Page Object pattern and how to realize some best practices - automated-tests

It seems to be a best practice to return from one page object the next page object when navigating from one to another
Martin Fowler: PageObject
"if you navigate to another page, the initial page object should return another page object for the new page"
Getting Started with Page Object Pattern
"When you navigate you should return a page object for the next page"
I have the situation where after the first page object a simple "OK"-dialog informs the user about a status. And after clicking "OK" the next page object comes. Because I have handle such simple "OK" dialog often in the app, I did it till now in a generic way (find inside a "OK"-dialog page object the dialog and press OK). With this approach I cannot return the next page object.
To solve this I see these opportunities where I like to get some feedback on which of them might be the "best" one and what additional ideas you have.
implement for each of these "OK" dialogs a dedicated one which returns the next specific page object
create a generic Simple dialog where the type parameter is the next page object and inject the logic to find and return the next page object somehow (possible? how to do?)
don't follow the best practices and use the current approach where the simple "OK" dialog page object doesn't return the next page object and the "test logic" is resolving and calling the next page object themselves
let the first page object handle the call of the simple "OK"-dialog page object and then return the next page object from here; this would be fine for a simple "OK"-dialog, but I assume an astonishing coding experience in case it is a more complex "Yes/No"-dialog.
Thanks for your helping thoughts and feedback.
ps: might be important: the programming language is C#

Selenium guides to return PageObject to "model the user's journey through our application"
Depending on the type of journey the following options work:
option two: If "Simple OK Dialog" is A PART of the user journey (i.e. you are going to use this page object in tests) you can use a generic class approach
SimpleDialogPage Declaration:
public class SimpleDialogPage<T>
{
private IWebDriver driver;
public SimpleDialogPage(IWebDriver driver){
this.driver = driver;
}
public T ClickOk(){
return new T(driver);
}
}
SimpleDialogPage Invocation from the "first page"
public SimpleDialogPage<TheNextPage> ClickNext(){
return new SimpleDialogPage<TheNextPage>(driver)
}
Option four: If "Simple OK Dialog" is NOT a part of the user journey (i.e. you are not going to use this page object in tests) just hide it behind the action
TheNextPage ClickNextAndConfirmAlert(){
SimpleDialogPage alert = _buttonNext.Click();
alert.ClickOk();
return new TheNextPage(driver);
}

Thanks #Anatoly!
I did it in a close way as you proposed.
The current draft needs some refactorings, but the overall concept seems to be fine:
using FlaUI.Core.AutomationElements;
using System;
namespace Bvms.UiAutomation.SDK.Commons
{
/// <summary>Enhanced simple dialog allowing to specify two delegates to define the next page objects
/// for the confirmation and cancellation case.</summary>
/// <typeparam name="T">Next page object in case of confirmation (OK).</typeparam>
/// <typeparam name="TU">Next page object in case of canceling (Cancel).</typeparam>
/// <seealso cref="Dialog" />
public class GenericDialog<T, TU> : Dialog where T : PageObject where TU : PageObject
{
protected readonly Func<T> _confirmationCommand;
protected readonly Func<TU> _cancelCommand;
public GenericDialog(Window parentWindow, string windowsTitle, TimeSpan timeout)
: this(parentWindow, windowsTitle, timeout, null, null)
{
}
public GenericDialog(Window parentWindow, string windowsTitle, TimeSpan timeout, Func<T> confirmationCommand, Func<TU> cancelCommand)
: base(parentWindow, windowsTitle, timeout)
{
_confirmationCommand = confirmationCommand;
_cancelCommand = cancelCommand;
}
public new virtual (T, TU) EndDialog(DialogCompletion dialogCompletion)
{
switch (dialogCompletion)
{
case DialogCompletion.Confirmation:
return (ConfirmDialog(), default);
case DialogCompletion.Cancellation:
return (default, CancelDialog());
default:
throw new ArgumentOutOfRangeException(nameof(dialogCompletion), dialogCompletion, null);
}
}
public virtual T ConfirmDialog()
{
if (null == _confirmationCommand)
{
throw new NotImplementedException($"Use ConfirmDialog(Func<T> confirmationCommand) or create with ctor: GenericDialog(Window parentWindow, string windowsTitle, TimeSpan timeout, Func<T> confirmationCommand, Func<TU> cancelCommand)");
}
return ConfirmDialog(_confirmationCommand);
}
public virtual TU CancelDialog()
{
if (null == _cancelCommand)
{
throw new NotImplementedException($"Use CancelDialog(Func<TU> cancelCommand) or create with ctor: GenericDialog(Window parentWindow, string windowsTitle, TimeSpan timeout, Func<T> confirmationCommand, Func<TU> cancelCommand)");
}
return CancelDialog(_cancelCommand);
}
public virtual T ConfirmDialog(Func<T> confirmationCommand)
{
base.EndDialog(DialogCompletion.Confirmation);
return confirmationCommand();
}
public virtual TU CancelDialog(Func<TU> cancelCommand)
{
base.EndDialog(DialogCompletion.Cancellation);
return cancelCommand();
}
}
}

