id': video_id,
            })

        subtitles = {}
        for subtitle in media.get('subtitles', []):
            subtitle_url = subtitle.get('url')
            if not subtitle_url:
                continue
            subtitles.setdefault(subtitle.get('language', 'enUS'), []).append({
                'url': subtitle_url,
                'ext': subtitle.get('format', 'ass'),
            })
        if not subtitles:
            subtitles = self.extract_subtitles(video_id, webpage)

        # webpage provide more accurate data than series_title from XML
        series = self._html_search_regex(
            r'(?s)<h\d[^>]+\bid=["\']showmedia_about_episode_num[^>]+>(.+?)</h\d',
            webpage, 'series', fatal=False)

        season = episode = episode_number = duration = None

        if isinstance(metadata, xml.etree.ElementTree.Element):
            season = xpath_text(metadata, 'series_title')
            episode = xpath_text(metadata, 'episode_title')
            episode_number = int_or_none(xpath_text(metadata, 'episode_number'))
            duration = float_or_none(media_metadata.get('duration'), 1000)

        if not episode:
            episode = media_metadata.get('title')
        if not episode_number:
            episode_number = int_or_none(media_metadata.get('episode_number'))
        thumbnail_url = try_get(media, lambda x: x['thumbnail']['url'])
        if thumbnail_url:
            thumbnails.append({
                'url': thumbnail_url,
                'width': 640,
                'height': 360
            })

        season_number = int_or_none(self._search_regex(
            r'(?s)<h\d[^>]+id=["\']showmedia_about_episode_num[^>]+>.+?</h\d>\s*<h4>\s*Season (\d+)',
            webpage, 'season number', default=None))

        info = self._search_json_ld(webpage, video_id, default={})

        return merge_dicts({
            'id': video_id,
            'title': video_title,
            'description': video_description,
            'duration': duration,
            'thumbnails': thumbnails,
            'uploader': video_uploader,
            'series': series,
            'season': season,
            'season_number': season_number,
            'episode': episode,
            'episode_number': episode_number,
            'subtitles': subtitles,
            'formats': formats,
        }, info)


class CrunchyrollShowPlaylistIE(CrunchyrollBaseIE):
    IE_NAME = 'crunchyroll:playlist'
    _VALID_URL = r'https?://(?:(?P<prefix>www|m)\.)?(?P<url>crunchyroll\.com/(?:\w{2}(?:-\w{2})?/)?(?!(?:news|anime-news|library|forum|launchcalendar|lineup|store|comics|freetrial|login|media-\d+))(?P<id>[\w\-]+))/?(?:\?|$)'

    _TESTS = [{
        'url': 'https://www.crunchyroll.com/a-bridge-to-the-starry-skies-hoshizora-e-kakaru-hashi',
        'info_dict': {
            'id': 'a-bridge-to-the-starry-skies-hoshizora-e-kakaru-hashi',
            'title': 'A Bridge to the Starry Skies - Hoshizora e Kakaru Hashi'
        },
        'playlist_count': 13,
    }, {
        # geo-restricted (US), 18+ maturity wall, non-premium available
        'url': 'http://www.crunchyroll.com/cosplay-complex-ova',
        'info_dict': {
            'id': 'cosplay-complex-ova',
            'title': 'Cosplay Complex OVA'
        },
        'playlist_count': 3,
        'skip': 'Georestricted',
    }, {
        # geo-restricted (US), 18+ maturity wall, non-premium will be available since 2015.11.14
        'url': 'http://www.crunchyroll.com/ladies-versus-butlers?skip_wall=1',
        'only_matching': True,
    }, {
        'url': 'http://www.crunchyroll.com/fr/ladies-versus-butlers',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        show_id = self._match_id(url)

        webpage = self._download_webpage(
            # https:// gives a 403, but http:// does not
            self._add_skip_wall(url).replace('https://', 'http://'), show_id,
            headers=self.geo_verification_headers())
        if re.search(r'<div id="preload-data">', webpage):
            return self._redirect_to_beta(webpage, CrunchyrollBetaShowIE.ie_key(), show_id)
        title = self._html_search_meta('name', webpage, default=None)

        episode_re = r'<li id="showview_videos_media_(\d+)"[^>]+>.*?<a href="([^"]+)"'
        season_re = r'<a [^>]+season-dropdown[^>]+>([^<]+)'
        paths = re.findall(f'(?s){episode_re}|{season_re}', webpage)

        entries, current_season = [], None
        for ep_id, ep, season in paths:
            if season:
                current_season = season
                continue
            entries.append(self.url_result(
                f'http://www.crunchyroll.com{ep}', CrunchyrollIE.ie_key(), ep_id, season=current_season))

        return {
            '_type': 'playlist',
            'id': show_id,
            'title': title,
            'entries': reversed(entries),
        }


