We Keep Coding

Editor Pane in IWpfTextView is not Editable

I am trying to create a Visual Studio extension that will open a document window containing a custom control of mine that hosts an editor pane. I can get the document to load, the correct content type loads with my custom content extensions (classifier, quick info), a window with my custom control showing the editor pane loads, but the text cannot be modified. I can select and highlight text, but no keys work, no commands, and no mouse input other than text selection.
The following is what I have in my Package class:
[PackageRegistration(UseManagedResourcesOnly = true, AllowsBackgroundLoading = true)]
[ProvideEditorExtension(typeof(MyCustomEditorFactory), ".cust")]
[Guid(MyCustomExtensionsPackage.PackageGuidString)]
public sealed class MyCustomExtensionsPackage : AsyncPackage
{
public const string PackageGuidString = "ec2c4646-d0cc-42c6-b0a6-d0ff3e318cef";
#region Package Members
protected override async Task InitializeAsync(CancellationToken cancellationToken, IProgress<ServiceProgressData> progress)
{
await this.JoinableTaskFactory.SwitchToMainThreadAsync(cancellationToken);
RegisterEditorFactory(new MyCustomEditorFactory(this));
}
#endregion
}
I have verified that the RegisterEditorFactory method is called, and that my editor factory is called when files with the .cust extension are opened.
The following is my editor factory:
using Microsoft.VisualStudio;
using Microsoft.VisualStudio.ComponentModelHost;
using Microsoft.VisualStudio.Editor;
using Microsoft.VisualStudio.Shell;
using Microsoft.VisualStudio.Shell.Interop;
using Microsoft.VisualStudio.Text.Editor;
using Microsoft.VisualStudio.TextManager.Interop;
using Microsoft.VisualStudio.Utilities;
using System;
using System.ComponentModel.Composition;
using System.Runtime.InteropServices;
[ComVisible(true)]
[Guid(EditorFactoryGuidString)]
public sealed class MyCustomEditorFactory : IVsEditorFactory
{
public const string EditorFactoryGuidString = "38AD95BA-8891-46A2-A5EA-25F5F36EEAE0";
private MyCustomExtensionsPackage _package;
private Microsoft.VisualStudio.OLE.Interop.IServiceProvider _vsServiceProvider;
[Import]
public IContentTypeRegistryService ContentTypeRegistry { get; set; }
[Import]
public ITextEditorFactoryService TextEditorFactory { get; set; }
public MyCustomEditorFactory(MyCustomExtensionsPackage package)
{
_package = package;
}
public int SetSite(Microsoft.VisualStudio.OLE.Interop.IServiceProvider psp)
{
_vsServiceProvider = psp;
return (VSConstants.S_OK);
}
public int Close()
{
return (VSConstants.S_OK);
}
public int MapLogicalView(ref Guid rguidLogicalView, out string pbstrPhysicalView)
{
pbstrPhysicalView = null;
return (VSConstants.LOGVIEWID_Primary == rguidLogicalView ? VSConstants.S_OK : VSConstants.E_NOTIMPL);
}
public int CreateEditorInstance(uint grfCreateDoc, string pszMkDocument, string pszPhysicalView, IVsHierarchy pvHier, uint itemid, IntPtr punkDocDataExisting, out IntPtr ppunkDocView, out IntPtr ppunkDocData, out string pbstrEditorCaption, out Guid pguidCmdUI, out int pgrfCDW)
{
ThreadHelper.ThrowIfNotOnUIThread();
ppunkDocView = IntPtr.Zero;
ppunkDocData = IntPtr.Zero;
pbstrEditorCaption = string.Empty;
pguidCmdUI = VSConstants.GUID_TextEditorFactory;
pgrfCDW = 0;
int retVal = VSConstants.E_FAIL;
if ((grfCreateDoc & (VSConstants.CEF_OPENFILE | VSConstants.CEF_SILENT)) != 0)
{
IVsTextLines textBuffer = null;
if (punkDocDataExisting == IntPtr.Zero)
{
IComponentModel mef = _package.GetService<SComponentModel, IComponentModel>();
mef.DefaultCompositionService.SatisfyImportsOnce(this);
IVsEditorAdaptersFactoryService eafs = mef.GetService<IVsEditorAdaptersFactoryService>();
textBuffer = eafs.CreateVsTextBufferAdapter(_vsServiceProvider, ContentTypeRegistry.GetContentType("CUST")) as IVsTextLines;
string fileText = System.IO.File.ReadAllText(pszMkDocument);
textBuffer.InitializeContent(fileText, fileText.Length);
string[] roles = new string[]
{
PredefinedTextViewRoles.Analyzable,
PredefinedTextViewRoles.Editable,
PredefinedTextViewRoles.Interactive,
PredefinedTextViewRoles.Document,
PredefinedTextViewRoles.PrimaryDocument
