yt-dlp/yt_dlp/extractor/abematv.py

import base64
import binascii
import functools
import hashlib
import hmac
import io
import json
import re
import struct
import time
import urllib.parse
import urllib.request
import urllib.response
import uuid

from .common import InfoExtractor
from ..aes import aes_ecb_decrypt
from ..utils import (
    ExtractorError,
    bytes_to_intlist,
    decode_base_n,
    int_or_none,
    intlist_to_bytes,
    OnDemandPagedList,
    time_seconds,
    traverse_obj,
    update_url_query,
)

# NOTE: network handler related code is temporary thing until network stack overhaul PRs are merged (#2861/#2862)


def add_opener(ydl, handler):
    ''' Add a handler for opening URLs, like _download_webpage '''
    # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
    # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
    assert isinstance(ydl._opener, urllib.request.OpenerDirector)
    ydl._opener.add_handler(handler)


def remove_opener(ydl, handler):
    '''
    Remove handler(s) for opening URLs
    @param handler Either handler object itself or handler type.
    Specifying handler type will remove all handler which isinstance returns True.
    '''
    # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L426
    # https://github.com/python/cpython/blob/main/Lib/urllib/request.py#L605
    opener = ydl._opener
    assert isinstance(ydl._opener, urllib.request.OpenerDirector)
    if isinstance(handler, (type, tuple)):
        find_cp = lambda x: isinstance(x, handler)
    else:
        find_cp = lambda x: x is handler

    removed = []
    for meth in dir(handler):
        if meth in ["redirect_request", "do_open", "proxy_open"]:
            # oops, coincidental match
            continue

        i = meth.find("_")
        protocol = meth[:i]
        condition = meth[i + 1:]

        if condition.startswith("error"):
            j = condition.find("_") + i + 1
            kind = meth[j + 1:]
            try:
                kind = int(kind)
            except ValueError:
                pass
            lookup = opener.handle_error.get(protocol, {})
            opener.handle_error[protocol] = lookup
        elif condition == "open":
            kind = protocol
            lookup = opener.handle_open
        elif condition == "response":
            kind = protocol
            lookup = opener.process_response
        elif condition == "request":
            kind = protocol
            lookup = opener.process_request
        else:
            continue

        handlers = lookup.setdefault(kind, [])
        if handlers:
            handlers[:] = [x for x in handlers if not find_cp(x)]

        removed.append(x for x in handlers if find_cp(x))

    if removed:
        for x in opener.handlers:
            if find_cp(x):
                x.add_parent(None)
        opener.handlers[:] = [x for x in opener.handlers if not find_cp(x)]


class AbemaLicenseHandler(urllib.request.BaseHandler):
    handler_order = 499
    STRTABLE = '123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz'
    HKEY = b'3AF0298C219469522A313570E8583005A642E73EDD58E3EA2FB7339D3DF1597E'

    def __init__(self, ie: 'AbemaTVIE'):
        # the protocol that this should really handle is 'abematv-license://'
        # abematv_license_open is just a placeholder for development purposes
        # ref. https://github.com/python/cpython/blob/f4c03484da59049eb62a9bf7777b963e2267d187/Lib/urllib/request.py#L510
        setattr(self, 'abematv-license_open', getattr(self, 'abematv_license_open'))
        self.ie = ie

    def _get_videokey_from_ticket(self, ticket):
        to_show = self.ie.get_param('verbose', False)
        media_token = self.ie._get_media_token(to_show=to_show)

        license_response = self.ie._download_json(
            'https://license.abema.io/abematv-hls', None, note='Requesting playback license' if to_show else False,
            query={'t': media_token},
            data=json.dumps({
                'kv': 'a',
                'lt': ticket
            }).encode('utf-8'),
            headers={
                'Content-Type': 'application/json',
            })

        res = decode_base_n(license_response['k'], table=self.STRTABLE)
        encvideokey = bytes_to_intlist(struct.pack('>QQ', res >> 64, res & 0xffffffffffffffff))

