From d05cfe06006c4a44032e95dde047d5e12be8674c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sat, 13 Sep 2014 20:59:16 +0700 Subject: [PATCH] [YoutubeDL/utils] Clarify rationale for URL escaping in comment, move escape routines to utils and add some tests --- test/test_utils.py | 31 +++++++++++++++++++++++++++++++ youtube_dl/YoutubeDL.py | 20 +++++++------------- youtube_dl/utils.py | 18 ++++++++++++++++++ 3 files changed, 56 insertions(+), 13 deletions(-) diff --git a/test/test_utils.py b/test/test_utils.py index 8d8997977..e90caed29 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -40,6 +40,8 @@ parse_iso8601, strip_jsonp, uppercase_escape, + escape_rfc3986, + escape_url, ) @@ -286,5 +288,34 @@ def test_uppercase_escape(self): self.assertEqual(uppercase_escape('aä'), 'aä') self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐') + def test_escape_rfc3986(self): + reserved = "!*'();:@&=+$,/?#[]" + unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~' + self.assertEqual(escape_rfc3986(reserved), reserved) + self.assertEqual(escape_rfc3986(unreserved), unreserved) + self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82') + self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82') + self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar') + self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar') + + def test_escape_url(self): + self.assertEqual( + escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'), + 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4' + ) + self.assertEqual( + escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'), + 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290' + ) + self.assertEqual( + escape_url('http://тест.рф/фрагмент'), + 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' + ) + self.assertEqual( + escape_url('http://тест.рф/абв?абв=абв#абв'), + 'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2' + ) + self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') + if __name__ == '__main__': unittest.main() 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 @@ compat_str, compat_urllib_error, compat_urllib_request, - compat_urllib_parse_urlparse, + escape_url, ContentTooShortError, date_from_str, DateRange, @@ -1243,20 +1243,14 @@ def line(format, idlen=20): 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: diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index d920c65a4..9124c3621 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -1418,6 +1418,24 @@ def uppercase_escape(s): lambda m: unicode_escape(m.group(0))[0], s) + +def escape_rfc3986(s): + """Escape non-ASCII characters as suggested by RFC 3986""" + if sys.version_info < (3, 0) and isinstance(s, unicode): + s = s.encode('utf-8') + return compat_urllib_parse.quote(s, "%/;:@&=+$,!~*'()?#[]") #"%/;:@&=+$,!~*'()?#[]+" #?#[]+ + + +def escape_url(url): + """Escape URL as suggested by RFC 3986""" + url_parsed = compat_urllib_parse_urlparse(url) + return url_parsed._replace( + path=escape_rfc3986(url_parsed.path), + params=escape_rfc3986(url_parsed.params), + query=escape_rfc3986(url_parsed.query), + fragment=escape_rfc3986(url_parsed.fragment) + ).geturl() + try: struct.pack(u'!I', 0) except TypeError: