Merge remote-tracking branch 'dstftw/escape-non-ascii-in-urls'

Conflicts:
	test/test_utils.py

diff --git a/test/test_utils.py b/test/test_utils.py
index 70050d2b836aedfd7310bbee2c2f07037d70c71a..3efbed29dd34de570f2db4e6eb4954ec2f4b9c6e 100644 (file)
--- a/test/test_utils.py
+++ b/test/test_utils.py
@@ -41,6 +41,8 @@ from youtube_dl.utils import (
     strip_jsonp,
     uppercase_escape,
     limit_length,
+    escape_rfc3986,
+    escape_url,
 )
 
 
@@ -294,5 +296,34 @@ class TestUtil(unittest.TestCase):
             limit_length('foo bar baz asd', 12).startswith('foo bar'))
         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 
+    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 553bf559b3b2e7b155b2e14b2f2f49eda91ba9e9..9519594c9ad2dab36e1d66fe2f016d7edc6e798d 100755 (executable)
--- a/youtube_dl/YoutubeDL.py
+++ b/youtube_dl/YoutubeDL.py
@@ -28,6 +28,7 @@ from .utils import (
     compat_str,
     compat_urllib_error,
     compat_urllib_request,
+    escape_url,
     ContentTooShortError,
     date_from_str,
     DateRange,
@@ -1241,6 +1242,25 @@ 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)
+        # 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()
+        url_escaped = escape_url(url)
+
+        # Substitute URL if any change after escaping
+        if url != url_escaped:
+            if isinstance(req, compat_str):
+                req = url_escaped
+            else:
+                req = compat_urllib_request.Request(
+                    url_escaped, data=req.data, headers=req.headers,
+                    origin_req_host=req.origin_req_host, unverifiable=req.unverifiable)
+
         return self._opener.open(req, timeout=self._socket_timeout)
 
     def print_debug_header(self):

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index 3ac0f1f541745b1ec34f1245574ea82387c3115d..b644f4e920bf0353658ec9920abdb0541dbaf0e2 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -1437,6 +1437,24 @@ 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: