Plack::Request - Portable HTTP request object from PSGI env hash

  use Plack::Request;

  my $app_or_middleware = sub {
      my $env = shift; # PSGI env

      my $req = Plack::Request->new($env);

      my $path_info = $req->path_info;
      my $query     = $req->param('query');

      my $res = $req->new_response(200); # new Plack::Response
      $res->finalize;
  };

Plack::Request provides a consistent API for request objects across web server environments.

Note that this module is intended to be used by Plack middleware developers and web application framework developers rather than application developers (end users).

Writing your web application directly using Plack::Request is certainly possible but not recommended: it's like doing so with mod_perl's Apache::Request: yet too low level.

If you're writing a web application, not a framework, then you're encouraged to use one of the web application frameworks that support PSGI, or see modules like HTTP::Engine to provide higher level Request and Response API on top of PSGI.

Some of the methods defined in the earlier versions are deprecated in version 0.99. Take a look at "INCOMPATIBILITIES".

Unless otherwise noted, all methods and attributes are read-only, and passing values to the method like an accessor doesn't work like you expect it to.

    Plack::Request->new( $env );

Creates a new request object.

env

Returns the shared PSGI environment hash reference. This is a reference, so writing to this environment passes through during the whole PSGI request/response cycle.

address

Returns the IP address of the client (REMOTE_ADDR).

remote_host

Returns the remote host (REMOTE_HOST) of the client. It may be empty, in which case you have to get the IP address using address method and resolve on your own.

method

Contains the request method (GET, POST, HEAD, etc).

protocol

Returns the protocol (HTTP/1.0 or HTTP/1.1) used for the current request.

request_uri

Returns the raw, undecoded request URI path. You probably do NOT want to use this to dispatch requests.

path_info

Returns PATH_INFO in the environment. Use this to get the local path for the requests.

path

Similar to path_info but returns / in case it is empty. In other words, it returns the virtual path of the request URI after $req->base. See "DISPATCHING" for details.

script_name

Returns SCRIPT_NAME in the environment. This is the absolute path where your application is hosted.

scheme

Returns the scheme (http or https) of the request.

secure

Returns true or false, indicating whether the connection is secure (https).

body, input

Returns psgi.input handle.

session

Returns (optional) psgix.session hash. When it exists, you can retrieve and store per-session data from and to this hash.

session_options

Returns (optional) psgix.session.options hash.

logger

Returns (optional) psgix.logger code reference. When it exists, your application is supposed to send the log message to this logger, using:

  $req->logger->({ level => 'debug', message => "This is a debug message" });

cookies

Returns a reference to a hash containing the cookies. Values are strings that are sent by clients and are URI decoded.

query_parameters

Returns a reference to a hash containing query string (GET) parameters. This hash reference is Hash::MultiValue object.

body_parameters

Returns a reference to a hash containing posted parameters in the request body (POST). As with query_parameters, the hash reference is a Hash::MultiValue object.

parameters

Returns a Hash::MultiValue hash reference containing (merged) GET and POST parameters.

content, raw_body

Returns the request content in an undecoded byte string for POST requests.

uri

Returns an URI object for the current request. The URI is constructed using various environment values such as SCRIPT_NAME, PATH_INFO, QUERY_STRING, HTTP_HOST, SERVER_NAME and SERVER_PORT.

Every time this method is called it returns a new, cloned URI object.

base

Returns an URI object for the base path of current request. This is like uri but only contains up to SCRIPT_NAME where your application is hosted at.

Every time this method is called it returns a new, cloned URI object.

user

Returns REMOTE_USER if it's set.

headers

Returns an HTTP::Headers object containing the headers for the current request.

uploads

Returns a reference to a hash containing uploads. The hash reference is a Hash::MultiValue object and values are Plack::Request::Upload objects.

content_encoding

Shortcut to $req->headers->content_encoding.

content_length

Shortcut to $req->headers->content_length.

content_type

Shortcut to $req->headers->content_type.

header

Shortcut to $req->headers->header.

referer

Shortcut to $req->headers->referer.

user_agent

Shortcut to $req->headers->user_agent.

param

Returns GET and POST parameters with a CGI.pm-compatible param method. This is an alternative method for accessing parameters in $req->parameters. Unlike CGI.pm, it does not allow setting or modifying query parameters.

