NAME

       HTTP::Async - process multiple HTTP requests in parallel without blocking.

SYNOPSIS

       Create an object and add some requests to it:

           use HTTP::Async;
           my $async = HTTP::Async->new;

           # create some requests and add them to the queue.
           $async->add( HTTP::Request->new( GET => 'http://www.perl.org/'         ) );
           $async->add( HTTP::Request->new( GET => 'http://www.ecclestoad.co.uk/' ) );

       and then EITHER process the responses as they come back:

           while ( my $response = $async->wait_for_next_response ) {
               # Do some processing with $response
           }

       OR do something else if there is no response ready:

           while ( $async->not_empty ) {
               if ( my $response = $async->next_response ) {
                   # deal with $response
               } else {
                   # do something else
               }
           }

       OR just use the async object to fetch stuff in the background and deal with the responses
       at the end.

           # Do some long code...
           for ( 1 .. 100 ) {
             some_function();
             $async->poke;            # lets it check for incoming data.
           }

           while ( my $response = $async->wait_for_next_response ) {
               # Do some processing with $response
           }

DESCRIPTION

       Although using the conventional "LWP::UserAgent" is fast and easy it does have some
       drawbacks - the code execution blocks until the request has been completed and it is only
       possible to process one request at a time.  "HTTP::Async" attempts to address these
       limitations.

       It gives you a 'Async' object that you can add requests to, and then get the requests off
       as they finish. The actual sending and receiving of the requests is abstracted. As soon as
       you add a request it is transmitted, if there are too many requests in progress at the
       moment they are queued. There is no concept of starting or stopping - it runs
       continuously.

       Whilst it is waiting to receive data it returns control to the code that called it meaning
       that you can carry out processing whilst fetching data from the network. All without
       forking or threading - it is actually done using "select" lists.

Default settings:

       There are a number of default settings that should be suitable for most uses.  However in
       some circumstances you might wish to change these.

                   slots:  20
                 timeout:  180 (seconds)
        max_request_time:  300 (seconds)
           max_redirects:  7
           poll_interval:  0.05 (seconds)
              proxy_host:  ''
              proxy_port:  ''
              local_addr:  ''
              local_port:  ''
              ssl_options: {}

METHODS

   new
           my $async = HTTP::Async->new( %args );

       Creates a new HTTP::Async object and sets it up. Variations from the default can be set by
       passing them in as %args.

   slots, timeout, max_request_time, poll_interval, max_redirects, proxy_host, proxy_port,
       local_addr, local_port, ssl_options
           $old_value = $async->slots;
           $new_value = $async->slots( $new_value );

       Get/setters for the $async objects config settings. Timeout is for inactivity and is in
       seconds.

       Slots is the maximum number of parallel requests to make.

   add
           my @ids      = $async->add(@requests);
           my $first_id = $async->add(@requests);

       Adds requests to the queues. Each request is given an unique integer id (for this $async)
       that can be used to track the requests if needed. If called in list context an array of
       ids is returned, in scalar context the id of the first request added is returned.

   add_with_opts
           my $id = $async->add_with_opts( $request, \%opts );

       This method lets you add a single request to the queue with options that differ from the
       defaults. For example you might wish to set a longer timeout or to use a specific proxy.
       Returns the id of the request.

   poke
           $async->poke;

       At fairly frequent intervals some housekeeping needs to performed - such as reading
       received data and starting new requests. Calling "poke" lets the object do this and then
       return quickly. Usually you will not need to use this as most other methods do it for you.

       You should use "poke" if your code is spending time elsewhere (ie not using the async
       object) to allow it to keep the data flowing over the network. If it is not used then the
       buffers may fill up and completed responses will not be replaced with new requests.

   next_response
           my $response          = $async->next_response;
           my ( $response, $id ) = $async->next_response;

       Returns the next response (as a HTTP::Response object) that is waiting, or returns undef
       if there is none. In list context it returns a (response, id) pair, or an empty list if
       none. Does not wait for a response so returns very quickly.

   wait_for_next_response
           my $response          = $async->wait_for_next_response( 3.5 );
           my ( $response, $id ) = $async->wait_for_next_response( 3.5 );

       As "next_response" but only returns if there is a next response or the time in seconds
       passed in has elapsed. If no time is given then it blocks. Whilst waiting it checks the
       queues every c<poll_interval> seconds. The times can be fractional seconds.

   to_send_count, to_return_count, in_progress_count and total_count
           my $pending = $async->to_send_count;

       Returns the number of items in the various stages of processing.

   info
           print $async->info;

       Prints a line describing what the current state is.

   empty, not_empty
           while ( $async->not_empty ) { ...; }
           while (1) { ...; last if $async->empty; }

       Returns true or false depending on whether there are request or responses still on the
       object.

   DESTROY
       The destroy method croaks if an object is destroyed but is not empty. This is to help with
       debugging.

SEE ALSO

       HTTP::Async::Polite - a polite form of this module. Slows the scraping down by domain so
       that the remote server is not overloaded.

GOTCHAS

       The responses may not come back in the same order as the requests were made.  For https
       requests to work, you must have Net::HTTPS::NB installed.

THANKS

       Egor Egorov contributed patches for proxies, catching connections that die before headers
       sent and more.

       Tomohiro Ikebe from livedoor.jp submitted patches (and a test) to properly handle 304
       responses.

       Naveed Massjouni for adding the https handling code.

       Alex Balhatchet for adding the https + proxy handling code, and for making the tests run
       ok in parallel.

       Josef Toman for fixing two bugs, one related to header handling and another related to
       producing an absolute URL correctly.

       Github user 'c00ler-' for adding LocalAddr and LocalPort support.

       rt.cpan.org user 'Florian (fschlich)' for typo in documentation.

       Heikki Vatiainen for the ssl_options support patch.

BUGS AND REPO

       Please submit all bugs, patches etc on github

       <https://github.com/evdb/HTTP-Async>

AUTHOR

       Edmund von der Burg "<evdb@ecclestoad.co.uk>".

       <http://www.ecclestoad.co.uk/>

LICENCE AND COPYRIGHT

       Copyright (c) 2006, Edmund von der Burg "<evdb@ecclestoad.co.uk>".  All rights reserved.

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

DISCLAIMER OF WARRANTY

       BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE SOFTWARE,
       TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE
       COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF
       ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
       WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO
       THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH YOU. SHOULD THE SOFTWARE PROVE
       DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR, OR CORRECTION.

       IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT
       HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE SOFTWARE AS PERMITTED BY
       THE ABOVE LICENCE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL,
       INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE
       SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR
       LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE SOFTWARE TO OPERATE WITH ANY
       OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
       SUCH DAMAGES.