};
IWpfTextView dataView = TextEditorFactory.CreateTextView(eafs.GetDataBuffer(textBuffer), TextEditorFactory.CreateTextViewRoleSet(roles));
dataView.Options.SetOptionValue(DefaultTextViewHostOptions.LineNumberMarginName, true);
dataView.Options.SetOptionValue(DefaultTextViewHostOptions.ShowCaretPositionOptionName, true);
dataView.Options.SetOptionValue(DefaultTextViewHostOptions.ChangeTrackingName, true);
dataView.Options.SetOptionValue(DefaultTextViewOptions.ViewProhibitUserInputName, false);
IWpfTextViewHost wpfHost = TextEditorFactory.CreateTextViewHost(dataView, false);
MyCustomEditor editor = new MyCustomEditor(wpfHost);
ppunkDocData = Marshal.GetIUnknownForObject(textBuffer);
ppunkDocView = Marshal.GetIUnknownForObject(editor);
retVal = VSConstants.S_OK;
}
else
{
//code for document already open
}
else
{
retVal = VSConstants.E_INVALIDARG;
}
}
return (retVal);
}
}
There's a lot to unpack there, but ultimately I'm just
Creating an IVsTextBuffer from an IVsEditorAdaptersFactoryService
Loading the contents of the text buffer
Creating an IWpfTextView with the IVsTextBuffer from a ITextEditorFactoryService
Creating an IWpfTextViewHost with the IWpfTextView from the same ITextEditorFactoryService
Creating my my custom window pane and passing in the IWpfTextViewHost
Returning the IVsTextBuffer as the document data and my custom window pane as the document view
My custom window pane code is simply:
using Microsoft.VisualStudio.Shell;
using Microsoft.VisualStudio.Text.Editor;
using System.Runtime.InteropServices;
[ComVisible(true)]
public sealed class MyCustomEditor : WindowPane
{
private IWpfTextViewHost _wpfHost;
public MyCustomEditor(IWpfTextViewHost wpfHost)
{
_wpfHost = wpfHost;
Content = new MyCustomEditorControl(wpfHost.HostControl);
}
}
I've omitted the code for the MyCustomEditorControl for the sake of brevity, but just assume it's a simple WPF UserControl with a ContentPresenter. The constructor just sets Content property of the ContentPresenter with the wpfHost.HostControl that's passed in.
All of this works insofar as displaying the contents of a file, but I just can't edit anything. It's like the input bindings aren't wired up or enabled, but I can't find what properties to set or enable on which objects, and the documentation is rather poor on this topic. I've gone through the Visual Studio extensions walkthroughs on MSDN, but it seems to be a mish-mash of pre-WPF and post-WPF APIs with no clear guidance on what the authoritative approach is.
Admittedly, I may be naive in assuming that the IWpfTextView would automatically handle user input, but in my defense it's created from an ITextEditorFactoryService and having "TextEditor" in the name carries certain implications.
I'm going to go through a rather exhaustive list of things I've tried and checked to hopefully help narrow down the possibilities or where my mistake is:
After the document window is open, I've checked the Running Document Table and there is an entry for it that is automatically created after CreateEditorInstance exits. It links to the data buffer that's returned, the correct file path, and has an edit lock but no read lock. I've inspected the RDT when C# files are opened in a normal editor window and those have read and write locks. I tried manually setting a read lock on the RDT as well, the counter increments, but that doesn't seem to make a difference.
I've inspected the return result from ITextBuffer.CheckEditAccess on the data buffer and it returns true so the text buffer is reporting it's editable.
The textview roles DO have an effect on the final rendered IWpfTextViewHost. If I add or omit roles, the appearance and functionality changes. For example, adding or removing "ZOOMABLE" adds or removes the dropdown box with the control zoom level. However, "EDITABLE" has no effect on editability.
The editor options DO have an effect. For example, adding or removing the line number margin options does show/hide the line numbers. However, ViewProhibitUserInputName has no effect on editability.
