# coding: utf-8 from __future__ import unicode_literals from .vimeo import VHXEmbedIE from ..utils import ExtractorError import re class IntlDropoutIE(VHXEmbedIE): IE_NAME = 'intldropout' IE_DESC = 'International Dropout.tv' _NETRC_MACHINE = 'intldropouttv' _LOGIN_URL = 'https://intl.dropout.tv/login' _LOGOUT_URL = 'https://intl.dropout.tv/logout' _VALID_URL = r'https://intl\.dropout\.tv/(?:[^/]+/season:[^/]+/)?videos/(?P.+)' _TESTS = [ { 'url': 'https://intl.dropout.tv/um-actually/season:1/videos/c-3po-s-origins-hp-lovecraft-the-food-album-with-weird-al-yankovic', 'md5': '8beaac579b6ba762f63cd452fd28dcce', 'info_dict': { 'id': '397785', 'ext': 'mp4', 'title': "C-3PO's Origins, HP Lovecraft, the Food Album (with Weird Al Yankovic)", 'thumbnail': r're:^https://vhx.imgix.net/.*\.jpg$', 'description': 'Caldwell Tanner, Siobhan Thompson, and Nate Dern inspect guns and review the Diagon Alley bar scene.', 'upload_date': '20181206', 'timestamp': 1544117975, } }, { 'url': 'https://intl.dropout.tv/videos/um-actually-behind-the-scenes', 'md5': 'b974927cd563423fe50945dbfdbb894c', 'info_dict': { 'id': '397943', 'ext': 'mp4', 'title': 'Um, Actually: Behind the Scenes', 'thumbnail': r're:^https://vhx.imgix.net/.*\.jpg$', 'description': 'What does it take to stump the nerdy? Mike Trapp and team pull back the curtain.', 'upload_date': '20181206', 'timestamp': 1544118409, } } ] def _real_initialize(self): self._login() def _login(self): email, password = self._get_login_info() if (email is None or password is None) and self._downloader.params.get('cookiefile') is None: raise ExtractorError('No login info available, needed for using %s.' % self.IE_NAME, expected=True) self._vhx_login(email, password, self._LOGIN_URL) def _real_extract(self, url): webpage = self._download_webpage(url, None) if "The device limit for your account has been reached" in webpage: raise ExtractorError('Device Limit reached', expected=True) if "Start your free trial" in webpage or "Start Free Trial" in webpage or "Sign in" in webpage: raise ExtractorError('You don\'t seem to be logged in', expected=True) video = self._html_search_regex(r']*"(?Phttps://embed.vhx.tv/videos/[0-9]+[^"]*)"[^>]*>', webpage, 'embed') video_id = self._search_regex(r'https://embed.vhx.tv/videos/(?P[0-9]+)', video, 'id') video_title = self._html_search_regex(r'

]*>(?P[^<]+)<', webpage, 'title', fatal=False) return self.url_result(video, video_id=video_id, video_title=video_title) class IntlDropoutPlaylistIE(IntlDropoutIE): IE_NAME = 'intldropout:playlist' _VALID_URL = r'https://intl\.dropout\.tv/(?P<id>.+)' _TESTS = [ { 'url': 'https://intl.dropout.tv/um-actually-the-web-series', 'md5': 'ebcd26ef54f546225e7cb96e79da31cc', 'playlist_count': 9, 'info_dict': { 'id': 'um-actually-the-web-series', 'title': 'Um, Actually: The Web Series', } }, { 'url': 'https://intl.dropout.tv/new-releases', 'md5': 'ebcd26ef54f546225e7cb96e79da31cc', 'playlist_count': 22, 'info_dict': { 'id': 'new-releases', 'title': 'New Releases', } }, { 'url': 'https://intl.dropout.tv/troopers/season:2', 'md5': 'ebcd26ef54f546225e7cb96e79da31cc', 'playlist_count': 10, 'info_dict': { 'id': 'troopers/season:2', 'title': 'Troopers', } } ] @classmethod def suitable(cls, url): return False if IntlDropoutIE.suitable(url) else super(IntlDropoutPlaylistIE, cls).suitable(url) def _real_extract(self, url): playlist_id = self._match_id(url) webpage = self._download_webpage(url, playlist_id) items = re.findall(r'browse-item-title[^>]+>[^<]*<a href="(?P<url>https://intl.dropout.tv/[^/]+/[^"]+)"', webpage) playlist_title = self._html_search_regex(r'<h1 class="[^"]*collection-title[^"]*"[^>]*>(?P<title>[^<]+)<', webpage, 'title') return self.playlist_from_matches(items, playlist_id=playlist_id, playlist_title=playlist_title)