]>
Commit | Line | Data |
---|---|---|
b4e74474 | 1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
f0a6c3d2 | 4 | from ..utils import ( |
9d6458a2 | 5 | ExtractorError, |
e897bd82 | 6 | determine_ext, |
9d6458a2 | 7 | qualities, |
f0a6c3d2 | 8 | ) |
b4e74474 | 9 | |
10 | ||
11 | class TeacherTubeIE(InfoExtractor): | |
df773c3d | 12 | _WORKING = False |
b4e74474 | 13 | IE_NAME = 'teachertube' |
14 | IE_DESC = 'teachertube.com videos' | |
15 | ||
1e07fea2 | 16 | _VALID_URL = r'https?://(?:www\.)?teachertube\.com/(viewVideo\.php\?video_id=|music\.php\?music_id=|video/(?:[\da-z-]+-)?|audio/)(?P<id>\d+)' |
b4e74474 | 17 | |
18 | _TESTS = [{ | |
837b0617 | 19 | # flowplayer |
b4e74474 | 20 | 'url': 'http://www.teachertube.com/viewVideo.php?video_id=339997', |
21 | 'md5': 'f9434ef992fd65936d72999951ee254c', | |
22 | 'info_dict': { | |
23 | 'id': '339997', | |
24 | 'ext': 'mp4', | |
9a2dc4f7 S |
25 | 'title': 'Measures of dispersion from a frequency table', |
26 | 'description': 'Measures of dispersion from a frequency table', | |
837b0617 | 27 | 'thumbnail': r're:https?://.*\.(?:jpg|png)', |
b4e74474 | 28 | }, |
f0a6c3d2 | 29 | }, { |
837b0617 | 30 | # jwplayer |
f0a6c3d2 S |
31 | 'url': 'http://www.teachertube.com/music.php?music_id=8805', |
32 | 'md5': '01e8352006c65757caf7b961f6050e21', | |
33 | 'info_dict': { | |
34 | 'id': '8805', | |
35 | 'ext': 'mp3', | |
36 | 'title': 'PER ASPERA AD ASTRA', | |
9a2dc4f7 | 37 | 'description': 'RADIJSKA EMISIJA ZRAKOPLOVNE TEHNI?KE ?KOLE P', |
f0a6c3d2 | 38 | }, |
1e07fea2 | 39 | }, { |
837b0617 | 40 | # unavailable video |
1e07fea2 | 41 | 'url': 'http://www.teachertube.com/video/intro-video-schleicher-297790', |
837b0617 | 42 | 'only_matching': True, |
b4e74474 | 43 | }] |
44 | ||
45 | def _real_extract(self, url): | |
211503c3 | 46 | video_id = self._match_id(url) |
b4e74474 | 47 | webpage = self._download_webpage(url, video_id) |
48 | ||
9d6458a2 S |
49 | error = self._search_regex( |
50 | r'<div\b[^>]+\bclass=["\']msgBox error[^>]+>([^<]+)', webpage, | |
51 | 'error', default=None) | |
52 | if error: | |
53 | raise ExtractorError('%s said: %s' % (self.IE_NAME, error), expected=True) | |
54 | ||
1a30deca | 55 | title = self._html_search_meta('title', webpage, 'title', fatal=True) |
9a2dc4f7 S |
56 | TITLE_SUFFIX = ' - TeacherTube' |
57 | if title.endswith(TITLE_SUFFIX): | |
58 | title = title[:-len(TITLE_SUFFIX)].strip() | |
59 | ||
60 | description = self._html_search_meta('description', webpage, 'description') | |
61 | if description: | |
62 | description = description.strip() | |
63 | ||
f0a6c3d2 S |
64 | quality = qualities(['mp3', 'flv', 'mp4']) |
65 | ||
9a2dc4f7 S |
66 | media_urls = re.findall(r'data-contenturl="([^"]+)"', webpage) |
67 | media_urls.extend(re.findall(r'var\s+filePath\s*=\s*"([^"]+)"', webpage)) | |
c7df67ed | 68 | media_urls.extend(re.findall(r'\'file\'\s*:\s*["\']([^"\']+)["\'],', webpage)) |
702e5220 | 69 | |
f0a6c3d2 S |
70 | formats = [ |
71 | { | |
72 | 'url': media_url, | |
73 | 'quality': quality(determine_ext(media_url)) | |
702e5220 | 74 | } for media_url in set(media_urls) |
f0a6c3d2 | 75 | ] |
b4e74474 | 76 | |
837b0617 S |
77 | thumbnail = self._og_search_thumbnail( |
78 | webpage, default=None) or self._html_search_meta( | |
79 | 'thumbnail', webpage) | |
80 | ||
b4e74474 | 81 | return { |
82 | 'id': video_id, | |
9a2dc4f7 | 83 | 'title': title, |
9a2dc4f7 | 84 | 'description': description, |
837b0617 S |
85 | 'thumbnail': thumbnail, |
86 | 'formats': formats, | |
b4e74474 | 87 | } |
88 | ||
89 | ||
31a196d7 | 90 | class TeacherTubeUserIE(InfoExtractor): |
df773c3d | 91 | _WORKING = False |
31a196d7 PP |
92 | IE_NAME = 'teachertube:user:collection' |
93 | IE_DESC = 'teachertube.com user and collection videos' | |
b4e74474 | 94 | |
31a196d7 | 95 | _VALID_URL = r'https?://(?:www\.)?teachertube\.com/(user/profile|collection)/(?P<user>[0-9a-zA-Z]+)/?' |
b4e74474 | 96 | |
1a30deca PH |
97 | _MEDIA_RE = r'''(?sx) |
98 | class="?sidebar_thumb_time"?>[0-9:]+</div> | |
99 | \s* | |
100 | <a\s+href="(https?://(?:www\.)?teachertube\.com/(?:video|audio)/[^"]+)" | |
101 | ''' | |
22a6f150 PH |
102 | _TEST = { |
103 | 'url': 'http://www.teachertube.com/user/profile/rbhagwati2', | |
104 | 'info_dict': { | |
105 | 'id': 'rbhagwati2' | |
106 | }, | |
107 | 'playlist_mincount': 179, | |
108 | } | |
7aeb67b3 | 109 | |
b4e74474 | 110 | def _real_extract(self, url): |
5ad28e7f | 111 | mobj = self._match_valid_url(url) |
b4e74474 | 112 | user_id = mobj.group('user') |
113 | ||
31a196d7 PP |
114 | urls = [] |
115 | webpage = self._download_webpage(url, user_id) | |
7aeb67b3 | 116 | urls.extend(re.findall(self._MEDIA_RE, webpage)) |
5f6a1245 | 117 | |
1a30deca | 118 | pages = re.findall(r'/ajax-user/user-videos/%s\?page=([0-9]+)' % user_id, webpage)[:-1] |
31a196d7 PP |
119 | for p in pages: |
120 | more = 'http://www.teachertube.com/ajax-user/user-videos/%s?page=%s' % (user_id, p) | |
1a30deca PH |
121 | webpage = self._download_webpage(more, user_id, 'Downloading page %s/%s' % (p, len(pages))) |
122 | video_urls = re.findall(self._MEDIA_RE, webpage) | |
123 | urls.extend(video_urls) | |
b4e74474 | 124 | |
1a30deca | 125 | entries = [self.url_result(vurl, 'TeacherTube') for vurl in urls] |
b4e74474 | 126 | return self.playlist_result(entries, user_id) |