X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyouporn.py;h=34ab878a4167580b50ef5b3d5bee94e58b02fe41;hb=fd5c4aab5958a2a086072488913cc190ff028bc3;hp=d10ebb0bf93fda6f86c9ac3f272416e1ace1aef7;hpb=589c33dadeec18a9d50713a4a200e3e2d9e297bb;p=youtube-dl diff --git a/youtube_dl/extractor/youporn.py b/youtube_dl/extractor/youporn.py index d10ebb0bf..34ab878a4 100644 --- a/youtube_dl/extractor/youporn.py +++ b/youtube_dl/extractor/youporn.py @@ -3,9 +3,9 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import compat_urllib_request from ..utils import ( int_or_none, + sanitized_Request, str_to_int, unescapeHTML, unified_strdate, @@ -15,31 +15,55 @@ from ..aes import aes_decrypt_text class YouPornIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?youporn\.com/watch/(?P\d+)/(?P[^/?#&]+)' - _TEST = { + _TESTS = [{ 'url': 'http://www.youporn.com/watch/505835/sex-ed-is-it-safe-to-masturbate-daily/', + 'md5': '3744d24c50438cf5b6f6d59feb5055c2', 'info_dict': { 'id': '505835', 'display_id': 'sex-ed-is-it-safe-to-masturbate-daily', 'ext': 'mp4', 'title': 'Sex Ed: Is It Safe To Masturbate Daily?', 'description': 'Love & Sex Answers: http://bit.ly/DanAndJenn -- Is It Unhealthy To Masturbate Daily?', - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', 'uploader': 'Ask Dan And Jennifer', 'upload_date': '20101221', 'average_rating': int, 'view_count': int, + 'comment_count': int, 'categories': list, 'tags': list, 'age_limit': 18, - } - } + }, + }, { + # Unknown uploader + 'url': 'http://www.youporn.com/watch/561726/big-tits-awesome-brunette-on-amazing-webcam-show/?from=related3&al=2&from_id=561726&pos=4', + 'info_dict': { + 'id': '561726', + 'display_id': 'big-tits-awesome-brunette-on-amazing-webcam-show', + 'ext': 'mp4', + 'title': 'Big Tits Awesome Brunette On amazing webcam show', + 'description': 'http://sweetlivegirls.com Big Tits Awesome Brunette On amazing webcam show.mp4', + 'thumbnail': r're:^https?://.*\.jpg$', + 'uploader': 'Unknown', + 'upload_date': '20111125', + 'average_rating': int, + 'view_count': int, + 'comment_count': int, + 'categories': list, + 'tags': list, + 'age_limit': 18, + }, + 'params': { + 'skip_download': True, + }, + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') display_id = mobj.group('display_id') - request = compat_urllib_request.Request(url) + request = sanitized_Request(url) request.add_header('Cookie', 'age_verified=1') webpage = self._download_webpage(request, display_id) @@ -51,7 +75,7 @@ class YouPornIE(InfoExtractor): links = [] sources = self._search_regex( - r'sources\s*:\s*({.+?})', webpage, 'sources', default=None) + r'(?s)sources\s*:\s*({.+?})', webpage, 'sources', default=None) if sources: for _, link in re.findall(r'[^:]+\s*:\s*(["\'])(http.+?)\1', sources): links.append(link) @@ -77,8 +101,9 @@ class YouPornIE(InfoExtractor): } # Video URL's path looks like this: # /201012/17/505835/720p_1500k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4 + # /201012/17/505835/vl_240p_240k_505835/YouPorn%20-%20Sex%20Ed%20Is%20It%20Safe%20To%20Masturbate%20Daily.mp4 # We will benefit from it by extracting some metadata - mobj = re.search(r'/(?P\d{3,4})[pP]_(?P\d+)[kK]_\d+/', video_url) + mobj = re.search(r'(?P\d{3,4})[pP]_(?P\d+)[kK]_\d+/', video_url) if mobj: height = int(mobj.group('height')) bitrate = int(mobj.group('bitrate')) @@ -90,41 +115,42 @@ class YouPornIE(InfoExtractor): formats.append(f) self._sort_formats(formats) - description = self._html_search_regex( - r'(?s)]+class=["\']video-description["\'][^>]*>(.+?)', - webpage, 'description', fatal=False) + description = self._og_search_description(webpage, default=None) thumbnail = self._search_regex( r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P.+?)\1', webpage, 'thumbnail', fatal=False, group='thumbnail') - uploader = self._search_regex( - r']+class=["\']videoInfoBy["\'][^>]*>\s*By:\s*\s*]+href="[^"]*">([^<]+)', + uploader = self._html_search_regex( + r'(?s)]+class=["\']submitByLink["\'][^>]*>(.+?)', webpage, 'uploader', fatal=False) upload_date = unified_strdate(self._html_search_regex( - r'(?s)]+class=["\']videoInfoTime["\'][^>]*>(.+?)', + r'(?s)]+class=["\']videoInfo(?:Date|Time)["\'][^>]*>(.+?)', webpage, 'upload date', fatal=False)) age_limit = self._rta_search(webpage) average_rating = int_or_none(self._search_regex( - r']+class=["\']videoInfoRating["\'][^>]*>\s*]+class=["\']videoRatingPercentage["\'][^>]*>(\d+)%', + r']+class=["\']videoRatingPercentage["\'][^>]*>(\d+)%', webpage, 'average rating', fatal=False)) view_count = str_to_int(self._search_regex( - r'(?s)]+class=["\']videoInfoViews["\'][^>]*>.*?([\d,.]+)\s*', - webpage, 'view count', fatal=False)) - - def extract_tag_box(title): - tag_box = self._search_regex( - (r']+class=["\']tagBoxTitle["\'][^>]*>\s*%s\b.*?\s*' - ']+class=["\']tagBoxContent["\']>(.+?)') % re.escape(title), - webpage, '%s tag box' % title, default=None) + r'(?s)]+class=(["\']).*?\bvideoInfoViews\b.*?\1[^>]*>.*?(?P[\d,.]+)<', + webpage, 'view count', fatal=False, group='count')) + comment_count = str_to_int(self._search_regex( + r'>All [Cc]omments? \(([\d,.]+)\)', + webpage, 'comment count', fatal=False)) + + def extract_tag_box(regex, title): + tag_box = self._search_regex(regex, webpage, title, default=None) if not tag_box: return [] return re.findall(r']+href=[^>]+>([^<]+)', tag_box) - categories = extract_tag_box('Category') - tags = extract_tag_box('Tags') + categories = extract_tag_box( + r'(?s)Categories:.*?]+>(.+?)', 'categories') + tags = extract_tag_box( + r'(?s)Tags:.*?\s*]+class=["\']tagBoxContent["\'][^>]*>(.+?)', + 'tags') return { 'id': video_id, @@ -136,6 +162,7 @@ class YouPornIE(InfoExtractor): 'upload_date': upload_date, 'average_rating': average_rating, 'view_count': view_count, + 'comment_count': comment_count, 'categories': categories, 'tags': tags, 'age_limit': age_limit,