From 6b883e15c8e03da5d280a78e7a56f844876e23d0 Mon Sep 17 00:00:00 2001 From: garret Date: Fri, 1 Nov 2024 16:11:10 +0000 Subject: [PATCH 1/4] [ie/radiko] rough patch to make timefree 30 possible naive approach that assumes things dont do anything unusual --- yt_dlp/extractor/radiko.py | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/yt_dlp/extractor/radiko.py b/yt_dlp/extractor/radiko.py index f94d6a3e7..393f81527 100644 --- a/yt_dlp/extractor/radiko.py +++ b/yt_dlp/extractor/radiko.py @@ -1,4 +1,5 @@ import base64 +import datetime import random import re import urllib.parse @@ -99,17 +100,25 @@ def _extract_full_key(self): self._FULL_KEY = full_key return full_key + def _get_broadcast_day(self, timestring): + dt = datetime.datetime.strptime(timestring, '%Y%m%d%H%M%S') + if dt.hour < 5: + dt -= datetime.timedelta(days=1) + return dt.strftime('%Y%m%d') + def _find_program(self, video_id, station, cursor): + broadcast_day = self._get_broadcast_day(cursor) + station_program = self._download_xml( - f'https://radiko.jp/v3/program/station/weekly/{station}.xml', video_id, - note=f'Downloading radio program for {station} station') + f'https://api.radiko.jp/program/v3/date/{broadcast_day}/station/{station}.xml', station, + note=f'Downloading programme data for {broadcast_day}') prog = None for p in station_program.findall('.//prog'): ft_str, to_str = p.attrib['ft'], p.attrib['to'] ft = unified_timestamp(ft_str, False) to = unified_timestamp(to_str, False) - if ft <= cursor and cursor < to: + if ft_str <= cursor and cursor < to_str: prog = p break if not prog: @@ -187,7 +196,7 @@ def _real_extract(self, url): station, timestring = self._match_valid_url(url).group('station', 'timestring') video_id = join_nonempty(station, timestring) vid_int = unified_timestamp(timestring, False) - prog, station_program, ft, radio_begin, radio_end = self._find_program(video_id, station, vid_int) + prog, station_program, ft, radio_begin, radio_end = self._find_program(video_id, station, timestring) auth_token, area_id = self._auth_client() From 5c4f4e77294a4e00a04bb2ef7547d4cbfe228675 Mon Sep 17 00:00:00 2001 From: garret1317 Date: Thu, 7 Nov 2024 15:20:03 +0000 Subject: [PATCH 2/4] Add error messages when programme is not available --- yt_dlp/extractor/radiko.py | 33 ++++++++++++++++++++++++++++++--- 1 file changed, 30 insertions(+), 3 deletions(-) diff --git a/yt_dlp/extractor/radiko.py b/yt_dlp/extractor/radiko.py index 393f81527..b6d894e12 100644 --- a/yt_dlp/extractor/radiko.py +++ b/yt_dlp/extractor/radiko.py @@ -38,6 +38,19 @@ class RadikoBaseIE(InfoExtractor): 'https://c-radiko.smartstream.ne.jp', ) + _JST = datetime.timezone(datetime.timedelta(hours=9)) + _has_tf30 = None + + def _check_account(self): + if self._has_tf30 is not None: + return self._has_tf30 + if self._get_cookies('https://radiko.jp').get('radiko_session') is None: + return + account_info = self._download_json('https://radiko.jp/ap/member/webapi/v2/member/login/check', + None, note='Checking account status', expected_status=400) + self._has_tf30 = account_info.get('timefreeplus') == '1' + return self._has_tf30 + def _negotiate_token(self): _, auth1_handle = self._download_webpage_handle( 'https://radiko.jp/v2/api/auth1', None, 'Downloading authentication page', @@ -104,14 +117,28 @@ def _get_broadcast_day(self, timestring): dt = datetime.datetime.strptime(timestring, '%Y%m%d%H%M%S') if dt.hour < 5: dt -= datetime.timedelta(days=1) - return dt.strftime('%Y%m%d') + return dt + + def _get_broadcast_day_end(self, dt): + dt += datetime.timedelta(days=1) + return datetime.datetime(dt.year, dt.month, dt.day, 5, 0, 0, tzinfo=self._JST) def _find_program(self, video_id, station, cursor): broadcast_day = self._get_broadcast_day(cursor) + broadcast_day_str = broadcast_day.strftime('%Y%m%d') + + broadcast_day_end = self._get_broadcast_day_end(broadcast_day) + now = datetime.datetime.now(tz=self._JST) + + if broadcast_day_end + datetime.timedelta(days=30) < now: + self.raise_no_formats('Programme is no longer available.', video_id=video_id, expected=True) + elif broadcast_day_end + datetime.timedelta(days=7) < now and not self._check_account(): + self.raise_login_required('Programme is only available with a Timefree 30 subscription', + metadata_available=True) station_program = self._download_xml( - f'https://api.radiko.jp/program/v3/date/{broadcast_day}/station/{station}.xml', station, - note=f'Downloading programme data for {broadcast_day}') + f'https://api.radiko.jp/program/v3/date/{broadcast_day_str}/station/{station}.xml', station, + note=f'Downloading programme data for {broadcast_day_str}') prog = None for p in station_program.findall('.