        h = hmac.new(
            binascii.unhexlify(self.HKEY),
            (license_response['cid'] + self.ie._DEVICE_ID).encode('utf-8'),
            digestmod=hashlib.sha256)
        enckey = bytes_to_intlist(h.digest())

        return intlist_to_bytes(aes_ecb_decrypt(encvideokey, enckey))

    def abematv_license_open(self, url):
        url = url.get_full_url() if isinstance(url, urllib.request.Request) else url
        ticket = urllib.parse.urlparse(url).netloc
        response_data = self._get_videokey_from_ticket(ticket)
        return urllib.response.addinfourl(io.BytesIO(response_data), headers={
            'Content-Length': len(response_data),
        }, url=url, code=200)


class AbemaTVBaseIE(InfoExtractor):
    _USERTOKEN = None
    _DEVICE_ID = None
    _MEDIATOKEN = None

    _SECRETKEY = b'v+Gjs=25Aw5erR!J8ZuvRrCx*rGswhB&qdHd_SYerEWdU&a?3DzN9BRbp5KwY4hEmcj5#fykMjJ=AuWz5GSMY-d@H7DMEh3M@9n2G552Us$$k9cD=3TxwWe86!x#Zyhe'

    @classmethod
    def _generate_aks(cls, deviceid):
        deviceid = deviceid.encode('utf-8')
        # add 1 hour and then drop minute and secs
        ts_1hour = int((time_seconds() // 3600 + 1) * 3600)
        time_struct = time.gmtime(ts_1hour)
        ts_1hour_str = str(ts_1hour).encode('utf-8')

        tmp = None

        def mix_once(nonce):
            nonlocal tmp
            h = hmac.new(cls._SECRETKEY, digestmod=hashlib.sha256)
            h.update(nonce)
            tmp = h.digest()

        def mix_tmp(count):
            nonlocal tmp
            for i in range(count):
                mix_once(tmp)

        def mix_twist(nonce):
            nonlocal tmp
            mix_once(base64.urlsafe_b64encode(tmp).rstrip(b'=') + nonce)

        mix_once(cls._SECRETKEY)
        mix_tmp(time_struct.tm_mon)
        mix_twist(deviceid)
        mix_tmp(time_struct.tm_mday % 5)
        mix_twist(ts_1hour_str)
        mix_tmp(time_struct.tm_hour % 5)

        return base64.urlsafe_b64encode(tmp).rstrip(b'=').decode('utf-8')

    def _get_device_token(self):
        if self._USERTOKEN:
            return self._USERTOKEN

        username, _ = self._get_login_info()
        AbemaTVBaseIE._USERTOKEN = username and self.cache.load(self._NETRC_MACHINE, username)
        if AbemaTVBaseIE._USERTOKEN:
            # try authentication with locally stored token
            try:
                self._get_media_token(True)
                return
            except ExtractorError as e:
                self.report_warning(f'Failed to login with cached user token; obtaining a fresh one ({e})')

        AbemaTVBaseIE._DEVICE_ID = str(uuid.uuid4())
        aks = self._generate_aks(self._DEVICE_ID)
        user_data = self._download_json(
            'https://api.abema.io/v1/users', None, note='Authorizing',
            data=json.dumps({
                'deviceId': self._DEVICE_ID,
                'applicationKeySecret': aks,
            }).encode('utf-8'),
            headers={
                'Content-Type': 'application/json',
            })
        AbemaTVBaseIE._USERTOKEN = user_data['token']

        # don't allow adding it 2 times or more, though it's guarded
        remove_opener(self._downloader, AbemaLicenseHandler)
        add_opener(self._downloader, AbemaLicenseHandler(self))

        return self._USERTOKEN

    def _get_media_token(self, invalidate=False, to_show=True):
        if not invalidate and self._MEDIATOKEN:
            return self._MEDIATOKEN

