Cool_Tools/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-12-01 10:47:30 +01:00
Code Issues Packages Projects Releases Wiki Activity

[appletrailers] Modernize

Browse Source
This commit is contained in:
Jaime Marquínez Ferrándiz 2014-03-27 21:10:51 +01:00
parent 214c22c704
commit 28acf5500a
1 changed files with 15 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
youtube_dl/extractor/appletrailers.py
View File

@ -16,9 +16,10 @@ class AppleTrailersIE(InfoExtractor):
"url": "http://trailers.apple.com/trailers/wb/manofsteel/", "url": "http://trailers.apple.com/trailers/wb/manofsteel/",
"playlist": [ "playlist": [
{ {
"file": "manofsteel-trailer4.mov",
"md5": "d97a8e575432dbcb81b7c3acb741f8a8", "md5": "d97a8e575432dbcb81b7c3acb741f8a8",
"info_dict": { "info_dict": {
"id": "manofsteel-trailer4",
"ext": "mov",
"duration": 111, "duration": 111,
"title": "Trailer 4", "title": "Trailer 4",
"upload_date": "20130523", "upload_date": "20130523",
@ -26,9 +27,10 @@ class AppleTrailersIE(InfoExtractor):
}, },
}, },
{ {
"file": "manofsteel-trailer3.mov",
"md5": "b8017b7131b721fb4e8d6f49e1df908c", "md5": "b8017b7131b721fb4e8d6f49e1df908c",
"info_dict": { "info_dict": {
"id": "manofsteel-trailer3",
"ext": "mov",
"duration": 182, "duration": 182,
"title": "Trailer 3", "title": "Trailer 3",
"upload_date": "20130417", "upload_date": "20130417",
@ -36,9 +38,10 @@ class AppleTrailersIE(InfoExtractor):
}, },
}, },
{ {
"file": "manofsteel-trailer.mov",
"md5": "d0f1e1150989b9924679b441f3404d48", "md5": "d0f1e1150989b9924679b441f3404d48",
"info_dict": { "info_dict": {
"id": "manofsteel-trailer",
"ext": "mov",
"duration": 148, "duration": 148,
"title": "Trailer", "title": "Trailer",
"upload_date": "20121212", "upload_date": "20121212",
@ -46,15 +49,16 @@ class AppleTrailersIE(InfoExtractor):
}, },
}, },
{ {
"file": "manofsteel-teaser.mov",
"md5": "5fe08795b943eb2e757fa95cb6def1cb", "md5": "5fe08795b943eb2e757fa95cb6def1cb",
"info_dict": { "info_dict": {
"id": "manofsteel-teaser",
"ext": "mov",
"duration": 93, "duration": 93,
"title": "Teaser", "title": "Teaser",
"upload_date": "20120721", "upload_date": "20120721",
"uploader_id": "wb", "uploader_id": "wb",
}, },
} },
] ]
} }
@ -65,16 +69,16 @@ def _real_extract(self, url):
movie = mobj.group('movie') movie = mobj.group('movie')
uploader_id = mobj.group('company') uploader_id = mobj.group('company')
playlist_url = compat_urlparse.urljoin(url, u'includes/playlists/itunes.inc') playlist_url = compat_urlparse.urljoin(url, 'includes/playlists/itunes.inc')
def fix_html(s): def fix_html(s):
s = re.sub(r'(?s)<script[^<]*?>.*?</script>', u'', s) s = re.sub(r'(?s)<script[^<]*?>.*?</script>', '', s)
s = re.sub(r'<img ([^<]*?)>', r'<img \1/>', s) s = re.sub(r'<img ([^<]*?)>', r'<img \1/>', s)
# The ' in the onClick attributes are not escaped, it couldn't be parsed # The ' in the onClick attributes are not escaped, it couldn't be parsed
# like: http://trailers.apple.com/trailers/wb/gravity/ # like: http://trailers.apple.com/trailers/wb/gravity/
def _clean_json(m): def _clean_json(m):
return u'iTunes.playURL(%s);' % m.group(1).replace('\'', '&#39;') return 'iTunes.playURL(%s);' % m.group(1).replace('\'', '&#39;')
s = re.sub(self._JSON_RE, _clean_json, s) s = re.sub(self._JSON_RE, _clean_json, s)
s = u'<html>' + s + u'</html>' s = '<html>' + s + u'</html>'
return s return s
doc = self._download_xml(playlist_url, movie, transform_source=fix_html) doc = self._download_xml(playlist_url, movie, transform_source=fix_html)
@ -82,7 +86,7 @@ def _clean_json(m):
for li in doc.findall('./div/ul/li'): for li in doc.findall('./div/ul/li'):
on_click = li.find('.//a').attrib['onClick'] on_click = li.find('.//a').attrib['onClick']
trailer_info_json = self._search_regex(self._JSON_RE, trailer_info_json = self._search_regex(self._JSON_RE,
on_click, u'trailer info') on_click, 'trailer info')
trailer_info = json.loads(trailer_info_json) trailer_info = json.loads(trailer_info_json)
title = trailer_info['title'] title = trailer_info['title']
video_id = movie + '-' + re.sub(r'[^a-zA-Z0-9]', '', title).lower() video_id = movie + '-' + re.sub(r'[^a-zA-Z0-9]', '', title).lower()
@ -98,8 +102,7 @@ def _clean_json(m):
first_url = trailer_info['url'] first_url = trailer_info['url']
trailer_id = first_url.split('/')[-1].rpartition('_')[0].lower() trailer_id = first_url.split('/')[-1].rpartition('_')[0].lower()
settings_json_url = compat_urlparse.urljoin(url, 'includes/settings/%s.json' % trailer_id) settings_json_url = compat_urlparse.urljoin(url, 'includes/settings/%s.json' % trailer_id)
settings_json = self._download_webpage(settings_json_url, trailer_id, u'Downloading settings json') settings = self._download_json(settings_json_url, trailer_id, 'Downloading settings json')
settings = json.loads(settings_json)
formats = [] formats = []
for format in settings['metadata']['sizes']: for format in settings['metadata']['sizes']:
@ -107,7 +110,6 @@ def _clean_json(m):
format_url = re.sub(r'_(\d*p.mov)', r'_h\1', format['src']) format_url = re.sub(r'_(\d*p.mov)', r'_h\1', format['src'])
formats.append({ formats.append({
'url': format_url, 'url': format_url,
'ext': determine_ext(format_url),
'format': format['type'], 'format': format['type'],
'width': format['width'], 'width': format['width'],
'height': int(format['height']), 'height': int(format['height']),