    $value  = $req->param( 'foo' );
    @values = $req->param( 'foo' );
    @params = $req->param;

upload

A convenient method to access $req->uploads.

    $upload  = $req->upload('field');
    @uploads = $req->upload('field');
    @fields  = $req->upload;

    for my $upload ( $req->upload('field') ) {
        print $upload->filename;
    }

new_response

  my $res = $req->new_response;

Creates a new Plack::Response object. Handy to remove dependency on Plack::Response in your code for easy subclassing and duck typing in web application frameworks, as well as overriding Response generation in middlewares.

Parameters that can take one or multiple values (i.e. parameters, query_parameters, body_parameters and uploads) store the hash reference as a Hash::MultiValue object. This means you can use the hash reference as a plain hash where values are always scalars (NOT array references), so you don't need to code ugly and unsafe ref ... eq 'ARRAY' anymore.

And if you explicitly want to get multiple values of the same key, you can call the get_all method on it, such as:

  my @foo = $req->query_parameters->get_all('foo');

You can also call get_one to always get one parameter independent of the context (unlike param), and even call mixed (with Hash::MultiValue 0.05 or later) to get the traditional hash reference,

  my $params = $req->parameters->mixed;

where values are either a scalar or an array reference depending on input, so it might be useful if you already have the code to deal with that ugliness.

The methods to parse request body (content, body_parameters and uploads) are carefully coded to save the parsed body in the environment hash as well as in the temporary buffer, so you can call them multiple times and create Plack::Request objects multiple times in a request and they should work safely, and won't parse request body more than twice for the efficiency.

If your application or framework wants to dispatch (or route) actions based on request paths, be sure to use $req->path_info not $req->uri->path.

This is because path_info gives you the virtual path of the request, regardless of how your application is mounted. If your application is hosted with mod_perl or CGI scripts, or even multiplexed with tools like Plack::App::URLMap, request's path_info always gives you the action path.

Note that path_info might give you an empty string, in which case you should assume that the path is /.

You will also want to use $req->base as a base prefix when building URLs in your templates or in redirections. It's a good idea for you to subclass Plack::Request and define methods such as:

  sub uri_for {
      my($self, $path, $args) = @_;
      my $uri = $self->base;
      $uri->path($uri->path . $path);
      $uri->query_form(@$args) if $args;
      $uri;
  }

So you can say:

  my $link = $req->uri_for('/logout', [ signoff => 1 ]);

and if $req->base is /app you'll get the full URI for /app/logout?signoff=1.

In version 0.99, many utility methods are removed or deprecated, and most methods are made read-only.

The following methods are deprecated: hostname, url_scheme, params, query_params, body_params, cookie and raw_uri. They will be removed in the next major release.

All parameter-related methods such as parameters, body_parameters, query_parameters and uploads now contains Hash::MultiValue objects, rather than scalar or an array reference depending on the user input which is insecure. See Hash::MultiValue for more about this change.

$req->path method had a bug, where the code and the document was mismatching. The document was suggesting it returns the sub request path after $req->base but the code was always returning the absolute URI path. The code is now updated to be an alias of $req->path_info but returns / in case it's empty. If you need the older behavior, just call $req->uri->path instead.

Cookie handling is simplified, and doesn't use CGI::Simple::Cookie anymore, which means you CAN NOT set array reference or hash reference as a cookie value and expect it be serialized. You're always required to set string value, and encoding or decoding them is totally up to your application or framework. Also, cookies hash reference now returns strings for the cookies rather than CGI::Simple::Cookie objects, which means you no longer have to write a wacky code such as:

  $v = $req->cookie->{foo} ? $req->cookie->{foo}->value : undef;

and instead, simply do:

  $v = $req->cookie->{foo};

Tatsuhiko Miyagawa

Kazuhiro Osawa

Tokuhiro Matsuno

Plack::Response HTTP::Request, Catalyst::Request

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.