yt-dlp/yt_dlp/extractor/iwara.py

121 lines
4.2 KiB
Python
Raw Normal View History

import re
from .common import InfoExtractor
from ..compat import compat_urllib_parse_urlparse
2017-02-05 14:48:13 +01:00
from ..utils import (
int_or_none,
mimetype2ext,
remove_end,
2018-07-21 13:02:41 +02:00
url_or_none,
unified_strdate,
strip_or_none,
2017-02-05 14:48:13 +01:00
)
class IwaraIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.|ecchi\.)?iwara\.tv/videos/(?P<id>[a-zA-Z0-9]+)'
_TESTS = [{
'url': 'http://iwara.tv/videos/amVwUl1EHpAD9RD',
2017-02-05 14:48:13 +01:00
# md5 is unstable
'info_dict': {
'id': 'amVwUl1EHpAD9RD',
'ext': 'mp4',
'title': '【MMD R-18】ガールフレンド carry_me_off',
'age_limit': 18,
'thumbnail': 'https://i.iwara.tv/sites/default/files/videos/thumbnails/7951/thumbnail-7951_0001.png',
'uploader': 'Reimu丨Action',
'upload_date': '20150828',
'description': 'md5:1d4905ce48c66c9299c617f08e106e0f',
},
}, {
'url': 'http://ecchi.iwara.tv/videos/Vb4yf2yZspkzkBO',
'md5': '7e5f1f359cd51a027ba4a7b7710a50f0',
'info_dict': {
'id': '0B1LvuHnL-sRFNXB1WHNqbGw4SXc',
'ext': 'mp4',
2017-02-05 14:48:13 +01:00
'title': '[3D Hentai] Kyonyu × Genkai × Emaki Shinobi Girls.mp4',
'age_limit': 18,
},
'add_ie': ['GoogleDrive'],
}, {
'url': 'http://www.iwara.tv/videos/nawkaumd6ilezzgq',
2017-02-05 14:48:13 +01:00
# md5 is unstable
'info_dict': {
'id': '6liAP9s2Ojc',
'ext': 'mp4',
2017-02-05 14:48:13 +01:00
'age_limit': 18,
'title': '[MMD] Do It Again Ver.2 [1080p 60FPS] (Motion,Camera,Wav+DL)',
'description': 'md5:590c12c0df1443d833fbebe05da8c47a',
'upload_date': '20160910',
'uploader': 'aMMDsork',
'uploader_id': 'UCVOFyOSCyFkXTYYHITtqB7A',
},
'add_ie': ['Youtube'],
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage, urlh = self._download_webpage_handle(url, video_id)
hostname = compat_urllib_parse_urlparse(urlh.geturl()).hostname
# ecchi is 'sexy' in Japanese
age_limit = 18 if hostname.split('.')[0] == 'ecchi' else 0
2017-02-05 14:48:13 +01:00
video_data = self._download_json('http://www.iwara.tv/api/video/%s' % video_id, video_id)
2017-02-05 14:48:13 +01:00
if not video_data:
iframe_url = self._html_search_regex(
r'<iframe[^>]+src=([\'"])(?P<url>[^\'"]+)\1',
webpage, 'iframe URL', group='url')
return {
'_type': 'url_transparent',
'url': iframe_url,
'age_limit': age_limit,
}
2022-04-04 10:27:35 +02:00
title = remove_end(self._html_extract_title(webpage), ' | Iwara')
thumbnail = self._html_search_regex(
r'poster=[\'"]([^\'"]+)', webpage, 'thumbnail', default=None)
uploader = self._html_search_regex(
r'class="username">([^<]+)', webpage, 'uploader', fatal=False)
upload_date = unified_strdate(self._html_search_regex(
r'作成日:([^\s]+)', webpage, 'upload_date', fatal=False))
description = strip_or_none(self._search_regex(
r'<p>(.+?(?=</div))', webpage, 'description', fatal=False,
flags=re.DOTALL))
2017-02-05 14:48:13 +01:00
formats = []
for a_format in video_data:
2018-07-21 13:02:41 +02:00
format_uri = url_or_none(a_format.get('uri'))
if not format_uri:
continue
2017-02-05 14:48:13 +01:00
format_id = a_format.get('resolution')
height = int_or_none(self._search_regex(
r'(\d+)p', format_id, 'height', default=None))
formats.append({
2018-07-21 13:02:41 +02:00
'url': self._proto_relative_url(format_uri, 'https:'),
2017-02-05 14:48:13 +01:00
'format_id': format_id,
'ext': mimetype2ext(a_format.get('mime')) or 'mp4',
'height': height,
'width': int_or_none(height / 9.0 * 16.0 if height else None),
2017-02-05 14:48:13 +01:00
'quality': 1 if format_id == 'Source' else 0,
})
self._sort_formats(formats)
return {
'id': video_id,
'title': title,
'age_limit': age_limit,
2017-02-05 14:48:13 +01:00
'formats': formats,
'thumbnail': self._proto_relative_url(thumbnail, 'https:'),
'uploader': uploader,
'upload_date': upload_date,
'description': description,
2017-02-05 14:48:13 +01:00
}