Detect Twebbrowser inputbox click in delphi XE2

Question

I am looking at creating autoform filler for Delphi, and obviously need a good method to capture which input boxes are the login boxes on each site so was wondering if I use a Twebbrowser component and load the page then click on the username and password boxes on the particular sites if I can then extract the form name and input box names that I clicked on.

In Short I need delphi to capture the name of the selected input box on a web page loaded into a twebbrowser component.

Any good methods to capture this information from the page loaded in a twebbrowser page would be appreciated!.

Answer 1

The following code shows how to find an INPUT element named 'input1':

var
  E : IHtmlElement;
  D : IHtmlDomNode;
  Doc2 : IHtmlDocument2;
  Doc3 : IHtmlDocument3;
  All : IHTMLElementCollection;
  i : Integer;

begin
  Doc3 := WebBrowser1.Document as IHtmlDocument3;
  D := Doc3.GetElementByID('input1') as IHtmlDomNode;
  if D <> Nil then begin
    ...

If you need to find more than one INPUT element or you wish to pattern-match the name of the INPUT element(s), you can do this by retrieving the document's IHtmlDocument2 interface and then iterating its all collection:

  Doc2 := WebBrowser1.Document as IHtmlDocument2;
  All := Doc2.all;
  for i := 0 to All.Length - 1 do begin
    E := All.Item(Null, i) as IHtmlElement;
    // Test E and do what you like with it
  end;

You could use a function like this to find the parent FORM element of an INPUT element

function GetParentFormElement(E : IHtmlElement) : IHtmlElement;
begin
  Result := Nil;
  while E <> Nil do begin
    if CompareText(E.tagName, 'form') = 0 then begin
      Result := E;
      exit;
    end;
    E := E.parentElement;
  end;
end;

and use it like this:

E := D as IHtmlElement;
E := GetParentFormElement(E);
Assert(E <> Nil);

not all forms have a name or id so how do I get the number or reference of a parent form if there are a number of forms in a page?

Equally, not all INPUT elements are contained in a FORM one. TBH, I don't know of a robust way of doing what you want which will survive a page's author making changes to it. Anyway, there must be some way of identifying a given INPUT element, otherwise the server wouldn't be able to extract the user's response, mustn't there? So it's just a question of figuring out what that might be for a particular page. It it is not in the attributes of the element, then maybe you could look for the text of a nearby text element - after all, there must be some kind of prompt to the user to tell them what to fill in where. But this is really a different issue than the substance of your original q, which I hope I have answered. If you need more help on this specific point, I suggest you ask in a new q. Make sure you include details (code) of what you've already tried, as qs lacking them tend not to be received well at SO.