from __future__ import unicode_literals import re from .common import InfoExtractor class BeegIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?beeg\.com/(?P\d+)' _TEST = { 'url': 'http://beeg.com/5416503', 'md5': '634526ae978711f6b748fe0dd6c11f57', 'info_dict': { 'id': '5416503', 'ext': 'mp4', 'title': 'Sultry Striptease', 'description': 'md5:6db3c6177972822aaba18652ff59c773', 'categories': list, # NSFW 'thumbnail': 're:https?://.*\.jpg$', 'age_limit': 18, } } def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') webpage = self._download_webpage(url, video_id) quality_arr = self._search_regex( r'(?s)var\s+qualityArr\s*=\s*{\s*(.+?)\s*}', webpage, 'quality formats') formats = [{ 'url': fmt[1], 'format_id': fmt[0], 'height': int(fmt[0][:-1]), } for fmt in re.findall(r"'([^']+)'\s*:\s*'([^']+)'", quality_arr)] self._sort_formats(formats) title = self._html_search_regex( r'([^<]+)\s*-\s*beeg\.?', webpage, 'title') description = self._html_search_regex( r'