Related

Ajax PageMethod accessing page-level private static property

I have a page that uses Ajax Page Methods. When the page first loads, the user is prompted to select a year. This is the only time that a PostBack occurs. The year is stored in a private static page-level integer property named SelectedYear. There are several page methods that pass data from the client to the server, but the year is always stored on the server, so that it won't have to be to be passed in again. The problem is, in a few cases, within the server WebMethod, the SelectedYear property seems to be reverting to 0. I can test for 0 and throw the error back to the client, but it would help if I could explain why it happened. At this point, I don't know. Any ideas? I'm a bit new to this style of programming. Here's a (very simplified) example of the code. The user MUST have selected a year in order to ever have reached the save function.
Here is my C# server code:
public partial class Default : System.Web.UI.Page
{
private static int SelectedYear;
protected void YearSelected(object sender, EventArgs e)
{
if (sender.Equals(btnCurrentYear))
SelectedYear = 2013;
else
SelectedYear = 2014;
}
[WebMethod]
public static bool Save(string FirstName, string LastName)
{
try
{
if (HttpContext.Current.User.Identity.IsAuthenticated)
//Right here, SelectedYear is sometimes 0.
SaveApplication(FirstName, LastName, SelectedYear);
else
throw new Exception("User is not logged in.");
}
catch (Exception ex)
{
throw;
}
}
}
Here is my JavaScript client code:
function Save(FirstName, LastName) {
PageMethods.Save(firstName, LastName, SaveSucceeded, SaveFailed);
}
function SaveSucceeded(result) {
//Notify user that save succeeded.
}
function SaveFailed(error) {
//Notify user that save failed.
}
Your problem is this:
private static int SelectedYear;
You'll want to remove the static. Static means it's global and will be shared for all users/requests... so when you set it to 2013 for one user, and another user hits that page who hasn't yet selected a year, it will be set to 0... for both of them. Yikes!
Trace through your postbacks to see what is happening to that variable during your AJAX methods.
You should consider storing the value in a session variable or maybe in a hidden field on the page.
More reading on a similar post: ASP.NET C# Static Variables are global?

