NAME
CGI::Auth::FOAF_SSL - authentication using FOAF+SSL (WebID)
SYNOPSIS
use CGI qw(:all);
use CGI::Auth::FOAF_SSL;
my $auth = CGI::Auth::FOAF_SSL->new_from_cgi( CGI->new );
print header('-type' => 'text/html', '-cookie' => $auth->cookie);
if (defined $auth && $auth->is_secure)
{
if (defined $auth->agent)
{
printf("<p>Hello <a href='%s'>%s</a>!</p>\n",
escapeHTML($auth->agent->homepage),
escapeHTML($auth->agent->name));
}
else
{
print "<p>Hello!</p>\n";
}
}
else
{
print "<p>Greetings stranger!</p>\n";
}
VERSION
1.001
DESCRIPTION
FOAF+SSL (a.k.a. WebID) is a simple authentication scheme described at http://esw.w3.org/topic/foaf+ssl. This module provides FOAF+SSL authentication for CGI scripts written in Perl.
This requires the web server to be using HTTPS and to be configured to request client certificates and to pass the certificate details on as environment variables for scripts. If you are using Apache, this means that you want to set the following directives in your SSL virtual host setup:
SSLEngine on
# SSLCipherSuite (see Apache documentation)
# SSLProtocol (see Apache documentation)
# SSLCertificateFile (see Apache documentation)
# SSLCertificateKeyFile (see Apache documentation)
SSLVerifyClient optional_no_ca
SSLVerifyDepth 1
SSLOptions +StdEnvVars +ExportCertData
Configuration
Constructors
$auth = CGI::Auth::FOAF_SSL->new($pem_encoded)
-
Performs FOAF+SSL authentication on a PEM-encoded key. If authentication is completely unsuccessful, returns undef. Otherwise, returns a CGI::Auth::FOAF_SSL object. Use
is_secure
to check if authentication was completely successful.You probably want to use
new_from_cgi
instead. $auth = CGI::Auth::FOAF_SSL->new_from_cgi($cgi_object)
-
Performs FOAF+SSL authentication on a CGI object. This is a wrapper around
new
which extracts the PEM-encoded client certificate from the CGI request. It has the same return values asnew
.If $cgi_object is omitted, uses
CGI->new
instead. $auth = CGI::Auth::FOAF_SSL->new_unauthenticated($pem_encoded)
-
Creates a CGI::Auth::FOAF_SSL object without doing any authentication.
It's very unlikely you want to do this. If you do create an unauthenticated object, then you'll probably want to do some authentication using the authenticate_by_XXX methods.
Public Methods
$bool = $auth->is_secure
-
Returns true iff the authentication process was completely successful.
$agent = $auth->agent
-
Returns a CGI::Auth::FOAF_SSL::Agent object which represents the agent making the request.
-
HTTP cookie related to the authentication process. Sending this to the client isn't strictly necessary, but it allows for a session to be established, greatly speeding up subsequent accesses. See also the COOKIES section of this documentation.
$ok = $auth->authenticate_by_uri($uri)
-
Checks if $uri claims that $auth's key identifies it.
This is only relevent if you constructed $auth using
new_unauthenticated
. $ok = $auth->authenticate_by_email($email_address)
-
Checks if $email_address claims that $auth's key identifies it (via WebFinger/Fingerpoint).
This is only relevent if you constructed $auth using
new_unauthenticated
. $ok = $auth->authenticate_by_sparql($uri, $endpoint)
-
Checks if $endpoint claims that $auth's key identifies $uri. $endpoint may be a SPARQL endpoint URI or an RDF::Trine::Model.
This is only relevent if you constructed $auth using
new_unauthenticated
.
Utility Methods
$model = $auth->get_trine_model($uri)
-
Get an RDF::Trine::Model corresponding to a URI.
$bi = $auth->make_bigint_from_node($trine_node)
-
Turns an RDF::Trine::Node::Literal object into a Math::BigInt representing the same number.
There are optional named parameters for providing a fallback in the case where $trine_node has an unrecognised datatype or is not a literal.
$bi = $auth->make_bigint_from_node( $trine_node, fallback=>$other_node, fallback_type=>'hex');
The authenticate_by_XXX methods use this.
$results = $auth->execute_query($sparql)
-
Returns the results of a SPARQL query. Uses the certificate subject's RDF file as a data source, or the certificate subject's SPARQL endpoint.
See RDF::TrineShortcuts function rdf_query for an explanation of the return format.
COOKIES
FOAF+SSL is entirely RESTful: there is no state kept between requests. This really simplifies authentication for both parties (client and server) for one-off requests. However, because FOAF+SSL requires the server to make various HTTP requests to authenticate the client, each request is slowed down significantly.
Cookies provide us with a way of speeding this up. Use of cookies is entirely optional, but greatly increases the speed of authentication for the second and subsequent requests a client makes. If your FOAF+SSL-secured service generally requires clients to make multiple requests in a short period, you should seriously consider using cookies to speed this up.
The method works like this: on the first request, authentication happens as normal. However, all RDF files relevant to authenticating the client are kept on disk (usually somewhere like '/tmp') in N-Triples format. They are associated with a session that is given a randomly generated identifier. This random identifier is sent the client as a cookie. On subsequent requests, the client includes the cookie and thus CGI::Auth::FOAF_SSL is able to retrieve the data it needs from disk in N-Triples format, rather than having to reach out onto the web for it again.
To use this feature, you must perform authentication before printing anything back to the client, use CGI::Auth::FOAF_SSL's cookie
method, and then pass that to the client as part of the HTTP response header.
use CGI qw(:all);
use CGI::Auth::FOAF_SSL;
my $auth = CGI::Auth::FOAF_SSL->new_from_cgi;
if (defined $auth && $auth->is_secure)
{
print header('-type' => 'text/html',
'-cookie' => $auth->cookie);
my $user = $auth->agent;
# ...
}
else # anonymous access
{
print header('-type' => 'text/html');
# ...
}
Old sessions are automatically purged after an hour of inactivity.
BUGS
Please report any bugs to http://rt.cpan.org/.
SEE ALSO
Helper module: CGI::Auth::FOAF_SSL::Agent
Related modules: CGI, RDF::Trine, RDF::ACL.
Information about FOAF+SSL: http://lists.foaf-project.org/mailman/listinfo/foaf-protocols, http://esw.w3.org/topic/foaf+ssl.
SSL in Apache: http://httpd.apache.org/docs/2.0/mod/mod_ssl.html.
Mailing list for general Perl RDF/SemWeb discussion and support: http://www.perlrdf.org/.
AUTHOR
Toby Inkster, <tobyink@cpan.org>
COPYRIGHT AND LICENSE
Copyright (C) 2009-2010 by Toby Inkster
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8 or, at your option, any later version of Perl 5 you may have available.