]>
Commit | Line | Data |
---|---|---|
001a5fd3 YCH |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
8eb7ba82 | 3 | import re |
001a5fd3 YCH |
4 | |
5 | from .common import InfoExtractor | |
6 | from ..compat import compat_urllib_parse_urlparse | |
caf0f5f8 YCH |
7 | from ..utils import ( |
8 | int_or_none, | |
9 | mimetype2ext, | |
10 | remove_end, | |
4ecf300d | 11 | url_or_none, |
8eb7ba82 B |
12 | unified_strdate, |
13 | strip_or_none, | |
caf0f5f8 | 14 | ) |
001a5fd3 YCH |
15 | |
16 | ||
17 | class IwaraIE(InfoExtractor): | |
18 | _VALID_URL = r'https?://(?:www\.|ecchi\.)?iwara\.tv/videos/(?P<id>[a-zA-Z0-9]+)' | |
19 | _TESTS = [{ | |
20 | 'url': 'http://iwara.tv/videos/amVwUl1EHpAD9RD', | |
caf0f5f8 | 21 | # md5 is unstable |
001a5fd3 YCH |
22 | 'info_dict': { |
23 | 'id': 'amVwUl1EHpAD9RD', | |
24 | 'ext': 'mp4', | |
25 | 'title': '【MMD R-18】ガールフレンド carry_me_off', | |
26 | 'age_limit': 18, | |
8eb7ba82 B |
27 | 'thumbnail': 'https://i.iwara.tv/sites/default/files/videos/thumbnails/7951/thumbnail-7951_0001.png', |
28 | 'uploader': 'Reimu丨Action', | |
29 | 'upload_date': '20150828', | |
30 | 'description': 'md5:1d4905ce48c66c9299c617f08e106e0f', | |
001a5fd3 YCH |
31 | }, |
32 | }, { | |
33 | 'url': 'http://ecchi.iwara.tv/videos/Vb4yf2yZspkzkBO', | |
34 | 'md5': '7e5f1f359cd51a027ba4a7b7710a50f0', | |
35 | 'info_dict': { | |
36 | 'id': '0B1LvuHnL-sRFNXB1WHNqbGw4SXc', | |
37 | 'ext': 'mp4', | |
caf0f5f8 | 38 | 'title': '[3D Hentai] Kyonyu × Genkai × Emaki Shinobi Girls.mp4', |
001a5fd3 YCH |
39 | 'age_limit': 18, |
40 | }, | |
41 | 'add_ie': ['GoogleDrive'], | |
42 | }, { | |
43 | 'url': 'http://www.iwara.tv/videos/nawkaumd6ilezzgq', | |
caf0f5f8 | 44 | # md5 is unstable |
001a5fd3 YCH |
45 | 'info_dict': { |
46 | 'id': '6liAP9s2Ojc', | |
47 | 'ext': 'mp4', | |
caf0f5f8 | 48 | 'age_limit': 18, |
001a5fd3 YCH |
49 | 'title': '[MMD] Do It Again Ver.2 [1080p 60FPS] (Motion,Camera,Wav+DL)', |
50 | 'description': 'md5:590c12c0df1443d833fbebe05da8c47a', | |
51 | 'upload_date': '20160910', | |
52 | 'uploader': 'aMMDsork', | |
53 | 'uploader_id': 'UCVOFyOSCyFkXTYYHITtqB7A', | |
54 | }, | |
55 | 'add_ie': ['Youtube'], | |
56 | }] | |
57 | ||
58 | def _real_extract(self, url): | |
59 | video_id = self._match_id(url) | |
60 | ||
61 | webpage, urlh = self._download_webpage_handle(url, video_id) | |
62 | ||
63 | hostname = compat_urllib_parse_urlparse(urlh.geturl()).hostname | |
64 | # ecchi is 'sexy' in Japanese | |
65 | age_limit = 18 if hostname.split('.')[0] == 'ecchi' else 0 | |
66 | ||
caf0f5f8 | 67 | video_data = self._download_json('http://www.iwara.tv/api/video/%s' % video_id, video_id) |
001a5fd3 | 68 | |
caf0f5f8 | 69 | if not video_data: |
001a5fd3 YCH |
70 | iframe_url = self._html_search_regex( |
71 | r'<iframe[^>]+src=([\'"])(?P<url>[^\'"]+)\1', | |
72 | webpage, 'iframe URL', group='url') | |
73 | return { | |
74 | '_type': 'url_transparent', | |
75 | 'url': iframe_url, | |
76 | 'age_limit': age_limit, | |
77 | } | |
78 | ||
79 | title = remove_end(self._html_search_regex( | |
80 | r'<title>([^<]+)</title>', webpage, 'title'), ' | Iwara') | |
81 | ||
63b1ad0f | 82 | thumbnail = self._html_search_regex( |
8eb7ba82 B |
83 | r'poster=[\'"]([^\'"]+)', webpage, 'thumbnail', default=None) |
84 | ||
85 | uploader = self._html_search_regex( | |
86 | r'class="username">([^<]+)', webpage, 'uploader', fatal=False) | |
87 | ||
88 | upload_date = unified_strdate(self._html_search_regex( | |
89 | r'作成日:([^\s]+)', webpage, 'upload_date', fatal=False)) | |
90 | ||
91 | description = strip_or_none(self._search_regex( | |
92 | r'<p>(.+?(?=</div))', webpage, 'description', fatal=False, | |
93 | flags=re.DOTALL)) | |
63b1ad0f | 94 | |
caf0f5f8 YCH |
95 | formats = [] |
96 | for a_format in video_data: | |
4ecf300d S |
97 | format_uri = url_or_none(a_format.get('uri')) |
98 | if not format_uri: | |
99 | continue | |
caf0f5f8 YCH |
100 | format_id = a_format.get('resolution') |
101 | height = int_or_none(self._search_regex( | |
102 | r'(\d+)p', format_id, 'height', default=None)) | |
103 | formats.append({ | |
4ecf300d | 104 | 'url': self._proto_relative_url(format_uri, 'https:'), |
caf0f5f8 YCH |
105 | 'format_id': format_id, |
106 | 'ext': mimetype2ext(a_format.get('mime')) or 'mp4', | |
107 | 'height': height, | |
2ab2c0d1 | 108 | 'width': int_or_none(height / 9.0 * 16.0 if height else None), |
caf0f5f8 YCH |
109 | 'quality': 1 if format_id == 'Source' else 0, |
110 | }) | |
111 | ||
112 | self._sort_formats(formats) | |
113 | ||
114 | return { | |
001a5fd3 YCH |
115 | 'id': video_id, |
116 | 'title': title, | |
117 | 'age_limit': age_limit, | |
caf0f5f8 | 118 | 'formats': formats, |
63b1ad0f | 119 | 'thumbnail': self._proto_relative_url(thumbnail, 'https:'), |
8eb7ba82 B |
120 | 'uploader': uploader, |
121 | 'upload_date': upload_date, | |
122 | 'description': description, | |
caf0f5f8 | 123 | } |