Send object to VIewModel with mvvm-light

I'm pretty new to MVVM light world, and after searches I can't find what I want to do.
My WP7 application contains a pivot, each pivot item content is View1 and viewmodel is VM1.
When loading my application, I'd like to create every pivot item with the same view and view model but with different parameter.
example :
PivotItem 1 -> send param "car" to the view model
PivotItem 2 -> send param "truck" to the view model, etc.
Google told me to use messaging but if I send 2 messages from my MainViewModel, both PivotItem1 and PivotItem2 ViewModel will receive these messages.
Am I wrong with this approach ?
Is there another solution to succeed ?
Thank you in advance for your answer.
PS : be indulgent, english is not my native language, don't hesitate to ask for further information.
Regards,
Aymeric Lagier
To seperate the messages use the second constructor signature whereby you can pass a token. This token can be anything but I generally use an enum to store all my message types within the system.
Create a static class in a common library and reference this in all projects where you need to send or receive messages.
The following code hopefully shows this approach, notice I am sending a string as a value within the message but this can be anything, even a complex object such as one of your business objects.
namespace MyProject.Common
{
public static class AppMessages
{
enum MessageTypes
{
ViewmodelA,
ViewmodelB
}
public static class ViewModelAUpdate
{
public static void Send(string value)
{
Messenger.Default.Send(value, MessageTypes.ViewmodelA);
}
public static void Register(object recipient, Action<string> action)
{
Messenger.Default.Register(recipient, MessageTypes.ViewmodelA, action);
}
}
public static class ViewModelBUpdate
{
public static void Send(string value)
{
Messenger.Default.Send(value, MessageTypes.ViewmodelB);
}
public static void Register(object recipient, Action<string> action)
{
Messenger.Default.Register(recipient, MessageTypes.ViewmodelB, action);
}
}
}
}
How about using a method to set the message you want to receive. (this could be done as a parameter in the constructor or a property as well)
public void RegisterForAppMessage(AppMessages.MessageTypes messageType)
{
switch (messageType)
{
case AppMessages.MessageTypes.PivotViewItem1Message:
AppMessages.PivotViewItem1Message.Register(this,DoSomethingWhenIRecievePivotViewItem1Messages)
break;
case AppMessages.MessageTypes.PivotViewItem2Message:
AppMessages.PivotViewItem2Message.Register(this,DoSomethingWhenIRecievePivotViewItem2Messages)
break;
}
}
private void DoSomethingWhenIRecievePivotViewItem2Messages(string obj)
{
// TODO: Implement this method
throw new NotImplementedException();
}
private void DoSomethingWhenIRecievePivotViewItem1Messages(string obj)
{
// TODO: Implement this method
throw new NotImplementedException();
}
Messaging sounds a bit heavy for this purpose. Could you simply inject a parameter into your ViewModel. If you already have MVVMLight you also have support for SimpleIOC. Maybe let the view locate its ViewModel when the view is resolved and there decide which parameter to use on the view model?
You can see an example of it here

