NAME
Catalyst::Response - stores output responding to the current client request
SYNOPSIS
$res = $c->response;
$res->body;
$res->code;
$res->content_encoding;
$res->content_length;
$res->content_type;
$res->cookies;
$res->header;
$res->headers;
$res->output;
$res->redirect;
$res->status;
$res->write;
DESCRIPTION
This is the Catalyst Response class, which provides methods for responding to the current client request. The appropriate Catalyst::Engine for your environment will turn the Catalyst::Response into a HTTP Response and return it to the client.
METHODS
$res->body( $text | $fh | $iohandle_object )
$c->response->body('Catalyst rocks!');
Sets or returns the output (text or binary data). If you are returning a large body, you might want to use a IO::Handle type of object (Something that implements the getline method in the same fashion), or a filehandle GLOB. These will be passed down to the PSGI handler you are using and might be optimized using server specific abilities (for example Twiggy will attempt to server a real local file in a non blocking manner).
If you are using a filehandle as the body response you are responsible for making sure it conforms to the PSGI specification with regards to content encoding. Unlike with scalar body values or when using the streaming interfaces we currently do not attempt to normalize and encode your filehandle. In general this means you should be sure to be sending bytes not UTF8 decoded multibyte characters.
Most of the time when you do:
open(my $fh, '<:raw', $path);
You should be fine. If you open a filehandle with a PerlIO layer you probably are not fine. You can usually fix this by explicitly using binmode to set the IOLayer to :raw. Its possible future versions of Catalyst will try to 'do the right thing'.
When using a IO::Handle type of object and no content length has been already set in the response headers Catalyst will make a reasonable attempt to determine the size of the Handle. Depending on the implementation of your handle object, setting the content length may fail. If it is at all possible for you to determine the content length of your handle object, it is recommended that you set the content length in the response headers yourself, which will be respected and sent by Catalyst in the response.
Please note that the object needs to implement getline
, not just read
. Older versions of Catalyst expected your filehandle like objects to do read. If you have code written for this expectation and you cannot change the code to meet the PSGI specification, you can try the following middleware Plack::Middleware::AdaptFilehandleRead which will attempt to wrap your object in an interface that so conforms.
Starting from version 5.90060, when using an IO::Handle object, you may want to use Plack::Middleware::XSendfile, to delegate the actual serving to the frontend server. To do so, you need to pass to body
an IO object with a path
method. This can be achieved in two ways.
Either using Plack::Util:
my $fh = IO::File->new($file, 'r');
Plack::Util::set_io_path($fh, $file);
Or using IO::File::WithPath
my $fh = IO::File::WithPath->new($file, 'r');
And then passing the filehandle to body and setting headers, if needed.
$c->response->body($fh);
$c->response->headers->content_type('text/plain');
$c->response->headers->content_length(-s $file);
$c->response->headers->last_modified((stat($file))[9]);
Plack::Middleware::XSendfile can be loaded in the application so:
__PACKAGE__->config(
psgi_middleware => [
'XSendfile',
# other middlewares here...
],
);
Beware that loading the middleware without configuring the webserver to set the request header X-Sendfile-Type
to a supported type (X-Accel-Redirect
for nginx, X-Sendfile
for Apache and Lighttpd), could lead to the disclosure of private paths to malicious clients setting that header.
Nginx needs the additional X-Accel-Mapping header to be set in the webserver configuration, so the middleware will replace the absolute path of the IO object with the internal nginx path. This is also useful to prevent a buggy app to server random files from the filesystem, as it's an internal redirect.
An nginx configuration for FastCGI could look so:
server {
server_name example.com;
root /my/app/root;
location /private/repo/ {
internal;
alias /my/app/repo/;
}
location /private/staging/ {
internal;
alias /my/app/staging/;
}
location @proxy {
include /etc/nginx/fastcgi_params;
fastcgi_param SCRIPT_NAME '';
fastcgi_param PATH_INFO $fastcgi_script_name;
fastcgi_param HTTP_X_SENDFILE_TYPE X-Accel-Redirect;
fastcgi_param HTTP_X_ACCEL_MAPPING /my/app=/private;
fastcgi_pass unix:/my/app/run/app.sock;
}
}
In the example above, passing filehandles with a local path matching /my/app/staging or /my/app/repo will be served by nginx. Passing paths with other locations will lead to an internal server error.