        AbemaTVBaseIE._MEDIATOKEN = self._download_json(
            'https://api.abema.io/v1/media/token', None, note='Fetching media token' if to_show else False,
            query={
                'osName': 'android',
                'osVersion': '6.0.1',
                'osLang': 'ja_JP',
                'osTimezone': 'Asia/Tokyo',
                'appId': 'tv.abema',
                'appVersion': '3.27.1'
            }, headers={
                'Authorization': f'bearer {self._get_device_token()}',
            })['token']

        return self._MEDIATOKEN

    def _call_api(self, endpoint, video_id, query=None, note='Downloading JSON metadata'):
        return self._download_json(
            f'https://api.abema.io/{endpoint}', video_id, query=query or {},
            note=note,
            headers={
                'Authorization': f'bearer {self._get_device_token()}',
            })

    def _extract_breadcrumb_list(self, webpage, video_id):
        for jld in re.finditer(
                r'(?is)</span></li></ul><script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
                webpage):
            jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
            if traverse_obj(jsonld, '@type') != 'BreadcrumbList':
                continue
            items = traverse_obj(jsonld, ('itemListElement', ..., 'name'))
            if items:
                return items
        return []


class AbemaTVIE(AbemaTVBaseIE):
    _VALID_URL = r'https?://abema\.tv/(?P<type>now-on-air|video/episode|channels/.+?/slots)/(?P<id>[^?/]+)'
    _NETRC_MACHINE = 'abematv'
    _TESTS = [{
        'url': 'https://abema.tv/video/episode/194-25_s2_p1',
        'info_dict': {
            'id': '194-25_s2_p1',
            'title': '第1話 「チーズケーキ」　「モーニング再び」',
            'series': '異世界食堂２',
            'series_number': 2,
            'episode': '第1話 「チーズケーキ」　「モーニング再び」',
            'episode_number': 1,
        },
        'skip': 'expired',
    }, {
        'url': 'https://abema.tv/channels/anime-live2/slots/E8tvAnMJ7a9a5d',
        'info_dict': {
            'id': 'E8tvAnMJ7a9a5d',
            'title': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
            'series': 'ゆるキャン△ SEASON２',
            'episode': 'ゆるキャン△ SEASON２ 全話一挙【無料ビデオ72時間】',
            'series_number': 2,
            'episode_number': 1,
            'description': 'md5:9c5a3172ae763278f9303922f0ea5b17',
        },
        'skip': 'expired',
    }, {
        'url': 'https://abema.tv/video/episode/87-877_s1282_p31047',
        'info_dict': {
            'id': 'E8tvAnMJ7a9a5d',
            'title': '第5話『光射す』',
            'description': 'md5:56d4fc1b4f7769ded5f923c55bb4695d',
            'thumbnail': r're:https://hayabusa\.io/.+',
            'series': '相棒',
            'episode': '第5話『光射す』',
        },
        'skip': 'expired',
    }, {
        'url': 'https://abema.tv/now-on-air/abema-anime',
        'info_dict': {
            'id': 'abema-anime',
            # this varies
            # 'title': '女子高生の無駄づかい 全話一挙【無料ビデオ72時間】',
            'description': 'md5:55f2e61f46a17e9230802d7bcc913d5f',
            'is_live': True,
        },
        'skip': 'Not supported until yt-dlp implements native live downloader OR AbemaTV can start a local HTTP server',
    }]
    _TIMETABLE = None

    def _perform_login(self, username, password):
        self._get_device_token()
        if self.cache.load(self._NETRC_MACHINE, username) and self._get_media_token():
            self.write_debug('Skipping logging in')
            return

        if '@' in username:  # don't strictly check if it's email address or not
            ep, method = 'user/email', 'email'
        else:
            ep, method = 'oneTimePassword', 'userId'

        login_response = self._download_json(
            f'https://api.abema.io/v1/auth/{ep}', None, note='Logging in',
            data=json.dumps({
                method: username,
                'password': password
            }).encode('utf-8'), headers={
                'Authorization': f'bearer {self._get_device_token()}',
                'Origin': 'https://abema.tv',
                'Referer': 'https://abema.tv/',
                'Content-Type': 'application/json',
            })

