[crackle] improve extraction

- extract vtt subtitles
- extract multiple resolutions for thumbnails
- pass geo verification proxy headers
- add support for mobile urls
This commit is contained in:
Remita Amine 2017-01-24 16:55:07 +01:00
parent 74af9c700d
commit 23b35a634e
1 changed files with 43 additions and 10 deletions

View File

@ -6,7 +6,7 @@ from ..utils import int_or_none
class CrackleIE(InfoExtractor): class CrackleIE(InfoExtractor):
_VALID_URL = r'(?:crackle:|https?://(?:www\.)?crackle\.com/(?:playlist/\d+/|(?:[^/]+/)+))(?P<id>\d+)' _VALID_URL = r'(?:crackle:|https?://(?:(?:www|m)\.)?crackle\.com/(?:playlist/\d+/|(?:[^/]+/)+))(?P<id>\d+)'
_TEST = { _TEST = {
'url': 'http://www.crackle.com/comedians-in-cars-getting-coffee/2498934', 'url': 'http://www.crackle.com/comedians-in-cars-getting-coffee/2498934',
'info_dict': { 'info_dict': {
@ -31,8 +31,32 @@ class CrackleIE(InfoExtractor):
} }
} }
_THUMBNAIL_RES = [
(120, 90),
(208, 156),
(220, 124),
(220, 220),
(240, 180),
(250, 141),
(315, 236),
(320, 180),
(360, 203),
(400, 300),
(421, 316),
(460, 330),
(460, 460),
(462, 260),
(480, 270),
(587, 330),
(640, 480),
(700, 330),
(700, 394),
(854, 480),
(1024, 1024),
(1920, 1080),
]
# extracted from http://legacyweb-us.crackle.com/flash/ReferrerRedirect.ashx # extracted from http://legacyweb-us.crackle.com/flash/ReferrerRedirect.ashx
_THUMBNAIL_TEMPLATE = 'http://images-us-am.crackle.com/%stnl_1920x1080.jpg?ts=20140107233116?c=635333335057637614'
_MEDIA_FILE_SLOTS = { _MEDIA_FILE_SLOTS = {
'c544.flv': { 'c544.flv': {
'width': 544, 'width': 544,
@ -61,17 +85,25 @@ class CrackleIE(InfoExtractor):
item = self._download_xml( item = self._download_xml(
'http://legacyweb-us.crackle.com/app/revamp/vidwallcache.aspx?flags=-1&fm=%s' % video_id, 'http://legacyweb-us.crackle.com/app/revamp/vidwallcache.aspx?flags=-1&fm=%s' % video_id,
video_id).find('i') video_id, headers=self.geo_verification_headers()).find('i')
title = item.attrib['t'] title = item.attrib['t']
subtitles = {} subtitles = {}
formats = self._extract_m3u8_formats( formats = self._extract_m3u8_formats(
'http://content.uplynk.com/ext/%s/%s.m3u8' % (config_doc.attrib['strUplynkOwnerId'], video_id), 'http://content.uplynk.com/ext/%s/%s.m3u8' % (config_doc.attrib['strUplynkOwnerId'], video_id),
video_id, 'mp4', m3u8_id='hls', fatal=None) video_id, 'mp4', m3u8_id='hls', fatal=None)
thumbnail = None thumbnails = []
path = item.attrib.get('p') path = item.attrib.get('p')
if path: if path:
thumbnail = self._THUMBNAIL_TEMPLATE % path for width, height in self._THUMBNAIL_RES:
res = '%dx%d' % (width, height)
thumbnails.append({
'id': res,
'url': 'http://images-us-am.crackle.com/%stnl_%s.jpg' % (path, res),
'width': width,
'height': height,
'resolution': res,
})
http_base_url = 'http://ahttp.crackle.com/' + path http_base_url = 'http://ahttp.crackle.com/' + path
for mfs_path, mfs_info in self._MEDIA_FILE_SLOTS.items(): for mfs_path, mfs_info in self._MEDIA_FILE_SLOTS.items():
formats.append({ formats.append({
@ -86,10 +118,11 @@ class CrackleIE(InfoExtractor):
if locale and v: if locale and v:
if locale not in subtitles: if locale not in subtitles:
subtitles[locale] = [] subtitles[locale] = []
subtitles[locale] = [{ for url_ext, ext in (('vtt', 'vtt'), ('xml', 'tt')):
'url': '%s/%s%s_%s.xml' % (config_doc.attrib['strSubtitleServer'], path, locale, v), subtitles.setdefault(locale, []).append({
'ext': 'ttml', 'url': '%s/%s%s_%s.%s' % (config_doc.attrib['strSubtitleServer'], path, locale, v, url_ext),
}] 'ext': ext,
})
self._sort_formats(formats, ('width', 'height', 'tbr', 'format_id')) self._sort_formats(formats, ('width', 'height', 'tbr', 'format_id'))
return { return {
@ -100,7 +133,7 @@ class CrackleIE(InfoExtractor):
'series': item.attrib.get('sn'), 'series': item.attrib.get('sn'),
'season_number': int_or_none(item.attrib.get('se')), 'season_number': int_or_none(item.attrib.get('se')),
'episode_number': int_or_none(item.attrib.get('ep')), 'episode_number': int_or_none(item.attrib.get('ep')),
'thumbnail': thumbnail, 'thumbnails': thumbnails,
'subtitles': subtitles, 'subtitles': subtitles,
'formats': formats, 'formats': formats,
} }