How can I take more control in ASP.NET?
I'm trying to build a very, very simple "micro-webapp" which I suspect will be of interest to a few Stack Overflow'rs if I ever get it done. I'm hosting it on my C# in Depth site, which is vanilla ASP.NET 3.5 (i.e. not MVC).
The flow is very simple:
- If a user enters the app with a URL which doesn't specify all the parameters (or if any of them are invalid) I want to just display the user input controls. (There are only two.)
- If a user enters the app with a URL which does have all the required parameters, I want to display the results and the input controls (so they can change the parameters)
Here are my self-imposed requirements (mixture of design and implementation):
- I want the submission to use GET rather than POST, mostly so users can bookmark the page easily.
- I don't want the URL to end up looking silly after submission, with extraneous bits and pieces on it. Just the main URL and the real parameters please.
- Ideally I'd like to avoid requiring JavaScript at all. There's no good reason for it in this app.
- I want to be able to access the controls during render time and set values etc. In particular, I want to be able to set the default values of the controls to the parameter values passed in, if ASP.NET can't do this automatically for me (within the other restrictions).
- I'm happy to do all the parameter validation myself, and I don't need much in the way of server side events. It's really simple to set everything on page load instead of attaching events to buttons etc.
Most of this is okay, but I haven't found any way of completely removing the viewstate and keeping the rest of the useful functionality. Using the post from this blog post I've managed to avoid getting any actual value for the viewstate - but it still ends up as a parameter on the URL, which looks really ugly.
If I make it a plain HTML form instead of an ASP.NET form (i.e. take out runat="server"
) then I don't get any magic viewstate - but then I can't access the controls programmatically.
I could do all of this by ignoring most of ASP.NET and building up an XML document with LINQ to XML, and implementing IHttpHandler
. That feels a bit low level though.
I realise that my problems could be solved by either relaxing my constraints (e.g. using POST and not caring about the surplus parameter) or by using ASP.NET MVC, but are my requirements really unreasonable?
Maybe ASP.NET just doesn't scale down to this sort of app? There's a very likely alternative though: I'm just being stupid, and there's a perfectly simple way of doing it that I just haven't found.
Any thoughts, anyone? (Cue comments of how the mighty are fallen, etc. That's fine - I hope I've never claimed to be an ASP.NET expert, as the truth is quite the opposite...)
Solution 1:
This solution will give you programmatic access to the controls in their entirety including all attributes on the controls. Also, only the text box values will appear in the URL upon submission so your GET request URL will be more "meaningful"
<%@ Page Language="C#" AutoEventWireup="true" CodeBehind="JonSkeetForm.aspx.cs" Inherits="JonSkeetForm" %>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" >
<head runat="server">
<title>Jon Skeet's Form Page</title>
</head>
<body>
<form action="JonSkeetForm.aspx" method="get">
<div>
<input type="text" ID="text1" runat="server" />
<input type="text" ID="text2" runat="server" />
<button type="submit">Submit</button>
<asp:Repeater ID="Repeater1" runat="server">
<ItemTemplate>
<div>Some text</div>
</ItemTemplate>
</asp:Repeater>
</div>
</form>
</body>
</html>
Then in your code-behind you can do everything you need on PageLoad
public partial class JonSkeetForm : System.Web.UI.Page
{
protected void Page_Load(object sender, EventArgs e)
{
text1.Value = Request.QueryString[text1.ClientID];
text2.Value = Request.QueryString[text2.ClientID];
}
}
If you don't want a form that has runat="server"
, then you should use HTML controls. It's easier to work with for your purposes. Just use regular HTML tags and put runat="server"
and give them an ID. Then you can access them programmatically and code without a ViewState
.
The only downside is that you won't have access to many of the "helpful" ASP.NET server controls like GridView
s. I included a Repeater
in my example because I'm assuming that you want to have the fields on the same page as the results and (to my knowledge) a Repeater
is the only DataBound control that will run without a runat="server"
attribute in the Form tag.