        AbemaTVBaseIE._USERTOKEN = login_response['token']
        self._get_media_token(True)
        self.cache.store(self._NETRC_MACHINE, username, AbemaTVBaseIE._USERTOKEN)

    def _real_extract(self, url):
        # starting download using infojson from this extractor is undefined behavior,
        # and never be fixed in the future; you must trigger downloads by directly specifying URL.
        # (unless there's a way to hook before downloading by extractor)
        video_id, video_type = self._match_valid_url(url).group('id', 'type')
        headers = {
            'Authorization': 'Bearer ' + self._get_device_token(),
        }
        video_type = video_type.split('/')[-1]

        webpage = self._download_webpage(url, video_id)
        canonical_url = self._search_regex(
            r'<link\s+rel="canonical"\s*href="(.+?)"', webpage, 'canonical URL',
            default=url)
        info = self._search_json_ld(webpage, video_id, default={})

        title = self._search_regex(
            r'<span\s*class=".+?EpisodeTitleBlock__title">(.+?)</span>', webpage, 'title', default=None)
        if not title:
            jsonld = None
            for jld in re.finditer(
                    r'(?is)<span\s*class="com-m-Thumbnail__image">(?:</span>)?<script[^>]+type=(["\']?)application/ld\+json\1[^>]*>(?P<json_ld>.+?)</script>',
                    webpage):
                jsonld = self._parse_json(jld.group('json_ld'), video_id, fatal=False)
                if jsonld:
                    break
            if jsonld:
                title = jsonld.get('caption')
        if not title and video_type == 'now-on-air':
            if not self._TIMETABLE:
                # cache the timetable because it goes to 5MiB in size (!!)
                self._TIMETABLE = self._download_json(
                    'https://api.abema.io/v1/timetable/dataSet?debug=false', video_id,
                    headers=headers)
            now = time_seconds(hours=9)
            for slot in self._TIMETABLE.get('slots', []):
                if slot.get('channelId') != video_id:
                    continue
                if slot['startAt'] <= now and now < slot['endAt']:
                    title = slot['title']
                    break

        # read breadcrumb on top of page
        breadcrumb = self._extract_breadcrumb_list(webpage, video_id)
        if breadcrumb:
            # breadcrumb list translates to: (e.g. 1st test for this IE)
            # Home > Anime (genre) > Isekai Shokudo 2 (series name) > Episode 1 "Cheese cakes" "Morning again" (episode title)
            # hence this works
            info['series'] = breadcrumb[-2]
            info['episode'] = breadcrumb[-1]
            if not title:
                title = info['episode']

        description = self._html_search_regex(
            (r'<p\s+class="com-video-EpisodeDetailsBlock__content"><span\s+class=".+?">(.+?)</span></p><div',
             r'<span\s+class=".+?SlotSummary.+?">(.+?)</span></div><div',),
            webpage, 'description', default=None, group=1)
        if not description:
            og_desc = self._html_search_meta(
                ('description', 'og:description', 'twitter:description'), webpage)
            if og_desc:
                description = re.sub(r'''(?sx)
                    ^(.+?)(?:
                        アニメの動画を無料で見るならABEMA！| # anime
                        等、.+ # applies for most of categories
                    )?
                ''', r'\1', og_desc)

        # canonical URL may contain series and episode number
        mobj = re.search(r's(\d+)_p(\d+)$', canonical_url)
        if mobj:
            seri = int_or_none(mobj.group(1), default=float('inf'))
            epis = int_or_none(mobj.group(2), default=float('inf'))
            info['series_number'] = seri if seri < 100 else None
            # some anime like Detective Conan (though not available in AbemaTV)
            # has more than 1000 episodes (1026 as of 2021/11/15)
            info['episode_number'] = epis if epis < 2000 else None

