Gui 反应迟钝

Question

First of all thanks for such a great community. I've learned a lot from your questions and answers here. This is my first question on SO, so please be gentle :)

Ok, but first things first:

-1st code version:

private async void buttonWebScrap_Click(object sender, EventArgs e)
{
    ClickLink("/ptk/sun/core/cookie/CookiesHandler.accept");

    await Task.Delay(750);

    if (_backgroundTaskRunning || !ClickLink("msisdn-change")) return;

    _backgroundTaskRunning = true;
    await LongTaskAsync();
}

private async Task LongTaskAsync()
{
    const string previous = "msisdn-pool-prev";
    const string next = "msisdn-pool-next";

    var tempNumbers = new List<object>();
    
    while (true)
    {
        await Task.Delay(750);

        var document = webBrowser.DocumentText;
        var htmlDoc = new HtmlAgilityPack.HtmlDocument();
        htmlDoc.LoadHtml(document);

        var numbers = htmlDoc.DocumentNode.SelectNodes("//a[starts-with(@id, 'msisdn')]");

        tempNumbers.AddRange(from number in numbers
                             where number.Id != previous && number.Id != next
                             select number.InnerText.RemoveEnters().RemoveSpaces().ReplaceSpecificChars());
        tempNumbers.Add("-------------------------");

        if (tempNumbers.Count >= 24)
        {
            listBoxNumbers.Items.AddRange(tempNumbers.ToArray());
            tempNumbers.Clear();
        }
        
        if (ClickLink(next) == false)
        {
            break;
        }
    }
} 

private bool ClickLink(string linkId)
{
    if (webBrowser.Document != null)
    {
        var elementById = webBrowser.Document.GetElementById(linkId);

        if (elementById != null)
        {
            elementById.InvokeMember("click");
        }
        else
        {
            return false;
        }

        if (webBrowser.Document.Window != null)
        {
            webBrowser.Document.Window.ScrollTo(0, 480);
        }
    }
    else
    {
        return false;
    }

    return true;
}

-2nd code version:

private void MainForm_Load(object sender, EventArgs e)
    {
    _webBrowserDocuments = new ConcurrentQueue<string>();
    _uiScheduler = TaskScheduler.FromCurrentSynchronizationContext();

    _progress = new Progress<string>();
    _progress.ProgressChanged += (o, s) => _objects.Add(s);

    _objects = new BindingList<string>();
    listBoxNumbers.DataSource = _objects;
}

private void MainForm_FormClosing(object sender, FormClosingEventArgs e)
{
    WebBrowserEmulation.Delete();
}

private async void buttonWebScrap_Click(object sender, EventArgs e)
{
    await WebBrowserClickLinkAsync("/ptk/sun/core/cookie/CookiesHandler.accept");

    if (_backgroundTaskRunning || !(await WebBrowserClickLinkAsync("msisdn-change"))) return;

    await Task.Delay(5000);
    var cts = new CancellationTokenSource();

    await WebBrowserDocumentDownloadAsync(cts);
    await DocumentParseAsync(_progress, cts);

    _backgroundTaskRunning = true;
}

private async Task DocumentParseAsync(IProgress<string> progress, CancellationTokenSource cts)
{
    await Task.Factory.StartNew(() =>
        {
            while (true)
            {
                string tempDocument;
                if (_webBrowserDocuments.TryDequeue(out tempDocument))
                {
                    var htmlDoc = new HtmlAgilityPack.HtmlDocument();
                    htmlDoc.LoadHtml(tempDocument);

                    var numbers = htmlDoc.DocumentNode.SelectNodes("//a[starts-with(@id, 'msisdn')]");

                    foreach (var number in numbers.Where(number => number.Id != Previous && number.Id != Next).
                                                   Select(x => x.InnerText.RemoveEnters().RemoveSpaces().ReplaceSpecificChars()))
                    {
                        progress.Report(number);
                    }

                    progress.Report("-------------------------");
                }

                if (cts.IsCancellationRequested)
                {
                    break;
                }
            }
        }, cts.Token);
}

private async Task WebBrowserDocumentDownloadAsync(CancellationTokenSource cts)
{
    await Task.Factory.StartNew(async () =>
        {
            while (true)
            {
                await Task.Delay(1000);

                _webBrowserDocuments.Enqueue(webBrowser.DocumentText);

                if (await WebBrowserClickLinkAsync(Next)) continue;
                cts.Cancel();
                break;
            }
        }, new CancellationToken(), TaskCreationOptions.None, _uiScheduler);
}

private async Task<bool> WebBrowserClickLinkAsync(string linkId)
{
    return await Task.Factory.StartNew(() =>
        {
            if (webBrowser.Document != null)
            {
                var elementById = webBrowser.Document.GetElementById(linkId);

                if (elementById != null)
                {
                    elementById.InvokeMember("click");
                }
                else
                {
                    return false;
                }

                if (webBrowser.Document.Window != null)
                {
                    webBrowser.Document.Window.ScrollTo(0, 480);
                }
            }
            else
            {
                return false;
            }

            return true;
        }, new CancellationToken(), TaskCreationOptions.None, _uiScheduler);
}

On the beggining everything is working just fine, but after webscraping about 500 of numbers, the "GUI" is a little bit sluggish. I don't know if it is with my "bad" understanding of async/await pattern, or something else. I thought that the second version will be better for this task - but it's still sluggish :/. Can someone help me with this?

Why I'm using webbrowser control instead of webclient? I know it will be a lot easier, but the site from which I'm webscraping is made with (as I see it) Java (jsessionId) + ajax, and there aren't "proper" links.

If you need some more details, just write ;)

Thanks in advance.

EDIT:

• second version is using Methods which are returning Task (or Task) to simplify the await with current SynchronizationContext from MainForm (only two of those)

• first version was the first approach of using await/async (as you can see the LongTaskAsync() method is async with await Task.Delay())

• this is a finished code (without some thinks like getting SynchronizationContext, setting ListBox.DataSource to BindList, etc.) with only 3 controls on winform - webbrowser, listbox and button ;)

Answer 1

The slowdown is likely due to adding your values to your user interface.

During your looping, you're adding the items to a list box:

    if (tempNumbers.Count >= 24)
    {
        listBoxNumbers.Items.AddRange(tempNumbers.ToArray());
        tempNumbers.Clear();
    }

As you get more and more results, the list box display will actually become a bottleneck, and cause things to slow down. Since the list box must always be updated on the UI thread, this will cause your UI to get less responsive over time.

Your second option will likely be even worse, as you add items to a BindingList<T> one item at a time in the second option, and each addition will cause a refresh of the UI.

This could be mitigated by using a ListView with VirtualMode set to true, as this prevents the addition of new items from forcing a refresh on the screen.