Take me over?
NAME
Catalyst::Plugin::Session - Generic Session plugin - ties together server side storage and client side state required to maintain session data.
SYNOPSIS
# To get sessions to "just work", all you need to do is use these plugins:
use Catalyst qw/
Session
Session::Store::FastMmap
Session::State::Cookie
/;
# you can replace Store::FastMmap with Store::File - both have sensible
# default configurations (see their docs for details)
# more complicated backends are available for other scenarios (DBI storage,
# etc)
# after you've loaded the plugins you can save session data
# For example, if you are writing a shopping cart, it could be implemented
# like this:
sub add_item : Local {
my ( $self, $c ) = @_;
my $item_id = $c->req->param("item");
# $c->session is a hash ref, a bit like $c->stash
# the difference is that it' preserved across requests
push @{ $c->session->{items} }, $item_id;
$c->forward("MyView");
}
sub display_items : Local {
my ( $self, $c ) = @_;
# values in $c->session are restored
$c->stash->{items_to_display} =
[ map { MyModel->retrieve($_) } @{ $c->session->{items} } ];
$c->forward("MyView");
}
DESCRIPTION
The Session plugin is the base of two related parts of functionality required for session management in web applications.
The first part, the State, is getting the browser to repeat back a session key, so that the web application can identify the client and logically string several requests together into a session.
The second part, the Store, deals with the actual storage of information about the client. This data is stored so that the it may be revived for every request made by the same client.
This plugin links the two pieces together.
RECOMENDED BACKENDS
- Session::State::Cookie
-
The only really sane way to do state is using cookies.
- Session::Store::File
-
A portable backend, based on Cache::File.
- Session::Store::FastMmap
-
A fast and flexible backend, based on Cache::FastMmap.
METHODS
- sessionid
-
An accessor for the session ID value.
- session
-
Returns a hash reference that might contain unserialized values from previous requests in the same session, and whose modified value will be saved for future requests.
This method will automatically create a new session and session ID if none exists.
- session_expires
- session_expires $reset
-
This method returns the time when the current session will expire, or 0 if there is no current session. If there is a session and it already expired, it will delete the session and return 0 as well.
If the
$reset
parameter is true, and there is a session ID the expiry time will be reset to the current time plus the time to live (see "CONFIGURATION"). This is used when creating a new session. - flash
-
This is like Ruby on Rails' flash data structure. Think of it as a stash that lasts for longer than one request, letting you redirect instead of forward.
The flash data will be cleaned up only on requests on which actually use $c->flash (thus allowing multiple redirections), and the policy is to delete all the keys which haven't changed since the flash data was loaded at the end of every request.
sub moose : Local { my ( $self, $c ) = @_; $c->flash->{beans} = 10; $c->response->redirect( $c->uri_for("foo") ); } sub foo : Local { my ( $self, $c ) = @_; my $value = $c->flash->{beans}; # ... $c->response->redirect( $c->uri_for("bar") ); } sub bar : Local { my ( $self, $c ) = @_; if ( exists $c->flash->{beans} ) { # false } }
- clear_flash
-
Zap all the keys in the flash regardless of their current state.
- keep_flash @keys
-
If you want to keep a flash key for the next request too, even if it hasn't changed, call
keep_flash
and pass in the keys as arguments. - delete_session REASON
-
This method is used to invalidate a session. It takes an optional parameter which will be saved in
session_delete_reason
if provided.NOTE: This method will also delete your flash data.
- session_delete_reason
-
This accessor contains a string with the reason a session was deleted. Possible values include:
address mismatch
session expired
- session_expire_key $key, $ttl
-
Mark a key to expire at a certain time (only useful when shorter than the expiry time for the whole session).
For example:
__PACKAGE__->config->{session}{expires} = 1000000000000; # forever # later $c->session_expire_key( __user => 3600 );
Will make the session data survive, but the user will still be logged out after an hour.
Note that these values are not auto extended.
INTERNAL METHODS
- setup
-
This method is extended to also make calls to
check_session_plugin_requirements
andsetup_session
. - check_session_plugin_requirements
-
This method ensures that a State and a Store plugin are also in use by the application.
- setup_session
-
This method populates
$c->config->{session}
with the default values listed in "CONFIGURATION". - prepare_action
-
This method is extended.
Its only effect is if the (off by default)
flash_to_stash
configuration parameter is on - then it will copy the contents of the flash to the stash at prepare time. - finalize_headers
-
This method is extended and will extend the expiry time before sending the response.
- finalize_body
-
This method is extended and will call finalize_session before the other finalize_body methods run. Here we persist the session data if a session exists.
- initialize_session_data
-
This method will initialize the internal structure of the session, and is called by the
session
method if appropriate. - create_session_id
-
Creates a new session ID using
generate_session_id
if there is no session ID yet. - validate_session_id SID
-
Make sure a session ID is of the right format.
