| Subject: | correctly resolve 'subdir/../index.html' to 'index.html' |
Hi,
you can test the bug by executing the test script given below:
perl URI.pl
gives
Test with URI 1.29 on MSWin32 with Perl 5.8.2
Base: http://www.someDomain.tld/subdir/
URL: NotHere/../index.html
-> http://www.somedomain.tld/subdir/index.html
URL: http://www.someDomain.tld/subdir/NotHere/../index.html
-> http://www.somedomain.tld/subdir/NotHere/../index.html
with the original URI 1.2(?:8|9) and
Test with URI 1.29 on MSWin32 with Perl 5.8.2
Base: http://www.someDomain.tld/subdir/
URL: NotHere/../index.html
-> http://www.somedomain.tld/subdir/index.html
URL: http://www.someDomain.tld/subdir/NotHere/../index.html
-> http://www.somedomain.tld/subdir/index.html
after applying my patch.
The patch can be applied by
patch -b _generic.pm _generic.patch
The patched URI/_generic.pm does exactly behave as before *unless* $URI::ABS_REMOTE_LEADING_DOTS is set to true, to stay compatible with the previous behaviour.
best,
rob.
########## test script URI.pl begin ##########
$^W = 1;
use strict;
require URI;
my $base = "http://www.someDomain.tld/subdir/";
my $url;
printf "Test with URI " . URI -> VERSION . " on $^O with Perl %vd\n", $^V;
print "Base: $base\n\n";
&fullurl("NotHere/../index.html");
&fullurl("http://www.someDomain.tld/subdir/NotHere/../index.html");
exit 0;
sub fullurl {
my $url = shift;
my $uri = URI -> new($url, $base);
local $URI::ABS_REMOTE_LEADING_DOTS = 1;
# required for the patched abs method in URI/_generic.pm
# to clean URLs like "dir/../index.html" if the URL is absolute
print "URL: $url\n -> ", $uri -> abs($base) -> canonical, "\n\n";
}
########## test script end ##########
124a125,127
> # modifications by roal to allow URLs like "dir/../index.html" to be cleaned *also if the base is absolute*
> # are marked with "ABS_REMOTE_DOTS"
> # ABS_REMOTE_LEADING_DOTS must be set to true to apply that
130c133,135
< return $self unless $URI::ABS_ALLOW_RELATIVE_SCHEME;
---
> # ABS_REMOTE_DOTS
> # return $self unless $URI::ABS_ALLOW_RELATIVE_SCHEME;
> return $self unless ($URI::ABS_ALLOW_RELATIVE_SCHEME || $URI::ABS_REMOTE_LEADING_DOTS);
138c143,148
< return $abs if $$self =~ m,^(?:$URI::scheme_re:)?//,o;
---
> # ABS_REMOTE_DOTS
> # return $abs if $$self =~ m,^(?:$URI::scheme_re:)?//,o;
> if ($$self =~ m,^(?:$URI::scheme_re:)?//,o) {
> return $abs unless ($URI::ABS_REMOTE_LEADING_DOTS && lc($self->host_port) eq lc($base->host_port));
> }
>
142c152,154
< return $abs if $path =~ m,^/,;
---
> # ABS_REMOTE_DOTS
> # return $abs if $path =~ m,^/,;
> return $abs if ($path =~ m,^/, && !$URI::ABS_REMOTE_LEADING_DOTS);
154c166,170
< $p .= $path;
---
>
> # ABS_REMOTE_DOTS
> # $p .= $path;
> $p = $path =~ m,^/, ? $path : $p . $path;
>