Fix metacafe.com and UTF8 output filenames

This commit is contained in:
Ricardo Garcia 2008-09-13 13:23:24 +02:00
parent 0ddf38df18
commit 2546e7679f

View File

@ -4,6 +4,7 @@
# License: Public domain code # License: Public domain code
import htmlentitydefs import htmlentitydefs
import httplib import httplib
import locale
import math import math
import netrc import netrc
import os import os
@ -592,7 +593,7 @@ class MetacafeIE(InfoExtractor):
"""Information Extractor for metacafe.com.""" """Information Extractor for metacafe.com."""
_VALID_URL = r'(?:http://)?(?:www\.)?metacafe\.com/watch/([^/]+)/([^/]+)/.*' _VALID_URL = r'(?:http://)?(?:www\.)?metacafe\.com/watch/([^/]+)/([^/]+)/.*'
_DISCLAIMER = 'http://www.metacafe.com/disclaimer' _DISCLAIMER = 'http://www.metacafe.com/family_filter/'
_youtube_ie = None _youtube_ie = None
def __init__(self, youtube_ie, downloader=None): def __init__(self, youtube_ie, downloader=None):
@ -631,10 +632,10 @@ class MetacafeIE(InfoExtractor):
# Confirm age # Confirm age
disclaimer_form = { disclaimer_form = {
'allowAdultContent': '1', 'filters': '0',
'submit': "Continue - I'm over 18", 'submit': "Continue - I'm over 18",
} }
request = urllib2.Request('http://www.metacafe.com/watch/', urllib.urlencode(disclaimer_form), std_headers) request = urllib2.Request('http://www.metacafe.com/', urllib.urlencode(disclaimer_form), std_headers)
try: try:
self.report_age_confirmation() self.report_age_confirmation()
disclaimer = urllib2.urlopen(request).read() disclaimer = urllib2.urlopen(request).read()
@ -684,7 +685,7 @@ class MetacafeIE(InfoExtractor):
video_url = '%s?__gda__=%s' % (mediaURL, gdaKey) video_url = '%s?__gda__=%s' % (mediaURL, gdaKey)
mobj = re.search(r'(?im)<meta name="title" content="Metacafe - ([^"]+)"', webpage) mobj = re.search(r'(?im)<title>(.*) - Video</title>', webpage)
if mobj is None: if mobj is None:
self.to_stderr(u'ERROR: unable to extract title') self.to_stderr(u'ERROR: unable to extract title')
return [None] return [None]
@ -909,7 +910,7 @@ if __name__ == '__main__':
'forcetitle': opts.gettitle, 'forcetitle': opts.gettitle,
'simulate': (opts.simulate or opts.geturl or opts.gettitle), 'simulate': (opts.simulate or opts.geturl or opts.gettitle),
'format': opts.format, 'format': opts.format,
'outtmpl': ((opts.outtmpl is not None and opts.outtmpl.decode()) 'outtmpl': ((opts.outtmpl is not None and opts.outtmpl.decode(locale.getdefaultlocale()[1]))
or (opts.usetitle and u'%(stitle)s-%(id)s.%(ext)s') or (opts.usetitle and u'%(stitle)s-%(id)s.%(ext)s')
or (opts.useliteral and u'%(title)s-%(id)s.%(ext)s') or (opts.useliteral and u'%(title)s-%(id)s.%(ext)s')
or u'%(id)s.%(ext)s'), or u'%(id)s.%(ext)s'),