This currently ensures that the session ID string is any amount of case insensitive hexadecimal characters.
- generate_session_id
-
This method will return a string that can be used as a session ID. It is supposed to be a reasonably random string with enough bits to prevent collision. It basically takes
session_hash_seed
and hashes it using SHA-1, MD5 or SHA-256, depending on the availability of these modules. - session_hash_seed
-
This method is actually rather internal to generate_session_id, but should be overridable in case you want to provide more random data.
Currently it returns a concatenated string which contains:
A counter
The current time
One value from
rand
.The stringified value of a newly allocated hash reference
The stringified value of the Catalyst context object
in the hopes that those combined values are entropic enough for most uses. If this is not the case you can replace
session_hash_seed
with e.g.sub session_hash_seed { open my $fh, "<", "/dev/random"; read $fh, my $bytes, 20; close $fh; return $bytes; }
Or even more directly, replace
generate_session_id
:sub generate_session_id { open my $fh, "<", "/dev/random"; read $fh, my $bytes, 20; close $fh; return unpack("H*", $bytes); }
Also have a look at Crypt::Random and the various openssl bindings - these modules provide APIs for cryptographically secure random data.
- finalize_session
-
Clean up the session during
finalize
.This clears the various accessors after saving to the store.
- dump_these
-
See "dump_these" in Catalyst - ammends the session data structure to the list of dumped objects if session ID is defined.
- calculate_extended_session_expires
- calculate_initial_session_expires
- create_session_id_if_needed
- delete_session_id
- extend_session_expires
- extend_session_id
- get_session_id
- reset_session_expires
- session_is_valid
- set_session_id
USING SESSIONS DURING PREPARE
The earliest point in time at which you may use the session data is after Catalyst::Plugin::Session's prepare_action
has finished.
State plugins must set $c->session ID before prepare_action
, and during prepare_action
Catalyst::Plugin::Session will actually load the data from the store.
sub prepare_action {
my $c = shift;
# don't touch $c->session yet!
$c->NEXT::prepare_action( @_ );
$c->session; # this is OK
$c->sessionid; # this is also OK
}
CONFIGURATION
$c->config->{session} = {
expires => 1234,
};
All configuation parameters are provided in a hash reference under the session
key in the configuration hash.
- expires
-
The time-to-live of each session, expressed in seconds. Defaults to 7200 (two hours).
- verify_address
-
When true,
<$c-
request->address>> will be checked at prepare time. If it is not the same as the address that initiated the session, the session is deleted.Defaults to false.
- verify_user_agent
-
When true,
<$c-
request->user_agent>> will be checked at prepare time. If it is not the same as the user agent that initiated the session, the session is deleted.Defaults to false.
- flash_to_stash
-
This option makes it easier to have actions behave the same whether they were forwarded to or redirected to. On prepare time it copies the contents of
flash
(if any) to the stash.
SPECIAL KEYS
The hash reference returned by $c->session
contains several keys which are automatically set:
- __expires
-
This key no longer exists. Use
session_expires
instead. - __updated
-
The last time a session was saved to the store.
- __created
-
The time when the session was first created.
- __address
-
The value of
$c->request->address
at the time the session was created. This value is only populated ifverify_address
is true in the configuration. - __user_agent
-
The value of
$c->request->user_agent
at the time the session was created. This value is only populated ifverify_user_agent
is true in the configuration.
CAVEATS
Round the Robin Proxies
verify_address
could make your site inaccessible to users who are behind load balanced proxies. Some ISPs may give a different IP to each request by the same client due to this type of proxying. If addresses are verified these users' sessions cannot persist.
To let these users access your site you can either disable address verification as a whole, or provide a checkbox in the login dialog that tells the server that it's OK for the address of the client to change. When the server sees that this box is checked it should delete the __address
special key from the session hash when the hash is first created.
Race Conditions
In this day and age where cleaning detergents and Dutch football (not the American kind) teams roam the plains in great numbers, requests may happen simultaneously. This means that there is some risk of session data being overwritten, like this:
request a starts, request b starts, with the same session ID
session data is loaded in request a
session data is loaded in request b
session data is changed in request a
request a finishes, session data is updated and written to store
request b finishes, session data is updated and written to store, overwriting changes by request a
If this is a concern in your application, a soon-to-be-developed locking solution is the only safe way to go. This will have a bigger overhead.
For applications where any given user is only making one request at a time this plugin should be safe enough.
AUTHORS
Andy Grundman
Christian Hansen
Yuval Kogman, nothingmuch@woobling.org
Sebastian Riedel
Tomas Doran (t0m) bobtfish@bobtfish.net
(current maintainer)
Sergio Salvi
And countless other contributers from #catalyst. Thanks guys!
COPYRIGHT & LICENSE
Copyright (c) 2005 the aforementioned authors. All rights
reserved. This program is free software; you can redistribute
it and/or modify it under the same terms as Perl itself.