System.Web.Routing with WebForms - picking up variables in the target page

I have the pattern User/{domain}/{username} set up via Routing. Everything works except for one thing. I can't figure out how to get the domain and username variables passed to my redirected page. Below is my GetHttpHandler method from my IRouteHandler implementation.
public IHttpHandler GetHttpHandler(RequestContext requestContext)
{
string basePath;
basePath = "~/UserPage.aspx";
string domain = requestContext.RouteData.GetRequiredString("domain");
string username = requestContext.RouteData.GetRequiredString("username");
string virtualPath =
string.Format(basePath + "?domain={0}&username={1}", domain, username);
return (Page)BuildManager.CreateInstanceFromVirtualPath(virtualPath, typeof(Page));
}
I get the error from the last line of code:
UserPage.aspx?domain=SOMEDOMAIN&username=SOMEUSER is not a valid virtual path.
So how are you supposed to pass variables to the target page? what am I missing?
I think I solved this one myself.
Found this loop
foreach (KeyValuePair<string, object> token in requestContext.RouteData.Values)
{
requestContext.HttpContext.Items.Add(token.Key, token.Value);
}
from http://www.codethinked.com/post/2008/08/20/Exploring-SystemWebRouting.aspx
Its like the 4th code sample down.
UPDATE:
Not sure if this will work... requestContext.HttpContext seems to be "readonly". Back to the drawing board.
UPDATE 2:
Looks like this will work if you add in a reference to System.Web.Abstractions
Started mucking around with things and saw the IHttpHandler interface provides the RequestContext to the GetHttpHandler method.
So, I modified my base page class (I always put a layer between System.Web.UI.Page and my own pages, calling it BasePage or similar just for the purpose). So I added a public property on PVBasePage to receive a RequestContext object.
public RequestContext RequestContext { get; set; }
Then, my Routing class code is as follows:
IHttpHandler IRouteHandler.GetHttpHandler(RequestContext requestContext)
{
// create the page object as my own page...
var page = BuildManager.CreateInstanceFromVirtualPath(VirtualPath
, typeof(PVBasePage)) as PVBasePage;
// pass in the request context
page.RequestContext = requestContext;
// return this page in the form of a IHttpHandler
return page as IHttpHandler;
}
So instead of, as in the sample code, creating the instance directly as the IHttpHandler, I create it as my own page. Set the request context property, and then return the page to the caller AS a IHttpHandler.
Tested and it works. WOO HOO!
Then in the instance page, you can hit the RequestContext.GetValues collection to read out your passed in parameters.
HTH
#B.Tyndall
I just got this working with a solution similar to yours.
found at: http://msmvps.com/blogs/luisabreu/archive/2008/03/12/using-the-routing-mvc-api-with-classic-asp-net.aspx
foreach (var aux in requestContext.RouteData.Values)
{
HttpContext.Current.Items[aux.Key] = aux.Value;
}
So in effect you're no longer using the Request.QueryString but instead Context.Items collection
HttpContext.Current.Items["RouteName"]
or
Context.Items["RouteName"]
It appears as though other are also taking the route (no pun intended) of putting the parameters in the context Items collection.
http://bbits.co.uk/blog/archive/2008/05/19/using-asp.net-routing-independent-of-mvc---passing-parameters-to.aspx
I combined a couple of these approaches for pages that have a specific parameter, I created a UserNameRouteHandler for pages that accept that type of parameter. In my PageBase class I checked the context items for that parameter and then set a property so that my pages that inherit from PageBase can use it.
public class UserNameRouteHandler : IRouteHandler
{
#region Implementation of IRouteHandler
public IHttpHandler GetHttpHandler(RequestContext requestContext)
{
string pageName = requestContext.RouteData.GetRequiredString("PageName");
string employeeUserName = requestContext.RouteData.GetRequiredString("UserName");
if(!string.IsNullOrEmpty(employeeUserName))
{
requestContext.HttpContext.Items["UserName"] = employeeUserName;
}
pageName = pageName.ToLower() == "home" ? "default" : pageName;
string virtualPath = string.Format("~/{0}.aspx", pageName);
return (Page)BuildManager.CreateInstanceFromVirtualPath(virtualPath, typeof(Page));
}
#endregion
}
And in my OnLoad of PageBase I set the property to pages that need it can have it...definitely looking for a more elegant solution though.
protected override void OnLoad(EventArgs e)
{
if (!IsPostBack)
{
if (Context.Items["UserName"] != null)
{
EmployeeUserName = Context.Items["UserName"].ToString();
}
}
base.OnLoad(e);
}