class CrunchyrollBetaBaseIE(CrunchyrollBaseIE):
    params = None

    def _get_params(self, lang):
        if not CrunchyrollBetaBaseIE.params:
            initial_state, app_config = self._get_beta_embedded_json(self._download_webpage(
                f'https://beta.crunchyroll.com/{lang}', None, note='Retrieving main page'), None)
            api_domain = app_config['cxApiParams']['apiDomain']
            basic_token = str(base64.b64encode(('%s:' % app_config['cxApiParams']['accountAuthClientId']).encode('ascii')), 'ascii')
            auth_response = self._download_json(
                f'{api_domain}/auth/v1/token', None, note='Authenticating with cookie',
                headers={
                    'Authorization': 'Basic ' + basic_token
                }, data='grant_type=etp_rt_cookie'.encode('ascii'))
            policy_response = self._download_json(
                f'{api_domain}/index/v2', None, note='Retrieving signed policy',
                headers={
                    'Authorization': auth_response['token_type'] + ' ' + auth_response['access_token']
                })
            cms = traverse_obj(policy_response, 'cms_beta', 'cms')
            bucket = cms['bucket']
            params = {
                'Policy': cms['policy'],
                'Signature': cms['signature'],
                'Key-Pair-Id': cms['key_pair_id']
            }
            locale = traverse_obj(initial_state, ('localization', 'locale'))
            if locale:
                params['locale'] = locale
            CrunchyrollBetaBaseIE.params = (api_domain, bucket, params)
        return CrunchyrollBetaBaseIE.params

    def _redirect_from_beta(self, url, lang, internal_id, display_id, is_episode, iekey):
        initial_state, app_config = self._get_beta_embedded_json(self._download_webpage(url, display_id), display_id)
        content_data = initial_state['content']['byId'][internal_id]
        if is_episode:
            video_id = content_data['external_id'].split('.')[1]
            series_id = content_data['episode_metadata']['series_slug_title']
        else:
            series_id = content_data['slug_title']
        series_id = re.sub(r'-{2,}', '-', series_id)
        url = f'https://www.crunchyroll.com/{lang}{series_id}'
        if is_episode:
            url = url + f'/{display_id}-{video_id}'
        self.to_screen(f'{display_id}: Not logged in. Redirecting to non-beta site - {url}')
        return self.url_result(url, iekey, display_id)


class CrunchyrollBetaIE(CrunchyrollBetaBaseIE):
    IE_NAME = 'crunchyroll:beta'
    _VALID_URL = r'https?://beta\.crunchyroll\.com/(?P<lang>(?:\w{2}(?:-\w{2})?/)?)watch/(?P<id>\w+)/(?P<display_id>[\w\-]*)/?(?:\?|$)'
    _TESTS = [{
        'url': 'https://beta.crunchyroll.com/watch/GY2P1Q98Y/to-the-future',
        'info_dict': {
            'id': 'GY2P1Q98Y',
            'ext': 'mp4',
            'duration': 1380.241,
            'timestamp': 1459632600,
            'description': 'md5:a022fbec4fbb023d43631032c91ed64b',
            'title': 'World Trigger Episode 73 – To the Future',
            'upload_date': '20160402',
            'series': 'World Trigger',
            'series_id': 'GR757DMKY',
            'season': 'World Trigger',
            'season_id': 'GR9P39NJ6',
            'season_number': 1,
            'episode': 'To the Future',
            'episode_number': 73,
            'thumbnail': r're:^https://beta.crunchyroll.com/imgsrv/.*\.jpeg$',
        },
        'params': {'skip_download': 'm3u8'},
    }, {
        'url': 'https://beta.crunchyroll.com/watch/GY2P1Q98Y/',
        'only_matching': True,
    }, {
        'url': 'https://beta.crunchyroll.com/pt-br/watch/G8WUN8VKP/the-ruler-of-conspiracy',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        lang, internal_id, display_id = self._match_valid_url(url).group('lang', 'id', 'display_id')

        if not self._get_cookies(url).get('etp_rt'):
            return self._redirect_from_beta(url, lang, internal_id, display_id, True, CrunchyrollIE.ie_key())

        api_domain, bucket, params = self._get_params(lang)

        episode_response = self._download_json(
            f'{api_domain}/cms/v2{bucket}/episodes/{internal_id}', display_id,
            note='Retrieving episode metadata', query=params)
        if episode_response.get('is_premium_only') and not episode_response.get('playback'):
            raise ExtractorError('This video is for premium members only.', expected=True)

        stream_response = self._download_json(
            f'{api_domain}{episode_response["__links__"]["streams"]["href"]}', display_id,
            note='Retrieving stream info', query=params)
        get_streams = lambda name: (traverse_obj(stream_response, name) or {}).items()

        requested_hardsubs = [('' if val == 'none' else val) for val in (self._configuration_arg('hardsub') or ['none'])]
        hardsub_preference = qualities(requested_hardsubs[::-1])
        requested_formats = self._configuration_arg('format') or ['adaptive_hls']

