archive
/
yt-dlp
mirror of https://github.com/yt-dlp/yt-dlp
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[globo] Separate article extractor

Browse Source
pull/8/head
Sergey M․ 9 years ago
parent 236cb2131b
commit ad607563a2
2 changed files with 79 additions and 66 deletions
Show Stats Download Patch File Download Diff File

5
youtube_dl/extractor/__init__.py
Unescape Escape View File

@ -212,7 +212,10 @@ from .gfycat import GfycatIE
from .giantbomb import GiantBombIE from .giantbomb import GiantBombIE
from .giga import GigaIE from .giga import GigaIE
from .glide import GlideIE from .glide import GlideIE
from .globo import GloboIE from .globo import (
GloboIE,
GloboArticleIE,
)
from .godtube import GodTubeIE from .godtube import GodTubeIE
from .goldenmoustache import GoldenMoustacheIE from .goldenmoustache import GoldenMoustacheIE
from .golem import GolemIE from .golem import GolemIE

140
youtube_dl/extractor/globo.py
Unescape Escape View File

@ -18,75 +18,52 @@ from ..utils import (
class GloboIE(InfoExtractor): class GloboIE(InfoExtractor):
_VALID_URL = 'https?://.+?\.globo\.com/(?P<id>.+)' _VALID_URL = '(?:globo:|https?://.+?\.globo\.com/(?:[^/]+/)*(?:v/(?:[^/]+/)?|videos/))(?P<id>\d{7,})'
_API_URL_TEMPLATE = 'http://api.globovideos.com/videos/%s/playlist' _API_URL_TEMPLATE = 'http://api.globovideos.com/videos/%s/playlist'
_SECURITY_URL_TEMPLATE = 'http://security.video.globo.com/videos/%s/hash?player=flash&version=17.0.0.132&resource_id=%s' _SECURITY_URL_TEMPLATE = 'http://security.video.globo.com/videos/%s/hash?player=flash&version=17.0.0.132&resource_id=%s'
_VIDEOID_REGEXES = [
r'\bdata-video-id="(\d+)"',
r'\bdata-player-videosids="(\d+)"',
r'<div[^>]+\bid="(\d+)"',
]
_RESIGN_EXPIRATION = 86400 _RESIGN_EXPIRATION = 86400
_TESTS = [ _TESTS = [{
{ 'url': 'http://globotv.globo.com/sportv/futebol-nacional/v/os-gols-de-atletico-mg-3-x-2-santos-pela-24a-rodada-do-brasileirao/3654973/',
'url': 'http://globotv.globo.com/sportv/futebol-nacional/v/os-gols-de-atletico-mg-3-x-2-santos-pela-24a-rodada-do-brasileirao/3654973/', 'md5': '03ebf41cb7ade43581608b7d9b71fab0',
'md5': '03ebf41cb7ade43581608b7d9b71fab0', 'info_dict': {
'info_dict': { 'id': '3654973',
'id': '3654973', 'ext': 'mp4',
'ext': 'mp4', 'title': 'Os gols de Atlético-MG 3 x 2 Santos pela 24ª rodada do Brasileirão',
'title': 'Os gols de Atlético-MG 3 x 2 Santos pela 24ª rodada do Brasileirão', 'duration': 251.585,
'duration': 251.585, 'uploader': 'SporTV',
'uploader': 'SporTV', 'uploader_id': 698,
'uploader_id': 698, 'like_count': int,
'like_count': int, }
} }, {
}, 'url': 'http://g1.globo.com/carros/autoesporte/videos/t/exclusivos-do-g1/v/mercedes-benz-gla-passa-por-teste-de-colisao-na-europa/3607726/',
{ 'md5': 'b3ccc801f75cd04a914d51dadb83a78d',
'url': 'http://g1.globo.com/carros/autoesporte/videos/t/exclusivos-do-g1/v/mercedes-benz-gla-passa-por-teste-de-colisao-na-europa/3607726/', 'info_dict': {
'md5': 'b3ccc801f75cd04a914d51dadb83a78d', 'id': '3607726',
'info_dict': { 'ext': 'mp4',
'id': '3607726', 'title': 'Mercedes-Benz GLA passa por teste de colisão na Europa',
'ext': 'mp4', 'duration': 103.204,
'title': 'Mercedes-Benz GLA passa por teste de colisão na Europa', 'uploader': 'Globo.com',
