John
You wrote in message news:
[hidden email]...
> Hi,
>
> I wanted to access a web page programmatically but I'm not sure how
> to do this.
>
> IWebBrowser2 new navigate: 'www.google.com'; document
>
> returns invalid call on
>
> IWebBrowser2>>navigate:
>
> I tried to see how the URLPresenter set up this control to an external
> address. Is there sample code somewhere in the hierarchy from which
> I could cook book a solution.
Just set the value of the model associated with the URLPresenter to the URL
you want to visit. For example:
ie := URLPresenter show.
ie value: 'www.google.com'.
As for accessing the documnt, well the IE control is asynchronous, so if you
attempt to access the document immediately after the navigation request you
will more than likely get back nil. You need to pause until the document is
available; there are lots of ways to do this, including listening for the
events, probably the #DocumentComplete: event, though you'd need to check
MSDN for further details about the events sent by the browser control. BTW:
The selectors of the events triggered can be found by evaluating an
expression such as:
ie view sink eventNames
Or for a little more detail:
ie view sink sourceTypeInfo printIDL
You may find it simpler, however, to just poll the control in a loop until
the download is complete. For example:
[ie view controlDispatch readState = READYSTATE_COMPLETE] whileFalse:
[SessionManager inputState pumpMessages].
You may need to disable parts of your applications user interface (if it has
one) while the document is downloading, as otherwise it will be "live". Note
that it is even necessary to wait for the initial about:blank document to
load, which is useful if one wants to push generated HTML directly into the
control:
When the document has loaded you can access the document object model:
ie view controlDispatch document
The result will be an IDispatch onto the HTMLDocument object. Any access to
the objects properties, or invocation of its methods, will have to go
through #doesNotUnderstand: or explicit calls via #invoke:etc #getProperty:
#setProperty:etc. If you are contemplating anything more than simple
manipulation of the DOM, you may want to generate interfaces for at least
part of the object model using the AX component wizard. Be warned, however,
that the full HTML DOM is huge. This is not particularly an issue in a
deployed app., because the stripper will be able to remove most of the
unused interfaces, but you may find it significantly bloats your development
image.
> Is there a better way to parse and html document?
If it is well formed, then the XML parser will do it, but that is a big
'If'. Maybe there is something else available from (for example)
dolphinharbour.org?
Regards
Blair
Locked
