X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2FYoutubeDL.py;h=9519594c9ad2dab36e1d66fe2f016d7edc6e798d;hb=e825c380821ca111740b4bf5d6d6de852a44a87c;hp=99cd05e6518a726a3b72e8f1d705d7733790ebe1;hpb=37419b4f9937f11ed3ca3545a32ed3451eb734ee;p=youtube-dl diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index 99cd05e65..9519594c9 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -28,7 +28,7 @@ from .utils import ( compat_str, compat_urllib_error, compat_urllib_request, - compat_urllib_parse_urlparse, + escape_url, ContentTooShortError, date_from_str, DateRange, @@ -1243,20 +1243,14 @@ class YoutubeDL(object): def urlopen(self, req): """ Start an HTTP download """ + # According to RFC 3986, URLs can not contain non-ASCII characters, however this is not + # always respected by websites, some tend to give out URLs with non percent-encoded + # non-ASCII characters (see telemb.py, ard.py [#3412]) # urllib chokes on URLs with non-ASCII characters (see http://bugs.python.org/issue3991) - # Working around by replacing request's original URL with escaped one - + # To work around aforementioned issue we will replace request's original URL with + # percent-encoded one url = req if isinstance(req, compat_str) else req.get_full_url() - - def escape(component): - return compat_cookiejar.escape_path(component.encode('utf-8')) - - url_parsed = compat_urllib_parse_urlparse(url) - url_escaped = url_parsed._replace( - path=escape(url_parsed.path), - query=escape(url_parsed.query), - fragment=escape(url_parsed.fragment) - ).geturl() + url_escaped = escape_url(url) # Substitute URL if any change after escaping if url != url_escaped: