[indavideo] Add support for generic embeds (closes #11989)

pull/16271/merge
Sergey M․ 2018-05-26 01:25:40 +07:00
parent 2a7c6befc1
commit aee36ca832
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
3 changed files with 41 additions and 58 deletions

View File

@ -469,10 +469,7 @@ from .imgur import (
)
from .ina import InaIE
from .inc import IncIE
from .indavideo import (
IndavideoIE,
IndavideoEmbedIE,
)
from .indavideo import IndavideoEmbedIE
from .infoq import InfoQIE
from .instagram import InstagramIE, InstagramUserIE
from .internazionale import InternazionaleIE

View File

@ -109,6 +109,7 @@ from .vice import ViceIE
from .xfileshare import XFileShareIE
from .cloudflarestream import CloudflareStreamIE
from .peertube import PeerTubeIE
from .indavideo import IndavideoEmbedIE
class GenericIE(InfoExtractor):
@ -2022,6 +2023,24 @@ class GenericIE(InfoExtractor):
},
'playlist_count': 2,
},
{
# Indavideo embed
'url': 'https://streetkitchen.hu/receptek/igy_kell_otthon_hamburgert_sutni/',
'info_dict': {
'id': '1693903',
'ext': 'mp4',
'title': 'Így kell otthon hamburgert sütni',
'description': 'md5:f5a730ecf900a5c852e1e00540bbb0f7',
'timestamp': 1426330212,
'upload_date': '20150314',
'uploader': 'StreetKitchen',
'uploader_id': '546363',
},
'add_ie': [IndavideoEmbedIE.ie_key()],
'params': {
'skip_download': True,
},
},
{
'url': 'http://share-videos.se/auto/video/83645793?uid=13',
'md5': 'b68d276de422ab07ee1d49388103f457',
@ -3044,6 +3063,11 @@ class GenericIE(InfoExtractor):
return self.playlist_from_matches(
peertube_urls, video_id, video_title, ie=PeerTubeIE.ie_key())
indavideo_urls = IndavideoEmbedIE._extract_urls(webpage)
if indavideo_urls:
return self.playlist_from_matches(
indavideo_urls, video_id, video_title, ie=IndavideoEmbedIE.ie_key())
sharevideos_urls = [mobj.group('url') for mobj in re.finditer(
r'<iframe[^>]+?\bsrc\s*=\s*(["\'])(?P<url>(?:https?:)?//embed\.share-videos\.se/auto/embed/\d+\?.*?\buid=\d+.*?)\1',
webpage)]

View File

@ -1,6 +1,8 @@
# coding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..compat import compat_str
from ..utils import (
@ -38,6 +40,20 @@ class IndavideoEmbedIE(InfoExtractor):
'only_matching': True,
}]
# Some example URLs covered by generic extractor:
# http://indavideo.hu/video/Vicces_cica_1
# http://index.indavideo.hu/video/2015_0728_beregszasz
# http://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko
# http://erotika.indavideo.hu/video/Amator_tini_punci
# http://film.indavideo.hu/video/f_hrom_nagymamm_volt
# http://palyazat.indavideo.hu/video/Embertelen_dal_Dodgem_egyuttes
@staticmethod
def _extract_urls(webpage):
return re.findall(
r'<iframe[^>]+\bsrc=["\'](?P<url>(?:https?:)?//embed\.indavideo\.hu/player/video/[\da-f]+)',
webpage)
def _real_extract(self, url):
video_id = self._match_id(url)
@ -110,57 +126,3 @@ class IndavideoEmbedIE(InfoExtractor):
'tags': tags,
'formats': formats,
}
class IndavideoIE(InfoExtractor):
_VALID_URL = r'https?://(?:.+?\.)?indavideo\.hu/video/(?P<id>[^/#?]+)'
_TESTS = [{
'url': 'http://indavideo.hu/video/Vicces_cica_1',
'md5': '8c82244ba85d2a2310275b318eb51eac',
'info_dict': {
'id': '1335611',
'display_id': 'Vicces_cica_1',
'ext': 'mp4',
'title': 'Vicces cica',
'description': 'Játszik a tablettel. :D',
'thumbnail': r're:^https?://.*\.jpg$',
'uploader': 'Jet_Pack',
'uploader_id': '491217',
'timestamp': 1390821212,
'upload_date': '20140127',
'duration': 7,
'age_limit': 0,
'tags': list,
},
}, {
'url': 'http://index.indavideo.hu/video/2015_0728_beregszasz',
'only_matching': True,
}, {
'url': 'http://auto.indavideo.hu/video/Sajat_utanfutoban_a_kis_tacsko',
'only_matching': True,
}, {
'url': 'http://erotika.indavideo.hu/video/Amator_tini_punci',
'only_matching': True,
}, {
'url': 'http://film.indavideo.hu/video/f_hrom_nagymamm_volt',
'only_matching': True,
}, {
'url': 'http://palyazat.indavideo.hu/video/Embertelen_dal_Dodgem_egyuttes',
'only_matching': True,
}]
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
embed_url = self._search_regex(
(r'<iframe[^>]+\bsrc=(["\'])(?P<url>(?:https?:)?//embed\.indavideo\.hu/player/video/.+?)\1',
r'<link[^>]+rel="video_src"[^>]+href="(?P<url>.+?)"'),
webpage, 'embed url', group='url')
return {
'_type': 'url_transparent',
'ie_key': 'IndavideoEmbed',
'url': embed_url,
'display_id': display_id,
}