我如何有效管理在主线程中运行的准“任务”

Question

I'm developing a small internal-use utility to monitor the uptime and responsiveness of several websites. We have very specialist requirements and criteria to test, hence developing a small application for this.

I have to use the WebBrowser Control to load the website as it has decencies on JavaScript and plug ins which are required. (The test has to be as real world as possible) As a result, my testing has to be run in the GUI thread, currently within the "DocumentCompleted" event.

Additionally, the website I'm testing is AJAX heavy, so I have to implement waiting periods within my logic to ensure that the DOM is loaded as I expect. For example, I check that the tags are present on the logon the page using a while loop.

My proof of concept works as expected and I'm very happy with the results. However, I'm struggling how to make this more 'task orientated'. For example, I need to implement a timeout period and the ability for the user to cancel what's going on. Eg, the logon page may have failed and so will never show the correct inputs.

At the moment, the only way I can see is to implement logic at every single step to check for a UserCancel variable or check the stopwatch.

I fear I've made a fundamentally bad choice in the architecture of my application, so I'm looking for advice on how to implement the above and improve the code.

---

private void TestSite(string url){

webBrowser.Naviate(url);
stopwatch.Start();

}

private void webBrowser_DocumentCompleted(object sender, WebBrowserDocumentCompletedEventArgs e) {

if(stopwatch.Elapsed.Seconds > timeOutValue || userCancel){
//Task timed out, or user has cancelled
return;
}

//Switch to the relevant application logic
//siteStatus is defined by where I expect the website to be at any particular time. 
switch(siteStatus)
{

     case siteStatus.LogonPage:
     if(DocumentTitle != ExpectedDocumentTitle){

     //Page is unexpected. Handle this and return

     }


     //Title is correct, wait until website has built the DOM
     while(stopwatch.Elapsed.Seconds < timeOutValue || !userCancel || webBrowser1.Document.GetElementsByTagName("input").Count() < 2){

     //Carry on
     Application.DoEvents();

     } 

     if(stopwatch.Elapsed.Seconds > timeOutValue){ return; }
     if(userCancel){ return; } 

     //Business logic goes here. Reports on time taken, ensures correct elements are on the page

     break;

     case siteStatus.Menu:

     //Essentially as above with different logic and tests

     break;

    //etc etc

}

Answer 1

The events based model is probably a better approach; rather than loops.

I would add some OnPropertyChanged event handlers to the div or container that is being updated by the ajax query as opposed to a loop.

private void DocumentCompletedHandler(object sender, EventArgs e)
{

    HtmlElement elem = webctrl.Document.GetElementById("ajaxDiv");

    if (elem != null)
    {
          elem.AttachEventHandler("onpropertychange", new EventHandler(ajaxEvent));
    }

}

with the handler

private void ajaxEvent(Object sender, EventArgs e)
{
      HtmlElement div = webctrl.Document.GetElementById("ajaxDiv");
      if (div == null) return;  // Not Found!?
      String contentLoaded = div.InnerHtml; // get the content loaded via ajax
}

With the same structure in mind, you can fire off an an interrupt event or something similar to abort the operation. Or you can use the stop command to stop all processing on the web control - Though I am not certain if this will stop ajax processing.

private void BtnAbort_Click(Object sender, EventArgs e)
{
   webctrl.Stop();
   // TODO : Remov any event handlers
}