NAME
HTTP::Server::Simple - Lightweight HTTP server
SYNOPSIS
use warnings;
use strict;
use HTTP::Server::Simple;
my $server = HTTP::Server::Simple->new();
$server->run();
However, normally you will sub-class the HTTP::Server::Simple::CGI module (see HTTP::Server::Simple::CGI);
package Your::Web::Server;
use base qw(HTTP::Server::Simple::CGI);
sub handle_request {
my ($self, $cgi) = @_;
#... do something, print output to default
# selected filehandle...
}
1;
DESCRIPTION
This is a simple standalone HTTP server. By default, it doesn't thread or fork. It does, however, act as a simple frontend which can be used to build a standalone web-based application or turn a CGI into one.
It is possible to use Net::Server classes to create forking, pre-forking, and other types of more complicated servers; see "net_server".
By default, the server traps a few signals:
- HUP
-
When you
kill -HUP
the server, it lets the current request finish being processed, then uses therestart
method to re-exec itself. Please note that in order to provide restart-on-SIGHUP, HTTP::Server::Simple sets a SIGHUP handler during initialisation. If your request handling code forks you need to make sure you reset this or unexpected things will happen if somebody sends a HUP to all running processes spawned by your app (e.g. by "kill -HUP <script>") - PIPE
-
If the server detects a broken pipe while writing output to the client, it ignores the signal. Otherwise, a client closing the connection early could kill the server.
EXAMPLE
#!/usr/bin/perl
{
package MyWebServer;
use HTTP::Server::Simple::CGI;
use base qw(HTTP::Server::Simple::CGI);
my %dispatch = (
'/hello' => \&resp_hello,
# ...
);
sub handle_request {
my $self = shift;
my $cgi = shift;
my $path = $cgi->path_info();
my $handler = $dispatch{$path};
if (ref($handler) eq "CODE") {
print "HTTP/1.0 200 OK\r\n";
$handler->($cgi);
} else {
print "HTTP/1.0 404 Not found\r\n";
print $cgi->header,
$cgi->start_html('Not found'),
$cgi->h1('Not found'),
$cgi->end_html;
}
}
sub resp_hello {
my $cgi = shift; # CGI.pm object
return if !ref $cgi;
my $who = $cgi->param('name');
print $cgi->header,
$cgi->start_html("Hello"),
$cgi->h1("Hello $who!"),
$cgi->end_html;
}
}
# start the server on port 8080
my $pid = MyWebServer->new(8080)->background();
print "Use 'kill $pid' to stop server.\n";
METHODS
HTTP::Server::Simple->new($port)
API call to start a new server. Does not actually start listening until you call ->run()
. If omitted, $port
defaults to 8080.
lookup_localhost
Looks up the local host's IP address, and returns it. For most hosts, this is 127.0.0.1
.
port [NUMBER]
Takes an optional port number for this server to listen on.
Returns this server's port. (Defaults to 8080)
host [address]
Takes an optional host address for this server to bind to.
Returns this server's bound address (if any). Defaults to undef
(bind to all interfaces).
background [ARGUMENTS]
Runs the server in the background, and returns the process ID of the started process. Any arguments will be passed through to "run".
run [ARGUMENTS]
Run the server. If all goes well, this won't ever return, but it will start listening for HTTP
requests. Any arguments passed to this will be passed on to the underlying Net::Server implementation, if one is used (see "net_server").
net_server
User-overridable method. If you set it to a Net::Server subclass, that subclass is used for the run
method. Otherwise, a minimal implementation is used as default.
restart
Restarts the server. Usually called by a HUP signal, not directly.
stdio_handle [FILEHANDLE]
When called with an argument, sets the socket to the server to that arg.
