]>
Commit | Line | Data |
---|---|---|
db4678e4 | 1 | import re |
2 | ||
e9b86526 | 3 | from .common import InfoExtractor |
38d15ba7 S |
4 | from ..utils import ( |
5 | determine_ext, | |
db4678e4 | 6 | extract_attributes, |
38d15ba7 S |
7 | int_or_none, |
8 | str_to_int, | |
db4678e4 | 9 | url_or_none, |
38d15ba7 S |
10 | urlencode_postdata, |
11 | ) | |
e9b86526 J |
12 | |
13 | ||
14 | class ManyVidsIE(InfoExtractor): | |
efc57145 | 15 | _VALID_URL = r'(?i)https?://(?:www\.)?manyvids\.com/video/(?P<id>\d+)' |
38d15ba7 S |
16 | _TESTS = [{ |
17 | # preview video | |
e9b86526 J |
18 | 'url': 'https://www.manyvids.com/Video/133957/everthing-about-me/', |
19 | 'md5': '03f11bb21c52dd12a05be21a5c7dcc97', | |
20 | 'info_dict': { | |
21 | 'id': '133957', | |
22 | 'ext': 'mp4', | |
efc57145 | 23 | 'title': 'everthing about me (Preview)', |
db4678e4 | 24 | 'uploader': 'ellyxxix', |
efc57145 S |
25 | 'view_count': int, |
26 | 'like_count': int, | |
27 | }, | |
38d15ba7 S |
28 | }, { |
29 | # full video | |
30 | 'url': 'https://www.manyvids.com/Video/935718/MY-FACE-REVEAL/', | |
db4678e4 | 31 | 'md5': 'bb47bab0e0802c2a60c24ef079dfe60f', |
38d15ba7 S |
32 | 'info_dict': { |
33 | 'id': '935718', | |
34 | 'ext': 'mp4', | |
35 | 'title': 'MY FACE REVEAL', | |
db4678e4 | 36 | 'description': 'md5:ec5901d41808b3746fed90face161612', |
37 | 'uploader': 'Sarah Calanthe', | |
38d15ba7 S |
38 | 'view_count': int, |
39 | 'like_count': int, | |
40 | }, | |
41 | }] | |
e9b86526 J |
42 | |
43 | def _real_extract(self, url): | |
e9b86526 | 44 | video_id = self._match_id(url) |
efc57145 | 45 | |
db4678e4 | 46 | real_url = 'https://www.manyvids.com/video/%s/gtm.js' % (video_id, ) |
47 | try: | |
48 | webpage = self._download_webpage(real_url, video_id) | |
49 | except Exception: | |
50 | # probably useless fallback | |
51 | webpage = self._download_webpage(url, video_id) | |
52 | ||
53 | info = self._search_regex( | |
54 | r'''(<div\b[^>]*\bid\s*=\s*(['"])pageMetaDetails\2[^>]*>)''', | |
55 | webpage, 'meta details', default='') | |
56 | info = extract_attributes(info) | |
57 | ||
58 | player = self._search_regex( | |
59 | r'''(<div\b[^>]*\bid\s*=\s*(['"])rmpPlayerStream\2[^>]*>)''', | |
60 | webpage, 'player details', default='') | |
61 | player = extract_attributes(player) | |
62 | ||
63 | video_urls_and_ids = ( | |
64 | (info.get('data-meta-video'), 'video'), | |
65 | (player.get('data-video-transcoded'), 'transcoded'), | |
66 | (player.get('data-video-filepath'), 'filepath'), | |
67 | (self._og_search_video_url(webpage, secure=False, default=None), 'og_video'), | |
68 | ) | |
69 | ||
70 | def txt_or_none(s, default=None): | |
46d09f87 | 71 | return (s.strip() or default) if isinstance(s, str) else default |
db4678e4 | 72 | |
73 | uploader = txt_or_none(info.get('data-meta-author')) | |
e9b86526 | 74 | |
db4678e4 | 75 | def mung_title(s): |
76 | if uploader: | |
77 | s = re.sub(r'^\s*%s\s+[|-]' % (re.escape(uploader), ), '', s) | |
78 | return txt_or_none(s) | |
efc57145 | 79 | |
db4678e4 | 80 | title = ( |
81 | mung_title(info.get('data-meta-title')) | |
82 | or self._html_search_regex( | |
