HTTP::GHTTP - Perl interface to the gnome ghttp library

  use HTTP::GHTTP;

  my $r = HTTP::GHTTP->new();
  $r->set_uri("http://axkit.org/");
  $r->process_request;
  print $r->get_body;

This is a fairly low level interface to the Gnome project's libghttp, which allows you to process HTTP requests to HTTP servers. There also exists a slightly higher level interface - a simple get() function which takes a URI as a parameter. This is not exported by default, you have to ask for it explicitly.

Constructor function - creates a new GHTTP object. If supplied a URI it will automatically call set_uri for you. If you also supply a list of key/value pairs it will set those as headers:

    my $r = HTTP::GHTTP->new(
        "http://axkit.com/",
        Connection => "close");

This sets the URI for the request This sets an outgoing HTTP request header

This sets the request type. The request types themselves are constants that are not exported by default. To export them, specify the :methods option in the import list:

    use HTTP::GHTTP qw/:methods/;
    my $r = HTTP::GHTTP->new();
    $r->set_uri('http://axkit.com/');
    $r->set_type(METHOD_HEAD);
    ...

The available methods are:

    METHOD_GET
    METHOD_POST
    METHOD_OPTIONS
    METHOD_HEAD
    METHOD_PUT
    METHOD_DELETE
    METHOD_TRACE
    METHOD_CONNECT
    METHOD_PROPFIND
    METHOD_PROPPATCH
    METHOD_MKCOL
    METHOD_COPY
    METHOD_MOVE
    METHOD_LOCK
    METHOD_UNLOCK

Some of these are for DAV.

This sets the body of a request, useful in POST and some of the DAV request types.

This sends the actual request to the server

This returns 2 values, a status code (numeric) and a status reason phrase. A simple example of the return values would be (200, OK).

This gets the value of an incoming HTTP response header

Returns a list of all the response header names in the order they came back. This method is only available in libghttp 1.08 and later - perl Makefile.PL should have reported whether it found it or not.

  my @headers = $r->get_headers;
  print join("\n", 
        map { "$_: " . $r->get_header($_) } @headers), "\n\n";

This gets the body of the response

If the response failed for some reason, this returns a textual error This sets an outgoing username and password for simple HTTP authentication

This sets your proxy server, use the form http://proxy:port If you have set a proxy and your proxy requires a username and password you can set it with this.

This is a low level interface useful only when doing async downloads. See ASYNC OPERATION.

This is a low level interface useful only when doing async downloads. See ASYNC OPERATION.

process returns undef for error, 1 for in progress, and zero for complete.

Returns an IO::Handle object that is the currently in progress socket. Useful only when doing async downloads. There appears to be some corruption when using the socket to retrieve file contents on more recent libghttp's.

This is only useful in async mode. It returns 3 values: The current processing stage (0 = none, 1 = request, 2 = response headers, 3 = response), the number of bytes read, and the number of bytes total.

This turns async mode on. There is no corresponding unset function.

Sets the download (and upload) chunk size in bytes for use in async mode. This may be a useful value to set for slow modems, or perhaps for a download progress bar, or just to allow periodic writes.

This does everything automatically for you, retrieving the body at the remote URI. Optionally pass in headers.

Its possible to use an asynchronous mode of operation with ghttp. Here's a brief example of how:

    my $r = HTTP::GHTTP->new("http://axkit.org/");
    $r->set_async; 
    $r->set_chunksize(1);
    $r->prepare;

    my $status;
    while ($status = $r->process) {
        # do something
        # you can do $r->get_body in here if you want to
        # but it always returns the entire body.
    }

    die "An error occured" unless defined $status;

    print $r->get_body;

Doing timeouts is an exercise for the reader (hint: lookup select() in perlfunc).

Note also that CW$sock above is an IO::Handle, not an IO::Socket, although you can probably get away with re-blessing it. Also note that by calling CW$r->get_socket() you load IO::Handle, which probably brings a lot of code with it, thereby obliterating a lot of the use for libghttp. So use at your own risk :-)

Matt Sergeant, matt@sergeant.org

This is free software, you may use it and distribute it under the same terms as Perl itself. Please be aware though that libghttp is licensed under the terms of the LGPL, a copy of which can be found in the libghttp distribution.

Probably many - this is my first adventure into XS.

libghttp doesn't support SSL. When libghttp does support SSL, so will HTTP::GHTTP. The author of libghttp, Chris Blizzard <blizzard@redhat.com> is looking for patches to support SSL, so get coding!

Benchmarking this sort of thing is often difficult, and I don't want to offend anyone. But as well as being lightweight (HTTP::GHTTP is about 4 times less code than either LWP::UserAgent, or HTTP::Lite), it is also in my tests significantly faster. Here are my benchmark results requesting http://localhost/ (the Apache Successful Install page):

    Benchmark: timing 1000 iterations of ghttp, lite, lwp...
         ghttp:  8 wallclock secs ( 0.96 usr +  1.16 sys =  2.12 CPU)
          lite: 21 wallclock secs ( 3.00 usr +  3.44 sys =  6.44 CPU)
           lwp: 18 wallclock secs ( 9.76 usr +  1.59 sys = 11.35 CPU)