From 82393e2bb2c499cad285beb07c222f501302d830 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Wed, 11 Nov 2015 21:02:05 +0600 Subject: [novamov] Follow continue-to-the-video button if any (Closes #7330) --- youtube_dl/extractor/novamov.py | 40 +++++++++++++++++++++++++++++++++------- 1 file changed, 33 insertions(+), 7 deletions(-) (limited to 'youtube_dl/extractor/novamov.py') diff --git a/youtube_dl/extractor/novamov.py b/youtube_dl/extractor/novamov.py index 04d779890..e0bf6d1bc 100644 --- a/youtube_dl/extractor/novamov.py +++ b/youtube_dl/extractor/novamov.py @@ -4,10 +4,14 @@ import re from .common import InfoExtractor from ..compat import ( + compat_urllib_request, compat_urlparse, ) from ..utils import ( ExtractorError, + NO_DEFAULT, + encode_dict, + urlencode_postdata, ) @@ -41,16 +45,38 @@ class NovaMovIE(InfoExtractor): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') - page = self._download_webpage( - 'http://%s/video/%s' % (self._HOST, video_id), video_id, 'Downloading video page') + url = 'http://%s/video/%s' % (self._HOST, video_id) - if re.search(self._FILE_DELETED_REGEX, page) is not None: - raise ExtractorError('Video %s does not exist' % video_id, expected=True) + webpage = self._download_webpage( + url, video_id, 'Downloading video page') - filekey = self._search_regex(self._FILEKEY_REGEX, page, 'filekey') + if re.search(self._FILE_DELETED_REGEX, webpage) is not None: + raise ExtractorError('Video %s does not exist' % video_id, expected=True) - title = self._html_search_regex(self._TITLE_REGEX, page, 'title', fatal=False) - description = self._html_search_regex(self._DESCRIPTION_REGEX, page, 'description', default='', fatal=False) + def extract_filekey(default=NO_DEFAULT): + return self._search_regex( + self._FILEKEY_REGEX, webpage, 'filekey', default=default) + + filekey = extract_filekey(default=None) + + if not filekey: + fields = self._hidden_inputs(webpage) + post_url = self._search_regex( + r']+action=(["\'])(?P.+?)\1', webpage, + 'post url', default=url, group='url') + if not post_url.startswith('http'): + post_url = compat_urlparse.urljoin(url, post_url) + request = compat_urllib_request.Request( + post_url, urlencode_postdata(encode_dict(fields))) + request.add_header('Content-Type', 'application/x-www-form-urlencoded') + request.add_header('Referer', post_url) + webpage = self._download_webpage( + request, video_id, 'Downloading continue to the video page') + + filekey = extract_filekey() + + title = self._html_search_regex(self._TITLE_REGEX, webpage, 'title', fatal=False) + description = self._html_search_regex(self._DESCRIPTION_REGEX, webpage, 'description', default='', fatal=False) api_response = self._download_webpage( 'http://%s/api/player.api.php?key=%s&file=%s' % (self._HOST, filekey, video_id), video_id, -- cgit v1.2.3 From 699ed30ceed8a0499d17c54742b63469e5ff08ee Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Wed, 11 Nov 2015 22:34:49 +0600 Subject: [novamov] Modernize --- youtube_dl/extractor/novamov.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'youtube_dl/extractor/novamov.py') diff --git a/youtube_dl/extractor/novamov.py b/youtube_dl/extractor/novamov.py index e0bf6d1bc..6b15fc2e5 100644 --- a/youtube_dl/extractor/novamov.py +++ b/youtube_dl/extractor/novamov.py @@ -42,8 +42,7 @@ class NovaMovIE(InfoExtractor): } def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') + video_id = self._match_id(url) url = 'http://%s/video/%s' % (self._HOST, video_id) -- cgit v1.2.3 From 5c2266df4b9aeb7881ed8c026a038e2a25e43734 Mon Sep 17 00:00:00 2001 From: Sergey M? Date: Sat, 21 Nov 2015 22:18:17 +0600 Subject: Switch codebase to use sanitized_Request instead of compat_urllib_request.Request [downloader/dash] Use sanitized_Request [downloader/http] Use sanitized_Request [atresplayer] Use sanitized_Request [bambuser] Use sanitized_Request [bliptv] Use sanitized_Request [brightcove] Use sanitized_Request [cbs] Use sanitized_Request [ceskatelevize] Use sanitized_Request [collegerama] Use sanitized_Request [extractor/common] Use sanitized_Request [crunchyroll] Use sanitized_Request [dailymotion] Use sanitized_Request [dcn] Use sanitized_Request [dramafever] Use sanitized_Request [dumpert] Use sanitized_Request [eitb] Use sanitized_Request [escapist] Use sanitized_Request [everyonesmixtape] Use sanitized_Request [extremetube] Use sanitized_Request [facebook] Use sanitized_Request [fc2] Use sanitized_Request [flickr] Use sanitized_Request [4tube] Use sanitized_Request [gdcvault] Use sanitized_Request [extractor/generic] Use sanitized_Request [hearthisat] Use sanitized_Request [hotnewhiphop] Use sanitized_Request [hypem] Use sanitized_Request [iprima] Use sanitized_Request [ivi] Use sanitized_Request [keezmovies] Use sanitized_Request [letv] Use sanitized_Request [lynda] Use sanitized_Request [metacafe] Use sanitized_Request [minhateca] Use sanitized_Request [miomio] Use sanitized_Request [meovideo] Use sanitized_Request [mofosex] Use sanitized_Request [moniker] Use sanitized_Request [mooshare] Use sanitized_Request [movieclips] Use sanitized_Request [mtv] Use sanitized_Request [myvideo] Use sanitized_Request [neteasemusic] Use sanitized_Request [nfb] Use sanitized_Request [niconico] Use sanitized_Request [noco] Use sanitized_Request [nosvideo] Use sanitized_Request [novamov] Use sanitized_Request [nowness] Use sanitized_Request [nuvid] Use sanitized_Request [played] Use sanitized_Request [pluralsight] Use sanitized_Request [pornhub] Use sanitized_Request [pornotube] Use sanitized_Request [primesharetv] Use sanitized_Request [promptfile] Use sanitized_Request [qqmusic] Use sanitized_Request [rtve] Use sanitized_Request [safari] Use sanitized_Request [sandia] Use sanitized_Request [shared] Use sanitized_Request [sharesix] Use sanitized_Request [sina] Use sanitized_Request [smotri] Use sanitized_Request [sohu] Use sanitized_Request [spankwire] Use sanitized_Request [sportdeutschland] Use sanitized_Request [streamcloud] Use sanitized_Request [streamcz] Use sanitized_Request [tapely] Use sanitized_Request [tube8] Use sanitized_Request [tubitv] Use sanitized_Request [twitch] Use sanitized_Request [twitter] Use sanitized_Request [udemy] Use sanitized_Request [vbox7] Use sanitized_Request [veoh] Use sanitized_Request [vessel] Use sanitized_Request [vevo] Use sanitized_Request [viddler] Use sanitized_Request [videomega] Use sanitized_Request [viewvster] Use sanitized_Request [viki] Use sanitized_Request [vk] Use sanitized_Request [vodlocker] Use sanitized_Request [voicerepublic] Use sanitized_Request [wistia] Use sanitized_Request [xfileshare] Use sanitized_Request [xtube] Use sanitized_Request [xvideos] Use sanitized_Request [yandexmusic] Use sanitized_Request [youku] Use sanitized_Request [youporn] Use sanitized_Request [youtube] Use sanitized_Request [patreon] Use sanitized_Request [extractor/common] Remove unused import [nfb] PEP 8 --- youtube_dl/extractor/novamov.py | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'youtube_dl/extractor/novamov.py') diff --git a/youtube_dl/extractor/novamov.py b/youtube_dl/extractor/novamov.py index 6b15fc2e5..6163e8855 100644 --- a/youtube_dl/extractor/novamov.py +++ b/youtube_dl/extractor/novamov.py @@ -3,14 +3,12 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import ( - compat_urllib_request, - compat_urlparse, -) +from ..compat import compat_urlparse from ..utils import ( ExtractorError, NO_DEFAULT, encode_dict, + sanitized_Request, urlencode_postdata, ) @@ -65,7 +63,7 @@ class NovaMovIE(InfoExtractor): 'post url', default=url, group='url') if not post_url.startswith('http'): post_url = compat_urlparse.urljoin(url, post_url) - request = compat_urllib_request.Request( + request = sanitized_Request( post_url, urlencode_postdata(encode_dict(fields))) request.add_header('Content-Type', 'application/x-www-form-urlencoded') request.add_header('Referer', post_url) -- cgit v1.2.3