83 | (r'<span[^>]+class=["\']item-title[^>]+>([^<]+)', | |
84 | r'<h2[^>]+class=["\']h2 m-0["\'][^>]*>([^<]+)'), | |
85 | webpage, 'title', default=None) | |
86 | or self._html_search_meta( | |
87 | 'twitter:title', webpage, 'title', fatal=True)) | |
88 | ||
89 | title = re.sub(r'\s*[|-]\s+ManyVids\s*$', '', title) or title | |
38d15ba7 S |
90 | |
91 | if any(p in webpage for p in ('preview_videos', '_preview.mp4')): | |
92 | title += ' (Preview)' | |
93 | ||
94 | mv_token = self._search_regex( | |
95 | r'data-mvtoken=(["\'])(?P<value>(?:(?!\1).)+)\1', webpage, | |
96 | 'mv token', default=None, group='value') | |
97 | ||
98 | if mv_token: | |
99 | # Sets some cookies | |
100 | self._download_webpage( | |
101 | 'https://www.manyvids.com/includes/ajax_repository/you_had_me_at_hello.php', | |
db4678e4 | 102 | video_id, note='Setting format cookies', fatal=False, |
103 | data=urlencode_postdata({ | |
38d15ba7 S |
104 | 'mvtoken': mv_token, |
105 | 'vid': video_id, | |
106 | }), headers={ | |
107 | 'Referer': url, | |
108 | 'X-Requested-With': 'XMLHttpRequest' | |
109 | }) | |
110 | ||
db4678e4 | 111 | formats = [] |
112 | for v_url, fmt in video_urls_and_ids: | |
113 | v_url = url_or_none(v_url) | |
114 | if not v_url: | |
115 | continue | |
116 | if determine_ext(v_url) == 'm3u8': | |
117 | formats.extend(self._extract_m3u8_formats( | |
118 | v_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
119 | m3u8_id='hls')) | |
120 | else: | |
121 | formats.append({ | |
122 | 'url': v_url, | |
123 | 'format_id': fmt, | |
124 | }) | |
125 | ||
126 | self._remove_duplicate_formats(formats) | |
127 | ||
128 | for f in formats: | |
129 | if f.get('height') is None: | |
130 | f['height'] = int_or_none( | |
131 | self._search_regex(r'_(\d{2,3}[02468])_', f['url'], 'video height', default=None)) | |
132 | if '/preview/' in f['url']: | |
133 | f['format_id'] = '_'.join(filter(None, (f.get('format_id'), 'preview'))) | |
134 | f['preference'] = -10 | |
135 | if 'transcoded' in f['format_id']: | |
136 | f['preference'] = f.get('preference', -1) - 1 | |
137 | ||
db4678e4 | 138 | def get_likes(): |
139 | likes = self._search_regex( | |
140 | r'''(<a\b[^>]*\bdata-id\s*=\s*(['"])%s\2[^>]*>)''' % (video_id, ), | |
141 | webpage, 'likes', default='') | |
142 | likes = extract_attributes(likes) | |
143 | return int_or_none(likes.get('data-likes')) | |
efc57145 | 144 | |
db4678e4 | 145 | def get_views(): |
146 | return str_to_int(self._html_search_regex( | |
147 | r'''(?s)<span\b[^>]*\bclass\s*=["']views-wrapper\b[^>]+>.+?<span\b[^>]+>\s*(\d[\d,.]*)\s*</span>''', | |
148 | webpage, 'view count', default=None)) | |
efc57145 | 149 | |
e9b86526 J |
150 | return { |
151 | 'id': video_id, | |
152 | 'title': title, | |
38d15ba7 | 153 | 'formats': formats, |
db4678e4 | 154 | 'description': txt_or_none(info.get('data-meta-description')), |
155 | 'uploader': txt_or_none(info.get('data-meta-author')), | |
156 | 'thumbnail': ( | |
157 | url_or_none(info.get('data-meta-image')) | |
158 | or url_or_none(player.get('data-video-screenshot'))), | |
159 | 'view_count': get_views(), | |
160 | 'like_count': get_likes(), | |
e9b86526 | 161 | } |