>
> Now if we can just figure out a logical way to lash all
> this together. . .
I've been working on a Web::Browser module to be a "tie together" class for
LWP::UserAgent and JavaScript.pm. It is intended to be a high-level
interface to web automation, plus an easier way to manipulate web page
contents using a well-known interface (the same one used by javascript).
It will require two more modules at least: an implementation of the HTML
Document Object Model and a set of bindings from the DOM to JavaScript.
My initial code can do things like:
$browser->document->forms(0)->set_value('username' => 'me');
$browser->document->forms(0)->submit();
or
$browser->document->link(14)->click();
which will load the page as the new document in $browser, but it is hacked
together (from HTML::Form, HTML::LinkExtor, HTML::TreeBuilder, etc) and
inflexible, so I started working on an HTML::DOM module built from the W3C
Document Object Model (HTML) Level 1 specification to inherit from XML::DOM
[which is based on the DOM (Core) Level 1]. I also have some code which
"publishes" part of the HTMLDocument object to a JavaScript.pm created
runtime/context and manipulates it from JavaScript code.
Do you think this is a good way to go about it? When it's done, I expect to
have 6 core modules sharing the work:
Web::Browser # High level interaction / Controller
LWP::UserAgent # (The real work) HTTP / Network manipulation
HTML::TreeBuilder # HTML Parsing into HTML::DOM
HTML::DOM # Document representation
HTML::DOM::JSBindings # JavaScript links to document
JavaScript # JavaScript processing engine
Three of them are already available on CPAN, so I'm currently focusing on
implementing the HTML::DOM module.
Any comments?
-Tom