I don't create a code window with CreateVsCodeWindowAdapter because the resulting IWpfTextViewHost that can be retrieved from GetWpfTextViewHost is already parented to another control. If I disconnect it with (IWpfTextViewHost.HostControl.Parent as Border).Child = null, then the contents of IWpfTextViewHost.HostControl gets disposed of for some reason. The IWpfTextViewHost.HostControl loads and it displays a window with margins and scrollbars, but the content is blank and if I click into the left margin of the content pane then Visual Studio throws an ObjectDisposedException. Please note I'm not nulling out the host control itself or its contents, I'm only telling its existing parent it no longer has a child.
I've inspected the editor GUID property of the IVsWindowFrame that opens my document and it's set to the value returned in the pguidCmdUI parameter, which I believe is correct.
I'm not certain if the value being set for the pguidCmdUI parameter in the editor factory CreateEditorInstance is correct. I'm aware it's used for command routing, which might explain why I can't type or use the mouse, but I'm not sure what the correct value should be. I read somewhere it's supposed to be the GUID of the factory that created the editor, so I set it to the text editor factory GUID since that's what created the IWpfTextView. I tried setting it to my editor factory GUID at the top of the file, as well as typeof(IWpfTextView).GUID, Guid.Empty, and assorted other GUIDs, but there's no change.
If I pass my IWpfTextView into IVsEditorAdapterFactoryService.GetViewAdapter, I get back null.
If I create a IVsTextView from IVsEditorAdapterFactoryService.CreateVsTextViewAdapter, I can have both an IVsTextView and an IWpfTextView, but the two don't know about each other and I don't know see any way to map them. I don't even know if they need to be mapped, or if IVsTextView needs to exist or is just the legacy text view interface.
When I inspect the created IVsTextView, it's base type is a SimpleTextViewWindow that is an undocumented class that's internal to Microsoft's Visual Studio implementation, it has WpfTextView and WpfTextViewHost properties, but they're both null (or throw an exception that they can't be read) and are unsettable. Microsoft obviously has some internal voodoo magic to map IVsTextViews to WpfTextViews in their own code, but I don't know what it is.
I'm aware of other code samples that create an IVsInvisibleEditor and get the data buffer from that, but that seems kind of hacky. I mean, it technically works and no one seems to know of any alternative because the documentation in this area is so poor, so I'm not judging or criticizing, it just seems like that shouldn't be necessary. It's my (possibly incorrect) understanding that an invisible editor is a virtual editor that is created for documents that are opened in memory but not hosted in a window, but I'm hosting mine in a window. Additionally all of the working code samples I could find are all copy/pasting from the same Microsoft VSIX code example. In that example, the virtual editor is created in a tool window that's created under the assumption that a document is already open in an existing editor window. So that tool window is creating a virtual editor side-by-side to a physical one for a tool window. If I follow the same invisible editor approach, then I'm effectively creating a virtual editor in memory and then a second physical document editor window. So basically, I'm creating two editors for one document. That doesn't seem right. Again, it technically works, but it seems off.
If the expectation is that we have to implement IOleCommandTarget on our editor window and manually handle every single keystroke, shortcut, and mouse button and manipulate the underlying text buffer directly, then that is going to be the most depressing thing ever. That would mean they tout being able to use WPF, but then give us a document view without a WPF control that natively handles text editing even though they exist in the framework.
Sorry for the ridiculously long post, but I wanted to provide as much information as I possibly could about where I'm at and what I've tried.

