I am using the WebBrowser control in my project to display complex HTML documents that are generated/manipulated at runtime.
I have noticed that constructing the DOM programmatically from C# by creating HtmlElement objects is about 3x slower than generating an HTML string and passing it to the WebBrowser, which in turn parses it to generate the DOM. Both ways create a noticeable delay when navigating between lengthy documents.
I am looking for the fastest way to switch between multiple documents in the same WebBrowser control, ideally without having to repeatedly generating the DOM tree for each document. Is it possible to cache a tree of HtmlElement objects somewhere in my program, and then re-insert them into the WebBrowser as needed?
I will describe the solution in terms of the native win32 COM APIs; it shouldn't be too difficult to write the interop to do it in C# (or find it at pinvoke.net). Alternatively, you may need to use the properites that the managed objects expose to get the native ones.
You're not likely to be able to build the DOM yourself faster than IE's parser, so create a blank HTMLDocument (which in native code would be CoCreateInstance(CLSID_HTMLDocument)) and QueryInterface() the HTMLDocument for its IMarkupServices implementation. Also create two IMarkupPointers using the IMarkupServices::CreateMarkupPointer() method.
Next call IMarkupServices::ParseString() to parse your HTML. This will give you a pointer to an IMarkupContainer that contains your DOM, as will as two IMarkupPointers that point to the beginning and end of you DOM. Now you can use IMarkupServices::Move() to move your data from one IMarkupContainer to another.
So the general scheme you would use is to have a single HTMLDocument which is your "display" document, and it's associated IMarkupContainer (which you can just QueryInterface() for). Then you have a vector or list or whatever of all the non-displaying markup containers. Then you just create a markup pointer for your display doc, call IMarkupPointer::MoveToContainer(displayDocumentContainer, true) and then use that to move stuff around from your display container to the not-displaying containers and vice-versa.
One thing to note: you must only access these objects on the thread you create them from, or acquire them on. All IE objects are STA objects. If you need multi-threaded access, you must marshal.
If you have specific follow up questions, let me know.
References:
IMarkupContainer
IMarkupServices
Introduction to Markup Services
This will do it
// On screen webbrowser control
webBrowserControl.Navigate("about:blank");
webBrowserControl.Document.Write("<div id=\"div1\">This will change</div>");
var elementToReplace = webBrowserControl.Document.GetElementById("div1");
var nodeToReplace = elementToReplace.DomElement as mshtml.IHTMLDOMNode;
// In memory webbrowser control to load fragement into
// It needs this base object as it is a COM control
var webBrowserFragement = new WebBrowser();
webBrowserFragement.Navigate("about:blank");
webBrowserFragement.Document.Write("<div id=\"div1\">Hello World!</div>");
var elementReplacement = webBrowserFragement.Document.GetElementById("div1");
var nodeReplacement = elementReplacement.DomElement as mshtml.IHTMLDOMNode;
// The magic happens here!
nodeToReplace.replaceNode(nodeReplacement);
I'd really need to know more about how you are generating these documents. It might be faster to get your data into a XML document and then use a XSL transform to convert the data to HTML and pass that to the WebBrowser control.
The nice thing about the XSLT implementation of .NET is that it takes the XSL source and compiles it to a temporary assembly to speed up the transforms.
If you decide to go that route look up the MVP.XML project which adds some nice exslt functionality to the stock XSLT implementation.
Maybe rather than caching the DOM you could just flip between several WebBrowser controls on the form - with only the active one being visible?
Could you do something like this?
Create the contents you want to display inside a DIV
Create secondary contents (in the background) inside non-visible DIVs
Swap the contents by playing with the visibility
Related
I am trying to pass a simple core data objects info from a tabBarController to its subviews so that they each reference a different attribute of that object. As a newbie, I'm not sure even where to start. It doesn't seem to be as simple as passing the data from one tableView to another...
Thank you for any help.
If you are sharing the same object between (most of the) the view controllers of your tab bar controller, maybe the best architecture for this would be to have one central data object.
A typical pattern is a singleton, some kind of data manager that provides the object, but maybe that is overkill. Another is to keep references to all view controllers and update them one by one when something changes - also not very elegant.
