lopar/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-14 13:27:27 +00:00
Code Releases Activity
73c566695f
youtube-dl/youtube_dl/extractor/appletrailers.py

139 lines
5.3 KiB
Python
Raw Normal View History

[appletrailers] Add support
2013-08-28 00:18:44 +00:00
import re
import xml.etree.ElementTree
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
import json
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
from .common import InfoExtractor
from ..utils import (
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
compat_urlparse,
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
determine_ext,
)
class AppleTrailersIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?trailers.apple.com/trailers/(?P<company>[^/]+)/(?P<movie>[^/]+)'
_TEST = {
u"url": u"http://trailers.apple.com/trailers/wb/manofsteel/",
u"playlist": [
{
u"file": u"manofsteel-trailer4.mov",
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
u"md5": u"d97a8e575432dbcb81b7c3acb741f8a8",
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
u"info_dict": {
u"duration": 111,
u"title": u"Trailer 4",
u"upload_date": u"20130523",
u"uploader_id": u"wb",
},
},
{
u"file": u"manofsteel-trailer3.mov",
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
u"md5": u"b8017b7131b721fb4e8d6f49e1df908c",
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
u"info_dict": {
u"duration": 182,
u"title": u"Trailer 3",
u"upload_date": u"20130417",
u"uploader_id": u"wb",
},
},
{
u"file": u"manofsteel-trailer.mov",
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
u"md5": u"d0f1e1150989b9924679b441f3404d48",
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
u"info_dict": {
u"duration": 148,
u"title": u"Trailer",
u"upload_date": u"20121212",
u"uploader_id": u"wb",
},
},
{
u"file": u"manofsteel-teaser.mov",
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
u"md5": u"5fe08795b943eb2e757fa95cb6def1cb",
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
u"info_dict": {
u"duration": 93,
u"title": u"Teaser",
u"upload_date": u"20120721",
u"uploader_id": u"wb",
},
}
]
}
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
_JSON_RE = r'iTunes.playURL\((.*?)\);'
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
movie = mobj.group('movie')
uploader_id = mobj.group('company')
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
playlist_url = compat_urlparse.urljoin(url, u'includes/playlists/itunes.inc')
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
playlist_snippet = self._download_webpage(playlist_url, movie)
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
playlist_cleaned = re.sub(r'(?s)<script[^<]*?>.*?</script>', u'', playlist_snippet)
playlist_cleaned = re.sub(r'<img ([^<]*?)>', r'<img \1/>', playlist_cleaned)
# The ' in the onClick attributes are not escaped, it couldn't be parsed
# with xml.etree.ElementTree.fromstring
# like: http://trailers.apple.com/trailers/wb/gravity/
def _clean_json(m):
return u'iTunes.playURL(%s);' % m.group(1).replace('\'', '&#39;')
playlist_cleaned = re.sub(self._JSON_RE, _clean_json, playlist_cleaned)
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
playlist_html = u'<html>' + playlist_cleaned + u'</html>'
doc = xml.etree.ElementTree.fromstring(playlist_html)
playlist = []
for li in doc.findall('./div/ul/li'):
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
on_click = li.find('.//a').attrib['onClick']
trailer_info_json = self._search_regex(self._JSON_RE,
on_click, u'trailer info')
trailer_info = json.loads(trailer_info_json)
title = trailer_info['title']
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
video_id = movie + '-' + re.sub(r'[^a-zA-Z0-9]', '', title).lower()
thumbnail = li.find('.//img').attrib['src']
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
upload_date = trailer_info['posted'].replace('-', '')
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
runtime = trailer_info['runtime']
m = re.search(r'(?P<minutes>[0-9]+):(?P<seconds>[0-9]{1,2})', runtime)
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
duration = None
if m:
duration = 60 * int(m.group('minutes')) + int(m.group('seconds'))
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
first_url = trailer_info['url']
[appletrailers] The request for the settings must have the trailer name in lower case (fixes #1329)
2013-09-29 18:59:19 +00:00
trailer_id = first_url.split('/')[-1].rpartition('_')[0].lower()
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
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 = json.loads(settings_json)
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
[appletrailers] Rework extraction (fixes #1387) The exraction was broken: * The includes page contains img elements that need to be fixed. * Use the 'itunes.inc' page, it contains a json dictionary for each trailer with information. * Get the formats from 'includes/settings{trailer_name}.json' * Use urljoin to allow urls with a fragment identifier to work Removed the thumbnail urls from the tests, they are different now.
2013-09-29 18:49:58 +00:00
formats = []
for format in settings['metadata']['sizes']:
# The src is a file pointing to the real video file
format_url = re.sub(r'_(\d*p.mov)', r'_h\1', format['src'])
formats.append({
'url': format_url,
'ext': determine_ext(format_url),
'format': format['type'],
'width': format['width'],
'height': int(format['height']),
})
formats = sorted(formats, key=lambda f: (f['height'], f['width']))
[appletrailers] Add support
2013-08-28 00:18:44 +00:00
info = {
'_type': 'video',
'id': video_id,
'title': title,
'formats': formats,
'title': title,
'duration': duration,
'thumbnail': thumbnail,
'upload_date': upload_date,
'uploader_id': uploader_id,
'user_agent': 'QuickTime compatible (youtube-dl)',
}
# TODO: Remove when #980 has been merged
info['url'] = formats[-1]['url']
info['ext'] = formats[-1]['ext']
playlist.append(info)
return {
'_type': 'playlist',
'id': movie,
'entries': playlist,
}
Copy Permalink