Enclosed is a patch for the "redirect" tag in WebUtils to improve the
resolution of relative uris. I hope this is helpful--let me know if
more information is needed or if I need to do something differently.
The first attachment to this message is a comparison of different
algorithms for uri-resolution, including the examples from RFC2396 and
the output of URI.pm. The column labeled WebUtils0 is from WebUtils
revision 1.5.
I fixed most of the differences between the output of WebUtils and URI.pm
with four changes to the redirect subroutine:
0. commented out the line adding a leading "./" (why was this needed?)
1. added missing backslash to "/./" regex
2. added a while loop around "/./" regex
3. removed 'g' modifier in while loops
The results of these changes are shown in the column labeled WebUtils1.
However, there were a few remaining differences in the handling of
"/./" and "/../", which I fixed by adding some lines to deal with
leading "/../" and to deal with query strings and fragments (#...).
The results of these changes are shown in the column labeled
WebUtils2. Note that there is one remaining difference, the case when
the uri consists solely of a fragment, "#s". But I decided not to
deal with this case since it causes a redirection loop and so no one
should be using it.
The second attachment is the script that produces the
comparisons--feel free to experiment with this script to try other
test cases and to test additional improvements to the uri-resolution
algorithm. If anyone finds a test case that is not handled properly
let me know and I will try to find a way to handle it.
The third attachment is my patch with the above fixes. The fourth
attachment is a patch that uses URI.pm directly. This is the most
straightforward solution, but I assume URI.pm is not being used so
that WebUtils.pm does not have to rely on this extra package being
installed.
Ken
P.S. Soon I am going on vacation until mid-Oct, so I may not be able
to reply to emails during that time.
BASE: /b/c/d;p?q#f
RELATIVE RFC URI WebUtils0 WebUtils1
WebUtils2
g /b/c/g /b/c/g NC: /c/ /b/c/g
/b/c/g
./g /b/c/g /b/c/g NC: /c/./ /b/c/g
/b/c/g
g/ /b/c/g/ /b/c/g/ NC: /c/ /b/c/g/
/b/c/g/
/g /g /g /g /g
/g
//g //g //g //g //g
//g
?y /b/c/?y /b/c/?y NC: /c/?y /b/c/?y
/b/c/?y
?y/./x /b/c/?y/./x /b/c/?y/./x NC: /c/?y/x NC: /b/c/?y/x
/b/c/?y/./x
?y/../x /b/c/?y/../x /b/c/?y/../x NC: /c/ NC: /b/c/x
/b/c/?y/../x
g?y /b/c/g?y /b/c/g?y NC: /c/g?y /b/c/g?y
/b/c/g?y
#s /b/c/d;p?q#s /b/c/d;p?q#s NC: /c/#s NC: /b/c/#s
NC: /b/c/#s
g#s /b/c/g#s /b/c/g#s NC: /c/g#s /b/c/g#s
/b/c/g#s
g?y#s /b/c/g?y#s /b/c/g?y#s NC: /c/g?y#s /b/c/g?y#s
/b/c/g?y#s
;x /b/c/;x /b/c/;x NC: /c/;x /b/c/;x
/b/c/;x
g;x /b/c/g;x /b/c/g;x NC: /c/g;x /b/c/g;x
/b/c/g;x
g;x?y#s /b/c/g;x?y#s /b/c/g;x?y#s NC: /c/g;x?y#s /b/c/g;x?y#s
/b/c/g;x?y#s
. /b/c/ /b/c/ NC: /c/. /b/c/
/b/c/
./ /b/c/ /b/c/ NC: /c/./ /b/c/
/b/c/
.. /b/ /b/ NC: / /b/
/b/
../ /b/ /b/ NC: / /b/
/b/
../g /b/g /b/g NC: / /b/g
/b/g
../.. / / / /
/
../../ / / / /
/
../../g /g /g NC: / /g
/g
/b/c/d;p?q#f /b/c/d;p?q#f NC: /c/ NC: /b/c/
/b/c/d;p?q#f
../../../g /../g /../g NC: / NC: /g
/../g
../../../../g /../../g /../../g NC: / NC: /g
/../../g
../../../../g/h/../i /../../g/i /../../g/i NC: / NC: /g/i
/../../g/i
../..g/../h /b/h /b/h NC: / /b/h
/b/h
/./g /./g /./g /./g /./g
/./g
/../g /../g /../g /../g /../g
/../g
g. /b/c/g. /b/c/g. NC: /c/g. /b/c/g.
/b/c/g.
.g /b/c/.g /b/c/.g NC: /c/.g /b/c/.g
/b/c/.g
g.. /b/c/g.. /b/c/g.. NC: /c/g.. /b/c/g..
/b/c/g..
..g /b/c/..g /b/c/..g NC: /c/..g /b/c/..g
/b/c/..g
./../g /b/g /b/g NC: /c/ /b/g
/b/g
./g/. /b/c/g/ /b/c/g/ NC: /c/./. /b/c/g/
/b/c/g/
g/./h /b/c/g/h /b/c/g/h NC: /c/./ /b/c/g/h
/b/c/g/h
g/../h /b/c/h /b/c/h NC: / /b/c/h
/b/c/h
g;x=1/./y /b/c/g;x=1/y /b/c/g;x=1/y NC: /c/g;x=1/y /b/c/g;x=1/y
/b/c/g;x=1/y
g;x=1/../y /b/c/y /b/c/y NC: /c/ /b/c/y
/b/c/y
g?y/./x /b/c/g?y/./x /b/c/g?y/./x NC: /c/g?y/x NC: /b/c/g?y/x
/b/c/g?y/./x
g?y/../x /b/c/g?y/../x /b/c/g?y/../x NC: /c/ NC: /b/c/x
/b/c/g?y/../x
g#s/./x /b/c/g#s/./x /b/c/g#s/./x NC: /c/g#s/x NC: /b/c/g#s/x
/b/c/g#s/./x
g#s/../x /b/c/g#s/../x /b/c/g#s/../x NC: /c/ NC: /b/c/x
/b/c/g#s/../x
Results marked 'NC:' are not compliant with examples from RFC 2396.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