Regarding CreateVsCodeWindowAdapter, see the "official" explanation how to make it work: https://developercommunity.visualstudio.com/t/projectionbuffertutorial-gives-error-in-dev16/498617

Dotliquid Template Render outputs filepath

I am currently trying to work with DotLiquid in C# and I've observed a behavior that I don't quite understand. Since I am not very familiar with C#, I cannot say for sure if my issue is with C# itself, or DotLiquid, so please bear with me. =)
I have a very basic index.liquid that I am trying to pass a Table-object to.
For the sake of getting started, I've overridden toString() to simply create a representation, for now, I'll later want to work with the actual object. When trying to work with the template, this is what I do:
public static void createHTML(DataTable table)
{
string templatePath = #"C:\Path\To\index.liquid";
var template = Template.Parse(templatePath);
template.Render(Hash.FromAnonymousObject(new
{
table = DataMapper.createTable(table).toString()
});
using (StreamWrite file = new StreamWriter(#"C:\Some\Path\test.html"))
{
file.write(template.Render());
}
}
Now, when I open this newly created test.html, all it contains is C:\Path\To\index.liquid, meaning I am somehow not loading my template correctly. Looking at Try to use DotLiquid with c#
I would have thought that I am loading the template correctly and using File.ReadAllText(templatePath)); showed me that templatePath is pointing to the correct file.
This shows me that I'm not understanding something very basic about Template.Parse() or Template.Render() where the source code doesn't provide me with the insight I'm missing, so hopefully, you can help me out.

Hopefully saving others from getting tripped up on this. The real reason why the output is the filepath is because Template.Parse(string source) expects the actual template contents, not a file path.
In order to accomplish what you are attempting, you need to use it this way:
Template template = Template.Parse(File.ReadAllText(templatePath));

It is hard to tell without the content of index.liquid, but there is already one thing to fix: you are calling Render two times, and the second without your object.
Try this:
public static void createHTML(DataTable table)
{
string templatePath = #"C:\Path\To\index.liquid";
var template = Template.Parse(templatePath);
using (StreamWrite file = new StreamWriter(#"C:\Some\Path\test.html"))
{
file.write(template.Render(Hash.FromAnonymousObject(new
{
table = DataMapper.createTable(table).toString()
})));
}
}
If this doesn't work, please update your question to add the content of index.liquid.

Run javascript in C# WPF Form

*Please note that this is not for a web based application, it's windows based.
I'm building an application where I will need the user to submit simple javascripts that will be run by the application.
The scripts will call functions that are part of the c# build.
An example:
C# code:
public void helloWorld()
{
Debug.WriteLine("hello world");
}
Javascript submitted by user:
helloWorld();
The JavaScript would be parsed by the application at runtime and then call the required functions in my C# code.
Why?..
My app will be used by people with very little programming experience, they enter very simple JavaScripts and the app will attempt to automate a few tasks on the users computer. So my reason for using JavaScript is because it's simple and very easy to learn for someone with little experience.

It sounds like you want a JavaScript parser for your application. To be honest, I dont think what you're doing is possible, considering the context of the script and your code is different. However, this project seems to be doing something that may get you to the right place:
http://javascriptdotnet.codeplex.com/
Personally, I would think making some kind of XML format would be useful (like how UrlRewriter.net makes rewriting URLs easy):
<xml>
<commands>
<!-- Expose a Set of Condition Objects to Select From -->
<if condition="YourApplication.Conditions.RightClickOnDesktop">
<print text="HelloWorld" />
</if>
</commands>

