[Yum-devel] [UG] encoding characters in URL

[Chris Lumens]

> Honestly, I don't think it even occurred to me to check.  The one
> warning I'll give is that you need to make sure you do it in such a
> way that it doesn't screw up the other protocols.  I'd probably do it
> in grabber.URLGrabber._parse_url.

That seemed really easy, which makes me wonder what I am forgetting.
Attached below is a patch to take care of the quoting and unquoting for
HTTP and HTTPS protocols only.  I don't think I see any way for URLs on
other protocols to get mangled.

- Chris


diff -ruN yum-2.1.13.orig/urlgrabber/grabber.py yum-2.1.13/urlgrabber/grabber.py
--- yum-2.1.13.orig/urlgrabber/grabber.py	2005-01-20 01:59:18.000000000 -0500
+++ yum-2.1.13/urlgrabber/grabber.py	2005-02-03 17:22:12.000000000 -0500
@@ -287,6 +287,7 @@
  import rfc822
  import time
  import string
+import urllib
  import urllib2
  from stat import *  # S_* and ST_*

@@ -570,7 +571,10 @@
          (url, parts) = self._parse_url(url)
          (scheme, host, path, parm, query, frag) = parts
          if filename is None:
-            filename = os.path.basename( path )
+            if scheme in [ 'http', 'https' ]:
+                filename = os.path.basename( urllib.unquote(path) )
+            else:
+                filename = os.path.basename( path )
          if scheme == 'file' and not opts.copy_local:
              # just return the name of the local file - don't make a
              # copy currently
@@ -662,6 +666,10 @@
              (scheme, host, path, parm, query, frag) = \
                                               urlparse.urlparse(url)
          path = os.path.normpath(path)
+
+        if scheme in [ 'http', 'https' ]:
+            path = urllib.quote(path)
+
          if '@' in host and auth_handler and scheme in ['http', 'https']:
              try:
                  user_pass, host = host.split('@', 1)