Returns the socket to the server; you should only use this for actual socket-related calls like getsockname
. If all you want is to read or write to the socket, you should use stdin_handle
and stdout_handle
to get the in and out filehandles explicitly.
stdin_handle
Returns a filehandle used for input from the client. By default, returns whatever was set with stdio_handle
, but a subclass could do something interesting here.
stdout_handle
Returns a filehandle used for output to the client. By default, returns whatever was set with stdio_handle
, but a subclass could do something interesting here.
IMPORTANT SUB-CLASS METHODS
A selection of these methods should be provided by sub-classes of this module.
handler
This method is called after setup, with no parameters. It should print a valid, full HTTP response to the default selected filehandle.
setup(name => $value, ...)
This method is called with a name => value list of various things to do with the request. This list is given below.
The default setup handler simply tries to call methods with the names of keys of this list.
ITEM/METHOD Set to Example
----------- ------------------ ------------------------
method Request Method "GET", "POST", "HEAD"
protocol HTTP version "HTTP/1.1"
request_uri Complete Request URI "/foobar/baz?foo=bar"
path Path part of URI "/foobar/baz"
query_string Query String undef, "foo=bar"
port Received Port 80, 8080
peername Remote name "200.2.4.5", "foo.com"
peeraddr Remote address "200.2.4.5", "::1"
peerport Remote port 42424
localname Local interface "localhost", "myhost.com"
headers([Header => $value, ...])
Receives HTTP headers and does something useful with them. This is called by the default setup()
method.
You have lots of options when it comes to how you receive headers.
You can, if you really want, define parse_headers()
and parse them raw yourself.
Secondly, you can intercept them very slightly cooked via the setup()
method, above.
Thirdly, you can leave the setup()
header as-is (or calling the superclass setup()
for unknown request items). Then you can define headers()
in your sub-class and receive them all at once.
Finally, you can define handlers to receive individual HTTP headers. This can be useful for very simple SOAP servers (to name a crack-fueled standard that defines its own special HTTP headers).
To do so, you'll want to define the header()
method in your subclass. That method will be handed a (key,value) pair of the header name and the value.
accept_hook
If defined by a sub-class, this method is called directly after an accept happens. An accept_hook to add SSL support might look like this:
sub accept_hook {
my $self = shift;
my $fh = $self->stdio_handle;
$self->SUPER::accept_hook(@_);
my $newfh =
IO::Socket::SSL->start_SSL( $fh,
SSL_server => 1,
SSL_use_cert => 1,
SSL_cert_file => 'myserver.crt',
SSL_key_file => 'myserver.key',
)
or warn "problem setting up SSL socket: " . IO::Socket::SSL::errstr();
$self->stdio_handle($newfh) if $newfh;
}
post_setup_hook
If defined by a sub-class, this method is called after all setup has finished, before the handler method.
print_banner
This routine prints a banner before the server request-handling loop starts.
Methods below this point are probably not terribly useful to define yourself in subclasses.
parse_request
Parse the HTTP request line. Returns three values, the request method, request URI and the protocol.
parse_headers
Parses incoming HTTP headers from STDIN, and returns an arrayref of (header => value)
pairs. See "headers" for possibilities on how to inspect headers.
setup_listener
This routine binds the server to a port and interface.
after_setup_listener
This method is called immediately after setup_listener. It's here just for you to override.
bad_request
This method should print a valid HTTP response that says that the request was invalid.
valid_http_method($method)
Given a candidate HTTP method in $method, determine if it is valid. Override if, for example, you'd like to do some WebDAV. The default implementation only accepts GET
, POST
, HEAD
, PUT
, and DELETE
.
AUTHOR
Copyright (c) 2004-2008 Jesse Vincent, <jesse@bestpractical.com>. All rights reserved.
Marcus Ramberg <drave@thefeed.no> contributed tests, cleanup, etc
Sam Vilain, <samv@cpan.org> contributed the CGI.pm split-out and header/setup API.
Example section by almut on perlmonks, suggested by Mark Fuller.
BUGS
There certainly are some. Please report them via rt.cpan.org
LICENSE
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.