Here is an example running a javascript code which, in turn, invokes a c# method
[System.Runtime.InteropServices.ComVisible(true)]
public class CSharpClass
{
public void MsgBox(string s)
{
MessageBox.Show(s);
}
}
-
Type scriptType = Type.GetTypeFromCLSID(Guid.Parse("0E59F1D5-1FBE-11D0-8FF2-00A0D10038BC"));
dynamic obj = Activator.CreateInstance(scriptType, false);
obj.Language = "Javascript";
obj.AddObject("mywindow", new CSharpClass(), true);
var result = obj.Eval(
#"
function test(){
mywindow.MsgBox('hello');
}
test();
"
);

Why do you "need the user to submit simple javascripts"? What is your application and what do users need it to do? Why have you decided a scripting language is the way to do this? I'm not saying that is the wrong answer, but that you have not justified this conclusion.
If your app will be used by "people with very little programming experience" I do not recommend implementing a scripting language. Basic concepts like source code and variables are very difficult for non-programmers to understand.
I suggest first investigating macro recording for user scripting. For .NET there is UI Automation and the White automation framework.

Silverlight C# - ComponentOne Spellchecker not loading dictionary

This may be a long shot, but I'm using ComponentOne's Spellchecker control for Silverlight. I made a test project, added a plain textbox and a button to it, added the references to the C1.Silverlight and C1.Silverlight.SpellChecker bits, and added the dictionary file to my project.
In the code, I called up the spellchecker on button1's click event and it worked SPLENDIDLY. The spellchecker dialog shows up, and works exactly as it should.
Since that test was successful, I then tried to implement this into my existing project. I've had no success for absolutely NO reason that I can determine, since I used the EXACT SAME code.
Here's the code I use to call the component:
using C1.Silverlight;
using C1.Silverlight.SpellChecker;
using C1.Silverlight.Resources;
public partial class MainPage : UserControl
{
C1SpellChecker spellChecker = new C1SpellChecker();
public MainPage()
{
InitializeComponent();
spellChecker.MainDictionary.LoadAsync("C1Spell_en-US.dct");
}
private void btnSpelling_Click(object sender, RoutedEventArgs e)
{
var dlg = new C1SpellDialog();
spellChecker.CheckControlAsync(txtArticle, false, dlg);
}
The references to C1.Silverlight and C1.Silverlight.Spellchecker are added to this project as well, and the dictionary as been added in the same fashion as well. The issue seems to be that for whatever reason the dictionary is not loading, because the spellChecker.Enabled method returns whether or not the main dictionary has been loaded. If I call MessageBox.Show("SpellChecker Enabled = " + spellChecker.Enabled.ToString()); it shows false, even though the call to load the dictionary is there (as you can see).
What would cause the dictionary to not load? Have I added it to my project incorrectly somehow?
EDIT: I suspect that I have added the dictionary to the project incorrectly, because the ComponentOne reference states:
If C1SpellChecker cannot find the
spelling dictionary, it will not throw
any exceptions. The Enabled property
will be set to false and the component
will not be able to spell-check any
text.
I just don't know what's wrong though because it was added in the same way that it was in the test project (Right clicked on the project.web->Add->Existing Item)
As always, thank you!
-Sootah

You could add the dictionary to the Silverlight app as an embedded resource and then load it using this code:
public MainPage()
{
InitializeComponent();
// load C1SpellChecker dictionary from embedded resource
var asm = this.GetType().Assembly;
foreach (var res in asm.GetManifestResourceNames())
{
if (res.EndsWith(".dct"))
{
using (var s = asm.GetManifestResourceStream(res))
{
sc.MainDictionary.Load(s);
break;
}
}
}
}

I think this post is duplicated in our forum as well, but will answer first here. Please try this:
1) Try to access the .dct file using your browser. If you cannot see it, it's probably because your web server is not serving that type of files. You need ton configure the web server to allow it.
2) verify the URL you are using is correct.http://helpcentral.componentone.com/CS/silverlight_161/f/78/p/86955/241328.aspx#241328
3) Check you are setting everything correctly: http://helpcentral.componentone.com/CS/silverlight_161/f/78/p/81924/227790.aspx#227790
Hope this helps!

