X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fextremetube.py;h=3403581fddf08a0928a8e4c5b22e740117646bd2;hb=5c2266df4b9aeb7881ed8c026a038e2a25e43734;hp=0f1eec40f86de4a7ac52507f77398b7d20039de2;hpb=d60703875307f969f84a8907a332adce7835c080;p=youtube-dl diff --git a/youtube_dl/extractor/extremetube.py b/youtube_dl/extractor/extremetube.py index 0f1eec40f..3403581fd 100644 --- a/youtube_dl/extractor/extremetube.py +++ b/youtube_dl/extractor/extremetube.py @@ -1,50 +1,92 @@ -import os +from __future__ import unicode_literals + import re from .common import InfoExtractor from ..utils import ( - compat_urllib_parse_urlparse, - compat_urllib_request, - compat_urllib_parse, + int_or_none, + sanitized_Request, + str_to_int, ) + class ExtremeTubeIE(InfoExtractor): - _VALID_URL = r'^(?:https?://)?(?:www\.)?(?Pextremetube\.com/video/.+?(?P[0-9]+))(?:[/?&]|$)' - _TEST = { - u'url': u'http://www.extremetube.com/video/music-video-14-british-euro-brit-european-cumshots-swallow-652431', - u'file': u'652431.mp4', - u'md5': u'1fb9228f5e3332ec8c057d6ac36f33e0', - u'info_dict': { - u"title": u"Music Video 14 british euro brit european cumshots swallow", - u"uploader": u"unknown", - u"age_limit": 18, + _VALID_URL = r'https?://(?:www\.)?extremetube\.com/(?:[^/]+/)?video/(?P[^/#?&]+)' + _TESTS = [{ + 'url': 'http://www.extremetube.com/video/music-video-14-british-euro-brit-european-cumshots-swallow-652431', + 'md5': '344d0c6d50e2f16b06e49ca011d8ac69', + 'info_dict': { + 'id': 'music-video-14-british-euro-brit-european-cumshots-swallow-652431', + 'ext': 'mp4', + 'title': 'Music Video 14 british euro brit european cumshots swallow', + 'uploader': 'unknown', + 'view_count': int, + 'age_limit': 18, } - } + }, { + 'url': 'http://www.extremetube.com/gay/video/abcde-1234', + 'only_matching': True, + }, { + 'url': 'http://www.extremetube.com/video/latina-slut-fucked-by-fat-black-dick', + 'only_matching': True, + }, { + 'url': 'http://www.extremetube.com/video/652431', + 'only_matching': True, + }] def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('videoid') - url = 'http://www.' + mobj.group('url') + video_id = self._match_id(url) - req = compat_urllib_request.Request(url) + req = sanitized_Request(url) req.add_header('Cookie', 'age_verified=1') webpage = self._download_webpage(req, video_id) - video_title = self._html_search_regex(r'

]*?title="([^"]+)"[^>]*>\1<', webpage, u'title') - uploader = self._html_search_regex(r'>Posted by:(?=<)(?:\s|<[^>]*>)*(.+?)\|', webpage, u'uploader', fatal=False) - video_url = compat_urllib_parse.unquote(self._html_search_regex(r'video_url=(.+?)&', webpage, u'video_url')) - path = compat_urllib_parse_urlparse( video_url ).path - extension = os.path.splitext( path )[1][1:] - format = path.split('/')[5].split('_')[:2] - format = "-".join( format ) + video_title = self._html_search_regex( + r'

]*?title="([^"]+)"[^>]*>', webpage, 'title') + uploader = self._html_search_regex( + r'Uploaded by:\s*\s*(.+?)\s*', + webpage, 'uploader', fatal=False) + view_count = str_to_int(self._html_search_regex( + r'Views:\s*\s*([\d,\.]+)', + webpage, 'view count', fatal=False)) + + flash_vars = self._parse_json( + self._search_regex( + r'var\s+flashvars\s*=\s*({.+?});', webpage, 'flash vars'), + video_id) + + formats = [] + for quality_key, video_url in flash_vars.items(): + height = int_or_none(self._search_regex( + r'quality_(\d+)[pP]$', quality_key, 'height', default=None)) + if not height: + continue + f = { + 'url': 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')) + f.update({ + 'format_id': '%dp-%dk' % (height, bitrate), + 'height': height, + 'tbr': bitrate, + }) + else: + f.update({ + 'format_id': '%dp' % height, + 'height': height, + }) + formats.append(f) + self._sort_formats(formats) return { 'id': video_id, 'title': video_title, + 'formats': formats, 'uploader': uploader, - 'url': video_url, - 'ext': extension, - 'format': format, - 'format_id': format, + 'view_count': view_count, 'age_limit': 18, }