What you really want is something like a global variable. You could (ab)use your app delegate (just give it a property that points to the object) or if you prefer even your tab bar controller (make a subclass, give it a property). In the latter case, every view controller could then get the object like this:
NSManagedObject *object = [(MyCustomTabBarController*)self.tabBarController object];
For example, you can check for changes and refresh your views in viewWillAppear.
A UITabBarController should be handling other view controllers, not handling data objects. How does the tab bar controller get the object reference in the first place? And what is the object you're sharing?
Let each of your subordinate VC's keep a pointer to the object, and then they can each follow the appropriate keypath to get to the entities they're designed to handle.
Tim Roadley's book Learning Core Data for iOS, in chapters 5 and 6, shows how to pass an object from one view controller (a table view) to a detail view. It doesn't sound like that's what you're asking, but just in case...
In response to comment:
I'm looking at a tableview, tap a cell, and then a tab bar controller slides in? That's not the usual visual metaphor for a tab bar; it's meant for changing modes for the entire program. See the Music app for a typical example: songs, playlists, artists.
But if you really need to do it that way, try this (I'm assuming you're using storyboards):
In prepareForSegue: in your tableview controller, tell the destination (tab bar controller) what object it's working with.
In the tab bar controller's -viewWillAppear, tell each of its tabs about the attribute: self.frobisherViewController.frobisher = self.myWidget.frobisher.
You could instead tell each of the component tabs about the top level object: self.frobisherViewController.widget = self.myWidget. But I like the first approach better because there is less linkage. The frobisherViewController now would need to know about both widgets and frobishers.
This ended up being very simple. I was trying to call the object in the child views initWithNibName which doesn't work. I ended up creating a setObject function and calling the properties I wanted in viewWillAppear.
Hope this helps someone.
I'm currently building a custom control to be used as an application's view navigator (classic OneUI style). First of all: this is a 8.5.3 based project, and unfortunately there's no way to use Extlib stuff or other extensions / plug-ins. So we have to build all the tricky stuff ourselves using only what came "out-of-the-box" with 8.5.3.
I'd llike to build my navigator using a repeat control containing <xp:link> controls. Datasource could be an array of javascript objects where each single object would look like this:
var navEntry = {"entryLabel" : "label-of-link",
"entryTarget" : "target-url-of-link",
"entryStyle" : "style-to-emphasize-selected-link"}
Each array element then would describe a single navigator entry.
Since the navigator will be used in all possible "DominoView" Xpages it yould make sense to build the array of JS objects at the Xpage level and then pass that info into the custom control.
I'm aware that there are multiple ways to do this, and one could be "Custom Control Properties". If there was a way to pass my JS object array.
(Hope I could make clear what I'm trying to do here...)
That object looks like a HashMap to me really. You should be able to pass that in to a custom control via custom property if you use the type java.util.HashMap I'd think. You'll need to type it in I'm sure. I've passed custom objects in by using java.lang.Object.
The custom control will get loaded during the Page Load event, and usually properties have to be available at that point. If they're loaded during the Render Response phase, that's too late. So your SSJS object will need to be Compute on Page Load.
To use contents of a repeat control, you would need to set repeatControls=true, otherwise the repeat is only built during render response. Until then it's just a single set of controls with no data in them. However, Im pretty sure repeatControls="true" means you only get the number of rows you define. You can't change it via a pager.
You can manually define the type of the custom property. For a standard SSJS Object you use "com.ibm.jscript.std.ObjectObject", for a SSJS Array you use "com.ibm.jscript.std.ArrayObject" etc. As editor for the custom property, I set the string editor ("String value").
I' working on a library control for Xpages and need some help in creating.
I would create a control which reads a configuration file and creates controls in a table, controls like Editboxes, checkboxgroups and so on.
so and now to my questions:
could I initiate controls from the Exlib or must I implement them all by my self?
if I could use them from the Exlib could anyone explain me how?
I hope its clear what i mean if not please ask me for further informations.
When creating your own components, if you're closely replicating some behavior that is already in an extension library component, I highly recommend you extend that component and just add what's needed to accommodate your different functionality. This makes things much easier and you don't have to code around every little scenario that the component might be placed in.