ASP.NET - Avoid hardcoding paths

I'm looking for a best practice solution that aims to reduce the amount of URLs that are hard-coded in an ASP.NET application.
For example, when viewing a product details screen, performing an edit on these details, and then submitting the changes, the user is redirected back to the product listing screen. Instead of coding the following:
Response.Redirect("~/products/list.aspx?category=books");
I would like to have a solution in place that allows me to do something like this:
Pages.GotoProductList("books");
where Pages is a member of the common base class.
I'm just spit-balling here, and would love to hear any other way in which anyone has managed their application redirects.
EDIT
I ended up creating the following solution: I already had a common base class, to which I added a Pages enum (thanks Mark), with each item having a System.ComponentModel.DescriptionAttribute attribute containing the page's URL:
public enum Pages
{
[Description("~/secure/default.aspx")]
Landing,
[Description("~/secure/modelling/default.aspx")]
ModellingHome,
[Description("~/secure/reports/default.aspx")]
ReportsHome,
[Description("~/error.aspx")]
Error
}
Then I created a few overloaded methods to handle different scenarios. I used reflection to get the URL of the page through it's Description attribute, and I pass query-string parameters as an anonymous type (also using reflection to add each property as a query-string parameter):
private string GetEnumDescription(Enum value)
{
Type type = value.GetType();
string name = Enum.GetName(type, value);
if (name != null)
{
FieldInfo field = type.GetField(name);
if (field != null)
{
DescriptionAttribute attr = Attribute.GetCustomAttribute(field, typeof(DescriptionAttribute)) as DescriptionAttribute;
if (attr != null)
return attr.Description;
}
}
return null;
}
protected string GetPageUrl(Enums.Pages target, object variables)
{
var sb = new StringBuilder();
sb.Append(UrlHelper.ResolveUrl(Helper.GetEnumDescription(target)));
if (variables != null)
{
sb.Append("?");
var properties = (variables.GetType()).GetProperties();
foreach (var property in properties)
sb.Append(string.Format("{0}={1}&", property.Name, property.GetValue(variables, null)));
}
return sb.ToString();
}
protected void GotoPage(Enums.Pages target, object variables, bool useTransfer)
{
if(useTransfer)
HttpContext.Current.Server.Transfer(GetPageUrl(target, variables));
else
HttpContext.Current.Response.Redirect(GetPageUrl(target, variables));
}
A typical call would then look like so:
GotoPage(Enums.Pages.Landing, new {id = 12, category = "books"});
Comments?
I'd suggest that you derive your own class ("MyPageClass") from the Page class and include this method there:
public class MyPageClass : Page
{
private const string productListPagePath = "~/products/list.aspx?category=";
protected void GotoProductList(string category)
{
Response.Redirect(productListPagePath + category);
}
}
Then, in your codebehind, make sure that your page derives from this class:
public partial class Default : MyPageClass
{
...
}
within that, you can redirect just by using:
GotoProductList("Books");
Now, this is a bit limited as is since you'll undoubtedly have a variety of other pages like the ProductList page. You could give each one of them its own method in your page class but this is kind of grody and not smoothly extensible.
I solve a problem kind of like this by keeping a db table with a page name/file name mapping in it (I'm calling external, dynamically added HTML files, not ASPX files so my needs are a bit different but I think the principles apply). Your call would then use either a string or, better yet, an enum to redirect:
protected void GoToPage(PageTypeEnum pgType, string category)
{
//Get the enum-to-page mapping from a table or a dictionary object stored in the Application space on startup
Response.Redirect(GetPageString(pgType) + category); // *something* like this
}
From your page your call would be: GoToPage(enumProductList, "Books");
The nice thing is that the call is to a function defined in an ancestor class (no need to pass around or create manager objects) and the path is pretty obvious (intellisense will limit your ranges if you use an enum).
Good luck!
You have a wealth of options availible, and they all start with creating a mapping dictionary, whereas you can reference a keyword to a hard URL. Whether you chose to store it in a configuration file or database lookup table, your options are endless.
You have a huge number of options available here. Database table or XML file are probably the most commonly used examples.
// Please note i have not included any error handling code.
public class RoutingHelper
{
private NameValueCollecton routes;
private void LoadRoutes()
{
//Get your routes from db or config file
routes = /* what ever your source is*/
}
public void RedirectToSection(string section)
{
if(routes == null) LoadRoutes();
Response.Redirect(routes[section]);
}
}
This is just sample code, and it can be implemented any way you wish. The main question you need to think about is where you want to store the mappings. A simple xml file could do it:
`<mappings>
<map name="Books" value="/products.aspx/section=books"/>
...
</mappings>`
and then just load that into your routes collection.
public class BasePage : Page
{
public virtual string GetVirtualUrl()
{
throw new NotImplementedException();
}
public void PageRedirect<T>() where T : BasePage, new()
{
T page = new T();
Response.Redirect(page.GetVirtualUrl());
}
}
public partial class SomePage1 : BasePage
{
protected void Page_Load()
{
// Redirect to SomePage2.aspx
PageRedirect<SomePage2>();
}
}
public partial class SomePage2 : BasePage
{
public override string GetVirtualUrl()
{
return "~/Folder/SomePage2.aspx";
}
}