Getting rid of Microsoft AJAX

We wrote a couple of controls using Microsoft AJAX (cs class + js class). Now I'm thinking about getting rid of it (just use jQuery), as it is bloated and we don't use UpdatePanel. My question is: how should I change my controls? Right now they implement IScriptControl, which uses Microsoft AJAX features (if I understand correctly, ScriptManager & ScriptControlDescriptor classes). What to use instead?
CLARIFICATION. I don't need some more JavaScript libraries - I'm already using jQuery and would like to minimize additional includes (unless they are really small). What I need is replacement for ScriptManager and IScriptControl interface. Things like:
Registering script references (and
not duplicating them).
Instantiating script class associated with control.
Binding my class to DOM element (what is the best way to do that using jQuery, btw?).
Initializing JS class fields.

On the other hand, you can try comet ajax approach, check these samples.

Ok, I finally finished it.
We ended up creating Page descendant with the some code in it to substitute for script manager (see below). We call to it from control's OnPreRender method using control's Page property. Thanks to guys from http://dj.codeplex.com/ for providing example of how to do it.
2, 3, 4. We used jQuery.data method to bind instances of script classes to DOM elements. We execute instantiation, initialization and binding code using jQuery.ready method. This code is added to control in its Render method using AddScript method (see below). Maybe later we would use JavaScriptSerializer for passing values from C# control to javascript classes, but at the moment we do it by hands, passing parameters to javascript class constructor.
HashSet<string> scriptReferences = new HashSet<string>();
HashSet<string> cssReferences = new HashSet<string>();
List<string> styles = new List<string>();
public void AddScriptReference(string url, bool resolve)
{
string realUrl = url;
if (resolve)
realUrl = ResolveClientUrl(url);
if (!scriptReferences.Contains(realUrl))
{
scriptReferences.Add(realUrl);
Header.Controls.Add(
new LiteralControl(
"<script type='text/javascript' src='" +
realUrl + "'></script>"));
}
}
public void AddCssReference(string url)
{
if (!cssReferences.Contains(url))
{
cssReferences.Add(url);
HtmlLink link = new HtmlLink();
//link.Href = ResolveClientUrl("~/jQuery-ui/css/ui-lightness/jquery-ui.css");
link.Href = url;
link.Attributes.Add("type", "text/css");
link.Attributes.Add("rel", "stylesheet");
Header.Controls.Add(link);
}
}
public void AddCssStyle(string style)
{
styles.Add(style);
}
protected override void OnPreRenderComplete(EventArgs e)
{
base.OnPreRenderComplete(e);
Header.Controls.Add(
new LiteralControl(
"<style type='text/css'>" + styles.Join("\n") + "</style>"
)
);
}
public static void AddScript(HtmlTextWriter writer, string script,
bool executeWhenReady)
{
writer.AddAttribute(HtmlTextWriterAttribute.Type, "text/javascript");
writer.RenderBeginTag(HtmlTextWriterTag.Script);
if (executeWhenReady)
{
writer.Write("$(function(){\n");
}
writer.Write(script);
if (executeWhenReady)
{
writer.Write("});\n");
}
writer.RenderEndTag();
}

jQuery and jQuery UI are very powerful. But you also have access to Moo Tools and Prototype. Which tools in MS AJAX are you using? There are pretty much something comparable in open source non-MS offerings across the board. They just require slightly more labor on your part to implement. You might also look at Telerik and Syncfusions controls. Telerik has a bunch of open source ajax offering for ASP.NET MVC which can be converted to ASP.NET Web Forms pretty easily.

You may be interested in the more lightweight conditional rendering feature in the upcoming release of .NET4 (along with Visual Studio 2010)
http://msdn.microsoft.com/en-us/magazine/ee335716.aspx
If you can wait, it might be a viable solution for you.

I'm not sure you need a replacement for IScriptControl and the Scriptmanager if you want to strictly use jQuery. I recommend you check out some http://blog.jeremymartin.name/2008/02/building-your-first-jquery-plugin-that.html'>tutorials on creating jQuery plugins.