Setting the body to a filehandle without the path
method bypasses the middleware completely.
For Apache and Lighttpd, the mapping doesn't apply and setting the X-Sendfile-Type is enough.
$res->has_body
Predicate which returns true when a body has been set.
$res->code
Alias for $res->status.
$res->content_encoding
Shortcut for $res->headers->content_encoding.
$res->content_length
Shortcut for $res->headers->content_length.
$res->content_type
Shortcut for $res->headers->content_type.
This value is typically set by your view or plugin. For example, Catalyst::Plugin::Static::Simple will guess the mime type based on the file it found, while Catalyst::View::TT defaults to text/html
.
$res->content_type_charset
Shortcut for $res->headers->content_type_charset;
$res->cookies
Returns a reference to a hash containing cookies to be set. The keys of the hash are the cookies' names, and their corresponding values are hash references used to construct a CGI::Simple::Cookie object.
$c->response->cookies->{foo} = { value => '123' };
The keys of the hash reference on the right correspond to the CGI::Simple::Cookie parameters of the same name, except they are used without a leading dash. Possible parameters are:
- value
- expires
- domain
- path
- secure
- httponly
$res->header
Shortcut for $res->headers->header.
$res->headers
Returns an HTTP::Headers object, which can be used to set headers.
$c->response->headers->header( 'X-Catalyst' => $Catalyst::VERSION );
$res->output
Alias for $res->body.
$res->redirect( $url, $status )
Causes the response to redirect to the specified URL. The default status is 302
.
$c->response->redirect( 'http://slashdot.org' );
$c->response->redirect( 'http://slashdot.org', 307 );
This is a convenience method that sets the Location header to the redirect destination, and then sets the response status. You will want to return
or $c->detach()
to interrupt the normal processing flow if you want the redirect to occur straight away.
Note: do not give a relative URL as $url, i.e: one that is not fully qualified (= http://...
, etc.) or that starts with a slash (= /path/here
). While it may work, it is not guaranteed to do the right thing and is not a standard behaviour. You may opt to use uri_for() or uri_for_action() instead.
Note: If $url is an object that does ->as_string (such as URI, which is what you get from ->uri_for) we automatically call that to stringify. This should ease the common case usage
return $c->res->redirect( $c->uri_for(...));
$res->location
Sets or returns the HTTP 'Location'.
$res->status
Sets or returns the HTTP status.
$c->response->status(404);
$res->code is an alias for this, to match HTTP::Response->code.
$res->write( $data )
Writes $data to the output stream. Calling this method will finalize your headers and send the headers and status code response to the client (so changing them afterwards is a waste... be sure to set your headers correctly first).
You may call this as often as you want throughout your response cycle. You may even set a 'body' afterward. So for example you might write your HTTP headers and the HEAD section of your document and then set the body from a template driven from a database. In some cases this can seem to the client as if you had a faster overall response (but note that unless your server support chunked body your content is likely to get queued anyway (Starman and most other http 1.1 webservers support this).
If there is an encoding set, we encode each line of the response (the default encoding is UTF-8).
$res->unencoded_write( $data )
Works just like ->write but we don't apply any content encoding to $data
. Use this if you are already encoding the $data or the data is arriving from an encoded storage.
$res->write_fh
Returns an instance of Catalyst::Response::Writer, which is a lightweight decorator over the PSGI $writer
object (see PSGI.pod\Delayed-Response-and-Streaming-Body).
In addition to proxying the write
and close
method from the underlying PSGI writer, this proxy object knows any application wide encoding, and provides a method write_encoded
that will properly encode your written lines based upon your encoding settings. By default in Catalyst responses are UTF-8 encoded and this is the encoding used if you respond via write_encoded
. If you want to handle encoding yourself, you can use the write
method directly.