But, if you are developing a component that is nothing like any of the extension library or core components then just ensure your component extends UIComponent or UIComponentBase. If going this route, you'll also need to create your own renderer which extends Renderer. This is what will build the on-screen representation of your component. Again, if there's already something in the core components or extension library components that closely mimics what you need then make your renderer extend that renderer. Also, don't forget to include the renderer definition in the faces-config file and the component definition in the xsp-config file or your component won't work.
As for initiating controls from the extlib.... I assume you mean can you inject them onto the page at runtime. If so the answer is absolutely yes. To add an input text field to the page where there is a container (i.e. panel, div, span, whatever) with an ID of "someContainer"
XspInputText input = new XspInputText();
input.setValue("someValue");
input.setId("someID");
UIComponent container = FacesContext.getCurrentInstance().getViewRoot().findComponent("someContainer");
container.getChildren().add(input);
To see the api for all of the core and extension library components take a look at the XPages Controls Documentation. For a more complete tutorial on creating your own components take a look at my blog for creating a custom component inside an nsf, the steps are pretty much the same for putting them into a library:
Part 1,
Part 2 and there is an example database in the Part 2 post.
This is a noobish question but I am getting started with MS Coded UI Tests. And I was wondering if there's a way to find page elements using XPath, instead of the default matching mechanism? What I want to do, is match a parent elements and programmatically navigate down the DOM tree to get the elements I want to work with. This can be easily done with Selenium, but I am not sure how to do it with Coded UI Tests.
Thanks
You should be able to manage navigating an xpath using the UITestControlCollection. Use CodedUI's recorder to get to the top level control, then use GetChildren to navigate your way. Keep in mind that the xpath changes because all the object types are similar, CodedUI's API doesn't distinguish.
Example:
HtmlDocument doc = this.UIYourWindowName.UIYourDocumentName; // mapped control
doc.Find();
UITestControl toline = new UITestControl(doc);
toline.SearchProperties["Id"] = "to_d"; // use the id of the top most control
UITestControlCollection toline1 = toline.GetChildren(); // get the child objects
toline1 = toline1[0].GetChildren(); // xpath: \\ctrl[#id='to_d']\item[0]
toline1 = toline1[0].GetChildren(); // ctrl[]\item[0]\item[0]
// and so on...
Our web application has some event code to "format" any text with pasted into a field so that any HTML styles do not break our data.
What would be a good way to pre-load the browser clipboard so that I can test pasting into the input field?
Is there any way to do it programmatic-ally or would I have the test script visit "a source page" and copy text before moving onto our application?
Any ideas or code snippets would be welcome.
Working with the clipboard will depend on your platform. E.g. on OS X, you can use pbcopy and Command-V:
open('|pbcopy', 'w') { |io| io << 'some text' }
browser.text_field(:name => 'q').send_keys([:command, 'v'])
I know there are equivalents on Linux (xclip?). Not sure about Windows.
I would consider using the .value= method to set the value. If it's been implemented the same as in watir, then it causes no events to be fired and sets the value directly, and then follow that up by sending an appropriate event (depending on what if any events are being monitored) such as onKeypress. I tried to figure out from the Watir-webdriver rdoc for textfield, if this distinction between .set and .value= has been maintained, but the way the doc describes them (at least there) makes them seem interchangable.. (Jarib can you clarify???)
Potentially you might need to first fire something like onFocus depending on the controls you are using. For example, as described in this SO case Setting a text field that has a JQuery mask on it for a jquery mask, they had to end up firing an unmask event to be able to even set the field.
This is a good case for utilizing the techniques described here How to find out which JavaScript events fired? and in the SO item linked in the comments for that question, to figure out just what events are fired when you manually paste something into a field. (note I would slueth it with both using the mouse, but also using something like tab to move between fields and set the focus, events common to those two methods are the ones most likely to be implemented by the controls.
I presume you have some kind of client side javascript that needs to check what was pasted into the field, and thus the reason for this test. If you are using standard HTML fields, with no javascript stuff, then I would consider this particular test case to be effectively the same as 'testing the browser' since supporting cut and paste in input fields is a standard browser function. In that case, you are sort of 'off the reservation' and I'd not bother with such a test case.