'duration': 103.204, 'uploader_id': 265,
'uploader': 'Globo.com', 'like_count': int,
'uploader_id': 265, }
'like_count': int, }, {
} 'url': 'http://globotv.globo.com/canal-brasil/sangue-latino/t/todos-os-videos/v/ator-e-diretor-argentino-ricado-darin-fala-sobre-utopias-e-suas-perdas/3928201/',
}, 'md5': 'c1defca721ce25b2354e927d3e4b3dec',
{ 'info_dict': {
'url': 'http://g1.globo.com/jornal-nacional/noticia/2014/09/novidade-na-fiscalizacao-de-bagagem-pela-receita-provoca-discussoes.html', 'id': '3928201',
'md5': '307fdeae4390ccfe6ba1aa198cf6e72b', 'ext': 'mp4',
'info_dict': { 'title': 'Ator e diretor argentino, Ricado Darín fala sobre utopias e suas perdas',
'id': '3652183', 'duration': 1472.906,
'ext': 'mp4', 'uploader': 'Canal Brasil',
'title': 'Receita Federal explica como vai fiscalizar bagagens de quem retorna ao Brasil de avião', 'uploader_id': 705,
'duration': 110.711, 'like_count': int,
'uploader': 'Rede Globo', }
'uploader_id': 196, }]
'like_count': int,
}
},
{
'url': 'http://globotv.globo.com/canal-brasil/sangue-latino/t/todos-os-videos/v/ator-e-diretor-argentino-ricado-darin-fala-sobre-utopias-e-suas-perdas/3928201/',
'md5': 'c1defca721ce25b2354e927d3e4b3dec',
'info_dict': {
'id': '3928201',
'ext': 'mp4',
'title': 'Ator e diretor argentino, Ricado Darín fala sobre utopias e suas perdas',
'duration': 1472.906,
'uploader': 'Canal Brasil',
'uploader_id': 705,
'like_count': int,
}
},
]
class MD5(): class MD5:
HEX_FORMAT_LOWERCASE = 0 HEX_FORMAT_LOWERCASE = 0
HEX_FORMAT_UPPERCASE = 1 HEX_FORMAT_UPPERCASE = 1
BASE64_PAD_CHARACTER_DEFAULT_COMPLIANCE = '' BASE64_PAD_CHARACTER_DEFAULT_COMPLIANCE = ''
@ -353,9 +330,6 @@ class GloboIE(InfoExtractor):
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
video_id = self._search_regex(self._VIDEOID_REGEXES, webpage, 'video id')
video = self._download_json( video = self._download_json(
self._API_URL_TEMPLATE % video_id, video_id)['videos'][0] self._API_URL_TEMPLATE % video_id, video_id)['videos'][0]
@ -417,3 +391,39 @@ class GloboIE(InfoExtractor):
'like_count': like_count, 'like_count': like_count,
'formats': formats 'formats': formats
} }
class GloboArticleIE(InfoExtractor):
_VALID_URL = 'https?://.+?\.globo\.com/(?:[^/]+/)*(?P<id>[^/]+)\.html'
_VIDEOID_REGEXES = [
r'\bdata-video-id=["\'](\d{7,})',
r'\bdata-player-videosids=["\'](\d{7,})',
r'\bvideosIDs\s*:\s*["\'](\d{7,})',
r'\bdata-id=["\'](\d{7,})',
r'<div[^>]+\bid=["\'](\d{7,})',
]
_TEST = {
'url': 'http://g1.globo.com/jornal-nacional/noticia/2014/09/novidade-na-fiscalizacao-de-bagagem-pela-receita-provoca-discussoes.html',
'md5': '307fdeae4390ccfe6ba1aa198cf6e72b',
'info_dict': {
'id': '3652183',
'ext': 'mp4',
'title': 'Receita Federal explica como vai fiscalizar bagagens de quem retorna ao Brasil de avião',
'duration': 110.711,
'uploader': 'Rede Globo',
'uploader_id': 196,
'like_count': int,
}
}
@classmethod
def suitable(cls, url):
return False if GloboIE.suitable(url) else super(GloboArticleIE, cls).suitable(url)
def _real_extract(self, url):
display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id)
video_id = self._search_regex(self._VIDEOID_REGEXES, webpage, 'video id')
return self.url_result('globo:%s' % video_id, 'Globo')

Write Preview
Loading…
Cancel
Save