X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fninegag.py;h=16a02ad7939082627ffb9edd41d7bf6e62fd1f6d;hb=9e1a5b845586a0a5431fb72467142046d8571e6f;hp=3352b833018315821d5964c8e3c907ceb362e21c;hpb=77477fa4c916599e7eaa236a3f3eb5703923cf91;p=youtube-dl diff --git a/youtube_dl/extractor/ninegag.py b/youtube_dl/extractor/ninegag.py index 3352b8330..16a02ad79 100644 --- a/youtube_dl/extractor/ninegag.py +++ b/youtube_dl/extractor/ninegag.py @@ -1,8 +1,10 @@ from __future__ import unicode_literals import re +import json from .common import InfoExtractor +from ..utils import str_to_int class NineGagIE(InfoExtractor): @@ -25,8 +27,7 @@ class NineGagIE(InfoExtractor): "thumbnail": "re:^https?://", }, 'add_ie': ['Youtube'] - }, - { + }, { 'url': 'http://9gag.tv/p/KklwM/alternate-banned-opening-scene-of-gravity?ref=fsidebar', 'info_dict': { 'id': 'KklwM', @@ -44,23 +45,14 @@ class NineGagIE(InfoExtractor): webpage = self._download_webpage(url, display_id) - youtube_id = self._html_search_regex( - r'(?s)id="jsid-video-post-container".*?data-external-id="([^"]+)"', - webpage, 'video ID') - title = self._html_search_regex( - r'(?s)id="jsid-video-post-container".*?data-title="([^"]+)"', - webpage, 'title', default=None) - if not title: - title = self._og_search_title(webpage) - description = self._html_search_regex( - r'(?s)
.*?

(.*?)

', webpage, - 'description', fatal=False) - view_count_str = self._html_search_regex( - r'

([0-9][0-9,]*) views

', webpage, 'view count', - fatal=False) - view_count = ( - None if view_count_str is None - else int(view_count_str.replace(',', ''))) + post_view = json.loads(self._html_search_regex( + r'var postView = new app\.PostView\({\s*post:\s*({.+?}),\s*posts:\s*prefetchedCurrentPost', webpage, 'post view')) + + youtube_id = post_view['videoExternalId'] + title = post_view['title'] + description = post_view['description'] + view_count = str_to_int(post_view['externalView']) + thumbnail = post_view.get('thumbnail_700w') or post_view.get('ogImageUrl') or post_view.get('thumbnail_300w') return { '_type': 'url_transparent', @@ -71,5 +63,5 @@ class NineGagIE(InfoExtractor): 'title': title, 'description': description, 'view_count': view_count, - 'thumbnail': self._og_search_thumbnail(webpage), + 'thumbnail': thumbnail, }