1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-11-21 15:45:36 +08:00
youtube-dl/youtube_dl/extractor/ninegag.py

105 lines
3.6 KiB
Python
Raw Normal View History

2014-01-29 01:55:06 +08:00
from __future__ import unicode_literals
2013-12-05 21:29:08 +08:00
import re
from .common import InfoExtractor
2014-04-15 20:49:38 +08:00
from ..utils import str_to_int
2013-12-05 21:29:08 +08:00
class NineGagIE(InfoExtractor):
IE_NAME = '9gag'
2015-09-23 03:58:47 +08:00
_VALID_URL = r'https?://(?:www\.)?9gag(?:\.com/tv|\.tv)/(?:p|embed)/(?P<id>[a-zA-Z0-9]+)(?:/(?P<display_id>[^?#/]+))?'
2013-12-05 21:29:08 +08:00
2014-04-11 07:25:24 +08:00
_TESTS = [{
2015-09-23 03:56:26 +08:00
'url': 'http://9gag.com/tv/p/Kk2X5/people-are-awesome-2013-is-absolutely-awesome',
'info_dict': {
'id': 'Kk2X5',
'ext': 'mp4',
'description': 'This 3-minute video will make you smile and then make you feel untalented and insignificant. Anyway, you should share this awesomeness. (Thanks, Dino!)',
'title': '\"People Are Awesome 2013\" Is Absolutely Awesome',
'uploader_id': 'UCdEH6EjDKwtTe-sO2f0_1XA',
'uploader': 'CompilationChannel',
'upload_date': '20131110',
2015-09-23 03:56:26 +08:00
'view_count': int,
2013-12-05 21:29:08 +08:00
},
2015-09-23 03:55:16 +08:00
'add_ie': ['Youtube'],
}, {
'url': 'http://9gag.com/tv/p/aKolP3',
'info_dict': {
'id': 'aKolP3',
'ext': 'mp4',
'title': 'This Guy Travelled 11 countries In 44 days Just To Make This Amazing Video',
'description': "I just saw more in 1 minute than I've seen in 1 year. This guy's video is epic!!",
'uploader_id': 'rickmereki',
'uploader': 'Rick Mereki',
'upload_date': '20110803',
'view_count': int,
},
'add_ie': ['Vimeo'],
2014-11-24 04:39:15 +08:00
}, {
2015-09-23 03:41:44 +08:00
'url': 'http://9gag.com/tv/p/KklwM',
'only_matching': True,
}, {
'url': 'http://9gag.tv/p/Kk2X5',
'only_matching': True,
2015-09-23 03:58:47 +08:00
}, {
'url': 'http://9gag.com/tv/embed/a5Dmvl',
'only_matching': True,
2014-04-11 07:25:24 +08:00
}]
2015-09-23 03:58:47 +08:00
_EXTERNAL_VIDEO_PROVIDER = {
'1': {
'url': '%s',
'ie_key': 'Youtube',
},
'2': {
'url': 'http://player.vimeo.com/video/%s',
'ie_key': 'Vimeo',
},
'3': {
'url': 'http://instagram.com/p/%s',
'ie_key': 'Instagram',
},
'4': {
'url': 'http://vine.co/v/%s',
'ie_key': 'Vine',
},
}
2013-12-05 21:29:08 +08:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
2015-09-23 03:40:06 +08:00
display_id = mobj.group('display_id') or video_id
2013-12-05 21:29:08 +08:00
2014-04-11 07:25:24 +08:00
webpage = self._download_webpage(url, display_id)
2013-12-05 21:29:08 +08:00
2015-09-23 03:48:13 +08:00
post_view = self._parse_json(
self._search_regex(
r'var\s+postView\s*=\s*new\s+app\.PostView\({\s*post:\s*({.+?})\s*,\s*posts:\s*prefetchedCurrentPost',
webpage, 'post view'),
display_id)
2014-04-15 20:49:38 +08:00
2015-09-23 02:20:18 +08:00
ie_key = None
source_url = post_view.get('sourceUrl')
2015-09-23 03:28:00 +08:00
if not source_url:
2015-09-23 02:20:18 +08:00
external_video_id = post_view['videoExternalId']
external_video_provider = post_view['videoExternalProvider']
source_url = self._EXTERNAL_VIDEO_PROVIDER[external_video_provider]['url'] % external_video_id
ie_key = self._EXTERNAL_VIDEO_PROVIDER[external_video_provider]['ie_key']
2014-04-15 20:49:38 +08:00
title = post_view['title']
2015-09-23 03:46:40 +08:00
description = post_view.get('description')
view_count = str_to_int(post_view.get('externalView'))
2014-04-15 20:49:38 +08:00
thumbnail = post_view.get('thumbnail_700w') or post_view.get('ogImageUrl') or post_view.get('thumbnail_300w')
2013-12-05 21:29:08 +08:00
return {
'_type': 'url_transparent',
2015-09-23 02:20:18 +08:00
'url': source_url,
'ie_key': ie_key,
2013-12-05 21:29:08 +08:00
'id': video_id,
2014-04-11 07:25:24 +08:00
'display_id': display_id,
'title': title,
2014-03-13 23:40:53 +08:00
'description': description,
'view_count': view_count,
2014-04-15 20:49:38 +08:00
'thumbnail': thumbnail,
2013-12-05 21:29:08 +08:00
}