From 01541feb1250cfbd32c2a181de66a0587ce730f9 Mon Sep 17 00:00:00 2001 From: remitamine Date: Sun, 6 Dec 2015 02:25:46 +0100 Subject: [PATCH] [novamov] simplify info extraction --- youtube_dl/extractor/movshare.py | 4 -- youtube_dl/extractor/novamov.py | 69 +++++++------------------------ youtube_dl/extractor/nowvideo.py | 5 --- youtube_dl/extractor/videoweed.py | 3 -- 4 files changed, 15 insertions(+), 66 deletions(-) diff --git a/youtube_dl/extractor/movshare.py b/youtube_dl/extractor/movshare.py index 6101063f2..3ac83a008 100644 --- a/youtube_dl/extractor/movshare.py +++ b/youtube_dl/extractor/movshare.py @@ -11,10 +11,6 @@ class MovShareIE(NovaMovIE): _HOST = 'www.movshare.net' - _FILE_DELETED_REGEX = r'>This file no longer exists on our servers.<' - _TITLE_REGEX = r'Title: ([^<]+)

' - _DESCRIPTION_REGEX = r'Description: ([^<]+)

' - _TEST = { 'url': 'http://www.movshare.net/video/559e28be54d96', 'md5': 'abd31a2132947262c50429e1d16c1bfd', diff --git a/youtube_dl/extractor/novamov.py b/youtube_dl/extractor/novamov.py index 6163e8855..01959632d 100644 --- a/youtube_dl/extractor/novamov.py +++ b/youtube_dl/extractor/novamov.py @@ -1,15 +1,9 @@ from __future__ import unicode_literals -import re - from .common import InfoExtractor -from ..compat import compat_urlparse from ..utils import ( ExtractorError, - NO_DEFAULT, - encode_dict, - sanitized_Request, - urlencode_postdata, + HEADRequest, ) @@ -22,11 +16,6 @@ class NovaMovIE(InfoExtractor): _HOST = 'www.novamov.com' - _FILE_DELETED_REGEX = r'This file no longer exists on our servers!' - _FILEKEY_REGEX = r'flashvars\.filekey="(?P[^"]+)";' - _TITLE_REGEX = r'(?s)
\s*

([^<]+)

' - _DESCRIPTION_REGEX = r'(?s)
\s*

[^<]+

([^<]+)

' - _TEST = { 'url': 'http://www.novamov.com/video/4rurhn9x446jj', 'md5': '7205f346a52bbeba427603ba10d4b935', @@ -42,53 +31,25 @@ class NovaMovIE(InfoExtractor): def _real_extract(self, url): video_id = self._match_id(url) - url = 'http://%s/video/%s' % (self._HOST, video_id) + video_data = self._download_json( + 'http://%s/mobile/ajax.php?videoId=%s' % (self._HOST, video_id), + video_id, 'Downloading video page') - webpage = self._download_webpage( - url, video_id, 'Downloading video page') + if video_data.get('error'): + raise ExtractorError( + '%s said: The video does not exist or has been deleted.' % self.IE_NAME, + expected=True) - if re.search(self._FILE_DELETED_REGEX, webpage) is not None: - raise ExtractorError('Video %s does not exist' % video_id, expected=True) + video_data = video_data['items'][0] - 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 = 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) - 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, - 'Downloading video api response') - - response = compat_urlparse.parse_qs(api_response) - - if 'error_msg' in response: - raise ExtractorError('%s returned error: %s' % (self.IE_NAME, response['error_msg'][0]), expected=True) - - video_url = response['url'][0] + request = HEADRequest('http://%s/mobile/%s' % (self._HOST, video_data['download'])) + # resolve the url so that we can detect the correct extension + head = self._request_webpage(request, video_id) + video_url = head.geturl() return { 'id': video_id, 'url': video_url, - 'title': title, - 'description': description + 'title': video_data['title'], + 'description': video_data.get('desc'), } diff --git a/youtube_dl/extractor/nowvideo.py b/youtube_dl/extractor/nowvideo.py index 57ee3d366..60d6e18e2 100644 --- a/youtube_dl/extractor/nowvideo.py +++ b/youtube_dl/extractor/nowvideo.py @@ -11,11 +11,6 @@ class NowVideoIE(NovaMovIE): _HOST = 'www.nowvideo.to' - _FILE_DELETED_REGEX = r'>This file no longer exists on our servers.<' - _FILEKEY_REGEX = r'var fkzd="([^"]+)";' - _TITLE_REGEX = r'

([^<]+)

' - _DESCRIPTION_REGEX = r'\s*

([^<]+)

' - _TEST = { 'url': 'http://www.nowvideo.ch/video/0mw0yow7b6dxa', 'md5': 'f8fbbc8add72bd95b7850c6a02fc8817', diff --git a/youtube_dl/extractor/videoweed.py b/youtube_dl/extractor/videoweed.py index ca2e50935..513630947 100644 --- a/youtube_dl/extractor/videoweed.py +++ b/youtube_dl/extractor/videoweed.py @@ -11,9 +11,6 @@ class VideoWeedIE(NovaMovIE): _HOST = 'www.videoweed.es' - _FILE_DELETED_REGEX = r'>This file no longer exists on our servers.<' - _TITLE_REGEX = r'

([^<]+)

' - _TEST = { 'url': 'http://www.videoweed.es/file/b42178afbea14', 'md5': 'abd31a2132947262c50429e1d16c1bfd',