//prog'): From f2821be22bef594d9c77c82d725b6fc98c43e9da Mon Sep 17 00:00:00 2001 From: garret1317 Date: Thu, 7 Nov 2024 17:38:26 +0000 Subject: [PATCH 3/4] Add support for username/password login --- yt_dlp/extractor/radiko.py | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/yt_dlp/extractor/radiko.py b/yt_dlp/extractor/radiko.py index b6d894e12..7efbce7c1 100644 --- a/yt_dlp/extractor/radiko.py +++ b/yt_dlp/extractor/radiko.py @@ -5,6 +5,7 @@ import urllib.parse from .common import InfoExtractor +from ..networking.exceptions import HTTPError from ..utils import ( ExtractorError, clean_html, @@ -13,12 +14,14 @@ try_call, unified_timestamp, update_url_query, + urlencode_postdata, ) from ..utils.traversal import traverse_obj class RadikoBaseIE(InfoExtractor): _GEO_BYPASS = False + _NETRC_MACHINE = 'radiko' _FULL_KEY = None _HOSTS_FOR_TIME_FREE_FFMPEG_UNSUPPORTED = ( 'https://c-rpaa.smartstream.ne.jp', @@ -41,13 +44,23 @@ class RadikoBaseIE(InfoExtractor): _JST = datetime.timezone(datetime.timedelta(hours=9)) _has_tf30 = None + def _perform_login(self, username, password): + try: + login_info = self._download_json('https://radiko.jp/ap/member/webapi/member/login', None, note='Logging in', + data=urlencode_postdata({'mail': username, 'pass': password})) + self._has_tf30 = '2' in login_info.get('privileges') + except ExtractorError as error: + if isinstance(error.cause, HTTPError) and error.cause.status == 401: + raise ExtractorError('Invalid username and/or password', expected=True) + raise + def _check_account(self): if self._has_tf30 is not None: return self._has_tf30 if self._get_cookies('https://radiko.jp').get('radiko_session') is None: return account_info = self._download_json('https://radiko.jp/ap/member/webapi/v2/member/login/check', - None, note='Checking account status', expected_status=400) + None, note='Checking account status from cookies', expected_status=400) self._has_tf30 = account_info.get('timefreeplus') == '1' return self._has_tf30 From 696bf76b8b75bc43ee7a071fb8fc346fb0da8ba7 Mon Sep 17 00:00:00 2001 From: garret1317 Date: Thu, 7 Nov 2024 18:06:52 +0000 Subject: [PATCH 4/4] better words --- yt_dlp/extractor/radiko.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/yt_dlp/extractor/radiko.py b/yt_dlp/extractor/radiko.py index 7efbce7c1..d26edb8a8 100644 --- a/yt_dlp/extractor/radiko.py +++ b/yt_dlp/extractor/radiko.py @@ -54,7 +54,7 @@ def _perform_login(self, username, password): raise ExtractorError('Invalid username and/or password', expected=True) raise - def _check_account(self): + def _check_tf30(self): if self._has_tf30 is not None: return self._has_tf30 if self._get_cookies('https://radiko.jp').get('radiko_session') is None: @@ -145,13 +145,13 @@ def _find_program(self, video_id, station, cursor): if broadcast_day_end + datetime.timedelta(days=30) < now: self.raise_no_formats('Programme is no longer available.', video_id=video_id, expected=True) - elif broadcast_day_end + datetime.timedelta(days=7) < now and not self._check_account(): + elif broadcast_day_end + datetime.timedelta(days=7) < now and not self._check_tf30(): self.raise_login_required('Programme is only available with a Timefree 30 subscription', metadata_available=True) station_program = self._download_xml( f'https://api.radiko.jp/program/v3/date/{broadcast_day_str}/station/{station}.xml', station, - note=f'Downloading programme data for {broadcast_day_str}') + note=f'Downloading programme information for {broadcast_day_str}') prog = None for p in station_program.findall('.//prog'):