        formats = []
        for stream_type, streams in get_streams('streams'):
            if stream_type not in requested_formats:
                continue
            for stream in streams.values():
                hardsub_lang = stream.get('hardsub_locale') or ''
                if hardsub_lang.lower() not in requested_hardsubs:
                    continue
                format_id = join_nonempty(stream_type, format_field(stream, 'hardsub_locale', 'hardsub-%s'))
                if not stream.get('url'):
                    continue
                if stream_type.endswith('hls'):
                    adaptive_formats = self._extract_m3u8_formats(
                        stream['url'], display_id, 'mp4', m3u8_id=format_id,
                        fatal=False, note=f'Downloading {format_id} HLS manifest')
                elif stream_type.endswith('dash'):
                    adaptive_formats = self._extract_mpd_formats(
                        stream['url'], display_id, mpd_id=format_id,
                        fatal=False, note=f'Downloading {format_id} MPD manifest')
                for f in adaptive_formats:
                    if f.get('acodec') != 'none':
                        f['language'] = stream_response.get('audio_locale')
                    f['quality'] = hardsub_preference(hardsub_lang.lower())
                formats.extend(adaptive_formats)
        self._sort_formats(formats)

        return {
            'id': internal_id,
            'title': '%s Episode %s – %s' % (
                episode_response.get('season_title'), episode_response.get('episode'), episode_response.get('title')),
            'description': try_get(episode_response, lambda x: x['description'].replace(r'\r\n', '\n')),
            'duration': float_or_none(episode_response.get('duration_ms'), 1000),
            'timestamp': parse_iso8601(episode_response.get('upload_date')),
            'series': episode_response.get('series_title'),
            'series_id': episode_response.get('series_id'),
            'season': episode_response.get('season_title'),
            'season_id': episode_response.get('season_id'),
            'season_number': episode_response.get('season_number'),
            'episode': episode_response.get('title'),
            'episode_number': episode_response.get('sequence_number'),
            'formats': formats,
            'thumbnails': [{
                'url': thumb.get('source'),
                'width': thumb.get('width'),
                'height': thumb.get('height'),
            } for thumb in traverse_obj(episode_response, ('images', 'thumbnail', ..., ...)) or []],
            'subtitles': {
                lang: [{
                    'url': subtitle_data.get('url'),
                    'ext': subtitle_data.get('format')
                }] for lang, subtitle_data in get_streams('subtitles')
            },
        }


class CrunchyrollBetaShowIE(CrunchyrollBetaBaseIE):
    IE_NAME = 'crunchyroll:playlist:beta'
    _VALID_URL = r'https?://beta\.crunchyroll\.com/(?P<lang>(?:\w{2}(?:-\w{2})?/)?)series/(?P<id>\w+)/(?P<display_id>[\w\-]*)/?(?:\?|$)'
    _TESTS = [{
        'url': 'https://beta.crunchyroll.com/series/GY19NQ2QR/Girl-Friend-BETA',
        'info_dict': {
            'id': 'GY19NQ2QR',
            'title': 'Girl Friend BETA',
        },
        'playlist_mincount': 10,
    }, {
        'url': 'https://beta.crunchyroll.com/it/series/GY19NQ2QR/Girl-Friend-BETA',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        lang, internal_id, display_id = self._match_valid_url(url).group('lang', 'id', 'display_id')

        if not self._get_cookies(url).get('etp_rt'):
            return self._redirect_from_beta(url, lang, internal_id, display_id, False, CrunchyrollShowPlaylistIE.ie_key())

        api_domain, bucket, params = self._get_params(lang)

        series_response = self._download_json(
            f'{api_domain}/cms/v2{bucket}/series/{internal_id}', display_id,
            note='Retrieving series metadata', query=params)

        seasons_response = self._download_json(
            f'{api_domain}/cms/v2{bucket}/seasons?series_id={internal_id}', display_id,
            note='Retrieving season list', query=params)

        def entries():
            for season in seasons_response['items']:
                episodes_response = self._download_json(
                    f'{api_domain}/cms/v2{bucket}/episodes?season_id={season["id"]}', display_id,
                    note=f'Retrieving episode list for {season.get("slug_title")}', query=params)
                for episode in episodes_response['items']:
                    episode_id = episode['id']
                    episode_display_id = episode['slug_title']
                    yield {
                        '_type': 'url',
                        'url': f'https://beta.crunchyroll.com/{lang}watch/{episode_id}/{episode_display_id}',
                        'ie_key': CrunchyrollBetaIE.ie_key(),
                        'id': episode_id,
                        'title': '%s Episode %s – %s' % (episode.get('season_title'), episode.get('episode'), episode.get('title')),
                        'description': try_get(episode, lambda x: x['description'].replace(r'\r\n', '\n')),
                        'duration': float_or_none(episode.get('duration_ms'), 1000),
                        'series': episode.get('series_title'),
                        'series_id': episode.get('series_id'),
                        'season': episode.get('season_title'),
                        'season_id': episode.get('season_id'),
                        'season_number': episode.get('season_number'),
                        'episode': episode.get('title'),
                        'episode_number': episode.get('sequence_number')
                    }

        return self.playlist_result(entries(), internal_id, series_response.get('title'))