Redirecting users from edit page back to calling page

I am working on a project management web application. The user has a variety of ways to display a list of tasks. When viewing a list page, they click on task and are redirected to the task edit page.
Since they are coming from a variety of ways, I am just curious as to the best way to redirect the user back to the calling page. I have some ideas, but would like to get other developers input.
Would you store the calling url in session? as a cookie? I like the concept of using an object handle the redirection.
I would store the referring URL using the ViewState. Storing this outside the scope of the page (i.e. in the Session state or cookie) may cause problems if more than one browser window is open.
The example below validates that the page was called internally (i.e. not requested directly) and bounces back to the referring page after the user submits their response.
public partial class _Default : System.Web.UI.Page
{
protected void Page_Load(object sender, EventArgs e)
{
if (Request.UrlReferrer == null)
{
//Handle the case where the page is requested directly
throw new Exception("This page has been called without a referring page");
}
if (!IsPostBack)
{
ReturnUrl = Request.UrlReferrer.PathAndQuery;
}
}
public string ReturnUrl
{
get { return ViewState["returnUrl"].ToString(); }
set { ViewState["returnUrl"] = value; }
}
protected void btn_Click(object sender, EventArgs e)
{
//Do what you need to do to save the page
//...
//Go back to calling page
Response.Redirect(ReturnUrl, true);
}
}
This message my be tagged asp.net but I think it is a platform independent issue that pains all new web developers as they seek a 'clean' way to do this.
I think the two options in achieving this are:
A param in the url
A url stored in the session
I don't like the url method, it is a bit messy, and you have to remember to include the param in every relevent URL.
I'd just use an object with static methods for this. The object would wrap around the session item you use to store redirect URLS.
The methods would probably be as follows (all public static):
setRedirectUrl(string URL)
doRedirect(string defaultURL)
setRedirectUrl would be called in any action that produces links / forms which need to redirect to a given url. So say you had a projects view action that generates a list of projects, each with tasks that can be performed on them (e.g. delete, edit) you would call RedirectClass.setRedirectUrl("/project/view-all") in the code for this action.
Then lets say the user clicks delete, they need to be redirected to the view page after a delete action, so in the delete action you would call RedirectClass.setRedirectUrl("/project/view-all"). This method would look to see if the redirect variable was set in the session. If so redirect to that URL. If not, redirect to the default url (the string passed to the setRedirectUrl method).
I agree with "rmbarnes.myopenid.com" regarding this issue as being platform independent.
I would store the calling page URL in the QueryString or in a hidden field (for example in ViewState for ASP.NET). If you will store it outside of the page scope (such as Session, global variable - Application State and so on) then it will not be just overkill as Tom said but it will bring you trouble.
What kind of trouble? Trouble if the user has more than one tab (window) of that browser open. The tabs (or windows) of the same browser will probably share the same session and the redirection will not be the one expected and all the user will feel is that it is a bug.
My 2 eurocents..
I personally would store the required redirection info in an object and handle globally. I would avoid using a QueryString param or the like since they could try bouncing themselves back to a page they are not supposed to (possible security issue?). You could then create a static method to handle the redirection object, which could read the information and act accordingly. This encapsulates your redirection process within one page.
Using an object also means you can later extend it if required (such as adding return messages and other info).
For example (this is a 2 minute rough guideline BTW!):
public partial class _Default : System.Web.UI.Page
{
void Redirect(string url, string messsage)
{
RedirectionParams paras = new RedirectionParams(url, messsage);
RedirectionHandler(paras); // pass to some global method (or this could BE the global method)
}
protected void Button1_Click(object sender, EventArgs e)
{
Redirect("mypage.aspx", "you have been redirected");
}
}
public class RedirectionParams
{
private string _url;
public string URL
{
get { return _url; }
set { _url = value; }
}
private string _message;
public string Message
{
get { return _message; }
set { _message = value; }
}
public RedirectionParams(string url, string message)
{
this.URL = url;
this.Message = message;
}
}

Resources