        is_live, m3u8_url = False, None
        if video_type == 'now-on-air':
            is_live = True
            channel_url = 'https://api.abema.io/v1/channels'
            if video_id == 'news-global':
                channel_url = update_url_query(channel_url, {'division': '1'})
            onair_channels = self._download_json(channel_url, video_id)
            for ch in onair_channels['channels']:
                if video_id == ch['id']:
                    m3u8_url = ch['playback']['hls']
                    break
            else:
                raise ExtractorError(f'Cannot find on-air {video_id} channel.', expected=True)
        elif video_type == 'episode':
            api_response = self._download_json(
                f'https://api.abema.io/v1/video/programs/{video_id}', video_id,
                note='Checking playability',
                headers=headers)
            ondemand_types = traverse_obj(api_response, ('terms', ..., 'onDemandType'))
            if 3 not in ondemand_types:
                # cannot acquire decryption key for these streams
                self.report_warning('This is a premium-only stream')
            info.update(traverse_obj(api_response, {
                'series': ('series', 'title'),
                'season': ('season', 'title'),
                'season_number': ('season', 'sequence'),
                'episode_number': ('episode', 'number'),
            }))
            if not title:
                title = traverse_obj(api_response, ('episode', 'title'))
            if not description:
                description = traverse_obj(api_response, ('episode', 'content'))

            m3u8_url = f'https://vod-abematv.akamaized.net/program/{video_id}/playlist.m3u8'
        elif video_type == 'slots':
            api_response = self._download_json(
                f'https://api.abema.io/v1/media/slots/{video_id}', video_id,
                note='Checking playability',
                headers=headers)
            if not traverse_obj(api_response, ('slot', 'flags', 'timeshiftFree'), default=False):
                self.report_warning('This is a premium-only stream')

            m3u8_url = f'https://vod-abematv.akamaized.net/slot/{video_id}/playlist.m3u8'
        else:
            raise ExtractorError('Unreachable')

        if is_live:
            self.report_warning("This is a livestream; yt-dlp doesn't support downloading natively, but FFmpeg cannot handle m3u8 manifests from AbemaTV")
            self.report_warning('Please consider using Streamlink to download these streams (https://github.com/streamlink/streamlink)')
        formats = self._extract_m3u8_formats(
            m3u8_url, video_id, ext='mp4', live=is_live)

        info.update({
            'id': video_id,
            'title': title,
            'description': description,
            'formats': formats,
            'is_live': is_live,
        })
        return info


class AbemaTVTitleIE(AbemaTVBaseIE):
    _VALID_URL = r'https?://abema\.tv/video/title/(?P<id>[^?/]+)'
    _PAGE_SIZE = 25

    _TESTS = [{
        'url': 'https://abema.tv/video/title/90-1597',
        'info_dict': {
            'id': '90-1597',
            'title': 'シャッフルアイランド',
        },
        'playlist_mincount': 2,
    }, {
        'url': 'https://abema.tv/video/title/193-132',
        'info_dict': {
            'id': '193-132',
            'title': '真心が届く~僕とスターのオフィス・ラブ!?~',
        },
        'playlist_mincount': 16,
    }, {
        'url': 'https://abema.tv/video/title/25-102',
        'info_dict': {
            'id': '25-102',
            'title': 'ソードアート・オンライン アリシゼーション',
        },
        'playlist_mincount': 24,
    }]

    def _fetch_page(self, playlist_id, series_version, page):
        programs = self._call_api(
            f'v1/video/series/{playlist_id}/programs', playlist_id,
            note=f'Downloading page {page + 1}',
            query={
                'seriesVersion': series_version,
                'offset': str(page * self._PAGE_SIZE),
                'order': 'seq',
                'limit': str(self._PAGE_SIZE),
            })
        yield from (
            self.url_result(f'https://abema.tv/video/episode/{x}')
            for x in traverse_obj(programs, ('programs', ..., 'id')))

    def _entries(self, playlist_id, series_version):
        return OnDemandPagedList(
            functools.partial(self._fetch_page, playlist_id, series_version),
            self._PAGE_SIZE)

    def _real_extract(self, url):
        playlist_id = self._match_id(url)
        series_info = self._call_api(f'v1/video/series/{playlist_id}', playlist_id)

        return self.playlist_result(
            self._entries(playlist_id, series_info['version']), playlist_id=playlist_id,
            playlist_title=series_info.get('title'),
            playlist_description=series_info.get('content'))