NAME

WWW::Mechanize::Plugin::DOM - HTML Document Object Model plugin for Mech

VERSION

0.004 (alpha)

SYNOPSIS

use WWW::Mechanize;

my $m = new WWW::Mechanize;

$m->use_plugin('DOM',
    script_handlers => {
        default => \&script_handler,
        qr/(?:^|\/)(?:x-)?javascript/ => \&script_handler,
    },
    event_attr_handlers => {
        default => \&event_attr_handler,
        qr/(?:^|\/)(?:x-)?javascript/ => \&event_attr_handler,
    },
);

sub script_handler {
        my($mech, $dom_tree, $code, $url, $line, $is_inline) = @_;
        # ... code to run the script ...
}

sub event_attr_handler {
        my($mech, $elem, $event_name, $code, $url, $line) = @_;
        # ... code that returns a coderef ...
}

$m->plugin('DOM')->tree; # DOM tree for the current page

DESCRIPTION

This is a plugin for WWW::Mechanize that provides support for the HTML Document Object Model. This is a part of the WWW::Mechanize::Plugin::JavaScript distribution, but it can be used on its own.

USAGE

To enable this plugin, use Mech's use_plugin method, as shown in the synopsis.

To access the DOM tree, use $mech->plugin('DOM')->tree, which returns an HTML::DOM object.

You may provide a subroutine that runs an inline script like this:

$mech->use_plugin('DOM',
    script_handlers => {
        qr/.../ => sub { ... },
        qr/.../ => sub { ... },
        # etc
    }
);

And a subroutine for turning HTML event attributes into subroutines, like this:

$mech->use_plugin('DOM',
    event_attr_handlers => {
        qr/.../ => sub { ... },
        qr/.../ => sub { ... },
        # etc
   }
);

In both cases, the qr/.../ should be a regular expression that matches the scripting language to which the handler applies, or the string 'default'. The scripting language will be either a MIME type or the contents of the language attribute if a script element's type attribute is not present. The subroutine specified as the 'default' will be used if there is no handler for the scripting language in question or if there is no Content-Script-Type header and, for script_handlers, the script element has no 'type' or 'language' attribute.

Each time you move to another page with WWW::Mechanize, a different copy of the DOM plugin object is created. So, if you must refer to it in a callback routine, don't use a closure, but get it from the $mech object that is passed as the first argument.

The line number passed to an event attribute handler requires HTML::DOM 0.012 or higher. It will be undef will lower versions.

THE 'LOAD' EVENT

Currently the (on)load event is triggered when the page finishes parsing. This plugin assumes that you're not going to be loading any images, etc.

PREREQUISITES

HTML::DOM 0.010 or later (0.012 or higher recommended)

WWW::Mechanize

The current stable release of WWW::Mechanize does not support plugins. See WWW::Mechanize::Plugin::JavaScript for more info.

BUGS

• The onunload event is not yet supported. The window object cannot yet be used as an EventTarget (actually, it's not currently part of the DOM plugin, but part of the JS plugin; this will change, though). Some events do not yet do everything they are supposed to; e.g., a link's click method does not go to the next page.

• This plugin does not yet provide WWW::Mechanize with all the necessary callback routines (for extract_images, etc.).

• Currently, external scripts referenced within a page are always read as Latin-1. This will be fixed.

SEE ALSO

WWW::Mechanize::Plugin::JavaScript

WWW::Mechanize

HTML::DOM