Solution 2:
You're definitely (IMHO) on the right track by not using runat="server" in your FORM tag. This just means you'll need to extract values from the Request.QueryString directly, though, as in this example:
In the .aspx page itself:
<%@ Page Language="C#" AutoEventWireup="true"
CodeFile="FormPage.aspx.cs" Inherits="FormPage" %>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<title>ASP.NET with GET requests and no viewstate</title>
</head>
<body>
<asp:Panel ID="ResultsPanel" runat="server">
<h1>Results:</h1>
<asp:Literal ID="ResultLiteral" runat="server" />
<hr />
</asp:Panel>
<h1>Parameters</h1>
<form action="FormPage.aspx" method="get">
<label for="parameter1TextBox">
Parameter 1:</label>
<input type="text" name="param1" id="param1TextBox" value='<asp:Literal id="Param1ValueLiteral" runat="server" />'/>
<label for="parameter1TextBox">
Parameter 2:</label>
<input type="text" name="param2" id="param2TextBox" value='<asp:Literal id="Param2ValueLiteral" runat="server" />'/>
<input type="submit" name="verb" value="Submit" />
</form>
</body>
</html>
and in the code-behind:
using System;
public partial class FormPage : System.Web.UI.Page {
private string param1;
private string param2;
protected void Page_Load(object sender, EventArgs e) {
param1 = Request.QueryString["param1"];
param2 = Request.QueryString["param2"];
string result = GetResult(param1, param2);
ResultsPanel.Visible = (!String.IsNullOrEmpty(result));
Param1ValueLiteral.Text = Server.HtmlEncode(param1);
Param2ValueLiteral.Text = Server.HtmlEncode(param2);
ResultLiteral.Text = Server.HtmlEncode(result);
}
// Do something with parameters and return some result.
private string GetResult(string param1, string param2) {
if (String.IsNullOrEmpty(param1) && String.IsNullOrEmpty(param2)) return(String.Empty);
return (String.Format("You supplied {0} and {1}", param1, param2));
}
}
The trick here is that we're using ASP.NET Literals inside the value="" attributes of the text inputs, so the text-boxes themselves don't have to runat="server". The results are then wrapped inside an ASP:Panel, and the Visible property set on page load depending whether you want to display any results or not.
Solution 3:
Okay Jon, the viewstate issue first:
I haven't checked if there's any kind of internal code change since 2.0 but here's how I handled getting rid of the viewstate a few years ago. Actually that hidden field is hardcoded inside HtmlForm so you should derive your new one and step into its rendering making the calls by yourself. Note that you can also leave __eventtarget and __eventtarget out if you stick to plain old input controls (which I guess you'd want to since it also helps not requiring JS on the client):
protected override void RenderChildren(System.Web.UI.HtmlTextWriter writer)
{
System.Web.UI.Page page = this.Page;
if (page != null)
{
onFormRender.Invoke(page, null);
writer.Write("<div><input type=\"hidden\" name=\"__eventtarget\" id=\"__eventtarget\" value=\"\" /><input type=\"hidden\" name=\"__eventargument\" id=\"__eventargument\" value=\"\" /></div>");
}
ICollection controls = (this.Controls as ICollection);
renderChildrenInternal.Invoke(this, new object[] {writer, controls});
if (page != null)
onFormPostRender.Invoke(page, null);
}
So you get those 3 static MethodInfo's and call them out skipping that viewstate part out ;)
static MethodInfo onFormRender;
static MethodInfo renderChildrenInternal;
static MethodInfo onFormPostRender;
and here's your form's type constructor:
static Form()
{
Type aspNetPageType = typeof(System.Web.UI.Page);
onFormRender = aspNetPageType.GetMethod("OnFormRender", BindingFlags.Instance | BindingFlags.NonPublic);
renderChildrenInternal = typeof(System.Web.UI.Control).GetMethod("RenderChildrenInternal", BindingFlags.Instance | BindingFlags.NonPublic);
onFormPostRender = aspNetPageType.GetMethod("OnFormPostRender", BindingFlags.Instance | BindingFlags.NonPublic);
}
If I'm getting your question right, you also want not to use POST as the action of your forms so here's how you'd do that:
protected override void RenderAttributes(System.Web.UI.HtmlTextWriter writer)
{
writer.WriteAttribute("method", "get");
base.Attributes.Remove("method");
// the rest of it...
}
I guess this is pretty much it. Let me know how it goes.
EDIT: I forgot the Page viewstate methods:
So your custom Form : HtmlForm gets its brand new abstract (or not) Page : System.Web.UI.Page :P
protected override sealed object SaveViewState()
{
return null;
}
protected override sealed void SavePageStateToPersistenceMedium(object state)
{
}
protected override sealed void LoadViewState(object savedState)
{
}
protected override sealed object LoadPageStateFromPersistenceMedium()
{
return null;
}
In this case I seal the methods 'cause you can't seal the Page (even if it isn't abstract Scott Guthrie will wrap it into yet another one :P) but you can seal your Form.