from __future__ import unicode_literals import re from .common import InfoExtractor from ..utils import ( parse_iso8601, str_to_int, ) class CrackedIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?cracked\.com/video_(?P\d+)_[\da-z-]+\.html' _TEST = { 'url': 'http://www.cracked.com/video_19006_4-plot-holes-you-didnt-notice-in-your-favorite-movies.html', 'md5': '4b29a5eeec292cd5eca6388c7558db9e', 'info_dict': { 'id': '19006', 'ext': 'mp4', 'title': '4 Plot Holes You Didn\'t Notice in Your Favorite Movies', 'description': 'md5:3b909e752661db86007d10e5ec2df769', 'timestamp': 1405659600, 'upload_date': '20140718', } } def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') webpage = self._download_webpage(url, video_id) video_url = self._html_search_regex( [r'var\s+CK_vidSrc\s*=\s*"([^"]+)"', r'([^<]+)'], webpage, 'title') description = self._search_regex( r'name="?(?:og:)?description"?\s+content="([^"]+)"', webpage, 'description', default=None) timestamp = self._html_search_regex( r'"date"\s*:\s*"([^"]+)"', webpage, 'upload date', fatal=False) if timestamp: timestamp = parse_iso8601(timestamp[:-6]) view_count = str_to_int(self._html_search_regex( r'([\d,\.]+) Views', webpage, 'view count', fatal=False)) comment_count = str_to_int(self._html_search_regex( r'([\d,\.]+)', webpage, 'comment count', fatal=False)) m = re.search(r'_(?P\d+)X(?P\d+)\.mp4$', video_url) if m: width = int(m.group('width')) height = int(m.group('height')) else: width = height = None return { 'id': video_id, 'url': video_url, 'title': title, 'description': description, 'timestamp': timestamp, 'view_count': view_count, 'comment_count': comment_count, 'height': height, 'width': width, }