you-get/src/you_get/extractors/embed.py

99 lines
3.5 KiB
Python
Raw Normal View History

__all__ = ['embed_download']
from ..common import *
from .bilibili import bilibili_download
from .iqiyi import iqiyi_download_by_vid
2016-03-03 23:49:47 +03:00
from .le import letvcloud_download_by_vu
2016-03-31 18:42:00 +03:00
from .netease import netease_download
from .qq import qq_download_by_vid
from .sina import sina_download_by_vid
from .tudou import tudou_download_by_id
from .vimeo import vimeo_download_by_id
from .yinyuetai import yinyuetai_download_by_id
from .youku import youku_download_by_vid
"""
refer to http://open.youku.com/tools
"""
youku_embed_patterns = [ 'youku\.com/v_show/id_([a-zA-Z0-9=]+)',
'player\.youku\.com/player\.php/sid/([a-zA-Z0-9=]+)/v\.swf',
'loader\.swf\?VideoIDS=([a-zA-Z0-9=]+)',
'player\.youku\.com/embed/([a-zA-Z0-9=]+)',
'YKU.Player\(\'[a-zA-Z0-9]+\',{ client_id: \'[a-zA-Z0-9]+\', vid: \'([a-zA-Z0-9]+)\''
]
"""
http://www.tudou.com/programs/view/html5embed.action?type=0&code=3LS_URGvl54&lcode=&resourceId=0_06_05_99
"""
tudou_embed_patterns = [ 'tudou\.com[a-zA-Z0-9\/\?=\&\.\;]+code=([a-zA-Z0-9_-]+)\&',
2015-12-11 21:31:13 +03:00
'www\.tudou\.com/v/([a-zA-Z0-9_-]+)/[^"]*v\.swf'
]
"""
refer to http://open.tudou.com/wiki/video/info
"""
tudou_api_patterns = [ ]
yinyuetai_embed_patterns = [ 'player\.yinyuetai\.com/video/swf/(\d+)' ]
iqiyi_embed_patterns = [ 'player\.video\.qiyi\.com/([^/]+)/[^/]+/[^/]+/[^/]+\.swf[^"]+tvId=(\d+)' ]
2016-03-31 18:42:00 +03:00
netease_embed_patterns = [ '(http://\w+\.163\.com/movie/[^\'"]+)' ]
vimeo_embed_patters = [ 'player\.vimeo\.com/video/(\d+)' ]
"""
check the share button on http://www.bilibili.com/video/av5079467/
"""
bilibili_embed_patterns = [ 'static\.hdslb\.com/miniloader\.swf.*aid=(\d+)' ]
def embed_download(url, output_dir = '.', merge = True, info_only = False ,**kwargs):
2016-03-31 18:42:00 +03:00
content = get_content(url, headers=fake_headers)
found = False
title = match1(content, '<title>([^<>]+)</title>')
2016-03-31 18:42:00 +03:00
vids = matchall(content, youku_embed_patterns)
for vid in set(vids):
found = True
youku_download_by_vid(vid, title=title, output_dir=output_dir, merge=merge, info_only=info_only)
vids = matchall(content, tudou_embed_patterns)
for vid in set(vids):
found = True
tudou_download_by_id(vid, title=title, output_dir=output_dir, merge=merge, info_only=info_only)
vids = matchall(content, yinyuetai_embed_patterns)
for vid in vids:
found = True
yinyuetai_download_by_id(vid, title=title, output_dir=output_dir, merge=merge, info_only=info_only)
vids = matchall(content, iqiyi_embed_patterns)
for vid in vids:
found = True
iqiyi_download_by_vid((vid[1], vid[0]), title=title, output_dir=output_dir, merge=merge, info_only=info_only)
2016-03-31 18:42:00 +03:00
urls = matchall(content, netease_embed_patterns)
for url in urls:
found = True
netease_download(url, title=title, output_dir=output_dir, merge=merge, info_only=info_only)
urls = matchall(content, vimeo_embed_patters)
for url in urls:
found = True
vimeo_download_by_id(url, title=title, output_dir=output_dir, merge=merge, info_only=info_only)
aids = matchall(content, bilibili_embed_patterns)
for aid in aids:
found = True
url = 'http://www.bilibili.com/video/av%s/' % aid
bilibili_download(url, output_dir=output_dir, merge=merge, info_only=info_only)
if not found:
raise NotImplementedError(url)
site_info = "any.any"
download = embed_download
download_playlist = playlist_not_supported('any.any')