X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fxtube.py;h=8cd3a06877251ac6d8ec75eea7fb3756b4026b37;hb=fae45ede082e50a4935e014b47cbfde925c0f8ed;hp=f9d98b83fb47aef2f927c21c102ced7943c11e53;hpb=d5524947b560c1d0e1dfa2ef7f1969efe07866fa;p=youtube-dl diff --git a/youtube_dl/extractor/xtube.py b/youtube_dl/extractor/xtube.py index f9d98b83f..8cd3a0687 100644 --- a/youtube_dl/extractor/xtube.py +++ b/youtube_dl/extractor/xtube.py @@ -1,20 +1,20 @@ from __future__ import unicode_literals +import itertools import re -import json from .common import InfoExtractor -from ..compat import ( - compat_urllib_request, -) +from ..compat import compat_urllib_parse_unquote from ..utils import ( + int_or_none, parse_duration, + sanitized_Request, str_to_int, ) class XTubeIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?(?Pxtube\.com/watch\.php\?v=(?P[^/?&]+))' + _VALID_URL = r'(?:xtube:|https?://(?:www\.)?xtube\.com/watch\.php\?.*\bv=)(?P[^/?&#]+)' _TEST = { 'url': 'http://www.xtube.com/watch.php?v=kVTUy_G222_', 'md5': '092fbdd3cbe292c920ef6fc6a8a9cdab', @@ -22,7 +22,7 @@ class XTubeIE(InfoExtractor): 'id': 'kVTUy_G222_', 'ext': 'mp4', 'title': 'strange erotica', - 'description': 'http://www.xtube.com an ET kind of thing', + 'description': 'contains:an ET kind of thing', 'uploader': 'greenshowers', 'duration': 450, 'age_limit': 18, @@ -30,41 +30,49 @@ class XTubeIE(InfoExtractor): } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') - url = 'http://www.' + mobj.group('url') + video_id = self._match_id(url) - req = compat_urllib_request.Request(url) + req = sanitized_Request('http://www.xtube.com/watch.php?v=%s' % video_id) req.add_header('Cookie', 'age_verified=1') webpage = self._download_webpage(req, video_id) - video_title = self._html_search_regex(r'

([^<]+)', webpage, 'title') + video_title = self._html_search_regex( + r'

([^<]+)', webpage, 'title') video_uploader = self._html_search_regex( - r'so_s\.addVariable\("owner_u", "([^"]+)', webpage, 'uploader', fatal=False) + [r"var\s+contentOwnerId\s*=\s*'([^']+)", + r'By:\s*([^<]+)', webpage, 'description', fatal=False) + r'

([^<]+)', + webpage, 'description', fatal=False) duration = parse_duration(self._html_search_regex( - r'Runtime: ([^<]+)

', webpage, 'duration', fatal=False)) - view_count = self._html_search_regex( - r'Views: ([\d,\.]+)

', webpage, 'view count', fatal=False) - if view_count: - view_count = str_to_int(view_count) - comment_count = self._html_search_regex( - r'
([\d,\.]+) Comments
', webpage, 'comment count', fatal=False) - if comment_count: - comment_count = str_to_int(comment_count) - - player_quality_option = json.loads(self._html_search_regex( - r'playerQualityOption = ({.+?});', webpage, 'player quality option')) - - QUALITIES = ['3gp', 'mp4_normal', 'mp4_high', 'flv', 'mp4_ultra', 'mp4_720', 'mp4_1080'] - formats = [ - { - 'url': furl, + r'Runtime: ([^<]+)

', + webpage, 'duration', fatal=False)) + view_count = str_to_int(self._html_search_regex( + r'Views: ([\d,\.]+)

', + webpage, 'view count', fatal=False)) + comment_count = str_to_int(self._html_search_regex( + r'
([\d,\.]+) Comments
', + webpage, 'comment count', fatal=False)) + + formats = [] + for format_id, video_url in re.findall( + r'flashvars\.quality_(.+?)\s*=\s*"([^"]+)"', webpage): + fmt = { + 'url': compat_urllib_parse_unquote(video_url), 'format_id': format_id, - 'preference': QUALITIES.index(format_id) if format_id in QUALITIES else -1, - } for format_id, furl in player_quality_option.items() - ] + } + m = re.search(r'^(?P\d+)[pP]', format_id) + if m: + fmt['height'] = int(m.group('height')) + formats.append(fmt) + + if not formats: + video_url = compat_urllib_parse_unquote(self._search_regex( + r'flashvars\.video_url\s*=\s*"([^"]+)"', + webpage, 'video URL')) + formats.append({'url': video_url}) + self._sort_formats(formats) return { @@ -82,43 +90,43 @@ class XTubeIE(InfoExtractor): class XTubeUserIE(InfoExtractor): IE_DESC = 'XTube user profile' - _VALID_URL = r'https?://(?:www\.)?xtube\.com/community/profile\.php\?(.*?)user=(?P[^&#]+)(?:$|[&#])' + _VALID_URL = r'https?://(?:www\.)?xtube\.com/profile/(?P[^/]+-\d+)' _TEST = { - 'url': 'http://www.xtube.com/community/profile.php?user=greenshowers', + 'url': 'http://www.xtube.com/profile/greenshowers-4056496', 'info_dict': { - 'id': 'greenshowers', + 'id': 'greenshowers-4056496', + 'age_limit': 18, }, 'playlist_mincount': 155, } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - username = mobj.group('username') - - profile_page = self._download_webpage( - url, username, note='Retrieving profile page') - - video_count = int(self._search_regex( - r'%s\'s Videos \(([0-9]+)\)' % username, profile_page, - 'video count')) - - PAGE_SIZE = 25 - urls = [] - page_count = (video_count + PAGE_SIZE + 1) // PAGE_SIZE - for n in range(1, page_count + 1): - lpage_url = 'http://www.xtube.com/user_videos.php?page=%d&u=%s' % (n, username) - lpage = self._download_webpage( - lpage_url, username, - note='Downloading page %d/%d' % (n, page_count)) - urls.extend( - re.findall(r'addthis:url="([^"]+)"', lpage)) - - return { - '_type': 'playlist', - 'id': username, - 'entries': [{ - '_type': 'url', - 'url': eurl, - 'ie_key': 'XTube', - } for eurl in urls] - } + user_id = self._match_id(url) + + entries = [] + for pagenum in itertools.count(1): + request = sanitized_Request( + 'http://www.xtube.com/profile/%s/videos/%d' % (user_id, pagenum), + headers={ + 'Cookie': 'popunder=4', + 'X-Requested-With': 'XMLHttpRequest', + 'Referer': url, + }) + + page = self._download_json( + request, user_id, 'Downloading videos JSON page %d' % pagenum) + + html = page.get('html') + if not html: + break + + for _, video_id in re.findall(r'data-plid=(["\'])(.+?)\1', html): + entries.append(self.url_result('xtube:%s' % video_id, XTubeIE.ie_key())) + + page_count = int_or_none(page.get('pageCount')) + if not page_count or pagenum == page_count: + break + + playlist = self.playlist_result(entries, user_id) + playlist['age_limit'] = 18 + return playlist