Encoding only applies to content types for which it matters. Currently the following content types are assumed to need encoding: text (including HTML), xml and javascript.
We provide access to this object so that you can properly close over it for use in asynchronous and nonblocking applications. For example (assuming you are using a supporting server, like Twiggy:
package AsyncExample::Controller::Root;
use Moose;
BEGIN { extends 'Catalyst::Controller' }
sub prepare_cb {
my $write_fh = pop;
return sub {
my $message = shift;
$write_fh->write("Finishing: $message\n");
$write_fh->close;
};
}
sub anyevent :Local :Args(0) {
my ($self, $c) = @_;
my $cb = $self->prepare_cb($c->res->write_fh);
my $watcher;
$watcher = AnyEvent->timer(
after => 5,
cb => sub {
$cb->(scalar localtime);
undef $watcher; # cancel circular-ref
});
}
Like the 'write' method, calling this will finalize headers. Unlike 'write' when you can this it is assumed you are taking control of the response so the body is never finalized (there isn't one anyway) and you need to call the close method.
$res->print( @data )
Prints @data to the output stream, separated by $,. This lets you pass the response object to functions that want to write to an IO::Handle.
$res->finalize_headers()
Writes headers to response if not already written
from_psgi_response
Given a PSGI response (either three element ARRAY reference OR coderef expecting a $responder) set the response from it.
Properly supports streaming and delayed response and / or async IO if running under an expected event loop.
If passed an object, will expect that object to do a method as_psgi
.
Example:
package MyApp::Web::Controller::Test;
use base 'Catalyst::Controller';
use Plack::App::Directory;
my $app = Plack::App::Directory->new({ root => "/path/to/htdocs" })
->to_app;
sub myaction :Local Args {
my ($self, $c) = @_;
$c->res->from_psgi_response($app->($c->req->env));
}
sub streaming_body :Local {
my ($self, $c) = @_;
my $psgi_app = sub {
my $respond = shift;
my $writer = $respond->([200,["Content-Type" => "text/plain"]]);
$writer->write("body");
$writer->close;
};
$c->res->from_psgi_response($psgi_app);
}
Please note this does not attempt to map or nest your PSGI application under the Controller and Action namespace or path. You may wish to review 'PSGI Helpers' under Catalyst::Utils for help in properly nesting applications.
NOTE If your external PSGI application returns a response that has a character set associated with the content type (such as "text/html; charset=UTF-8") we set $c->clear_encoding to remove any additional content type encoding processing later in the application (this is done to avoid double encoding issues).
NOTE If your external PSGI application is streaming, we assume you completely handle the entire jobs (including closing the stream). This will also bypass the output finalization methods on Catalyst (such as 'finalize_body' which gets called but then skipped when it finds that output is already finished.) Its possible this might cause issue with some plugins that want to do 'things' during those finalization methods. Just understand what is happening.
encodable_content_type
This is a regular expression used to determine of the current content type should be considered encodable. Currently we apply default encoding (usually UTF8) to text type contents. Here's the default regular expression:
This would match content types like:
text/plain
text/html
text/xml
application/javascript
application/xml
application/vnd.user+xml
NOTE: We don't encode JSON content type responses by default since most of the JSON serializers that are commonly used for this task will do so automatically and we don't want to double encode. If you are not using a tool like JSON to produce JSON type content, (for example you are using a template system, or creating the strings manually) you will need to either encoding the body yourself:
$c->response->body( $c->encoding->encode( $body, $c->_encode_check ) );
Or you can alter the regular expression using this attribute.
encodable_response
Given a Catalyst::Response return true if its one that can be encoded.
make sure there is an encoding set on the response
make sure the content type is encodable
make sure no content type charset has been already set to something different from the global encoding
make sure no content encoding is present.
Note this does not inspect a body since we do allow automatic encoding on streaming type responses.
DEMOLISH
Ensures that the response is flushed and closed at the end of the request.
meta
Provided by Moose
AUTHORS
Catalyst Contributors, see Catalyst.pm
COPYRIGHT
This library is free software. You can redistribute it and/or modify it under the same terms as Perl itself.