|
Hi All,
I am working on developing a application (Windows Application) where in the application gets the content (majorly flash content) from a content server and takes the snapshot and gets the RGB data out into a buffer at x fps. This buffer is then passed to a display stack which handles the off-screen rendering.
I am using IECore as the renderer. I am able to capture the snapshot of the web content and save it as bmp file (shown below).
To get the content
AxWebBrowser web = new AxWebBrowser();
web.SetBounds(0, 0, 1024, 600);
web.Navigate2(ref URL, ref oNull, ref oNull, ref oNull, ref oNull);
Application.Run();
private Graphics grp;
private IntPtr hdc;
private IViewObject vw;
private IHTMLDocument2 doc;
public void Capture()
{
doc = (IHTMLDocument2)web .Document;
web.SetBounds(0, 0, width, height);
image = new Bitmap(width, height);
grp = Graphics.FromImage(image);
_RECTL bounds;
bounds.left = 0;
bounds.top = 0;
bounds.right = width;
bounds.bottom = height;
hdc = grp.GetHdc();
vw = doc2 as IViewObject;
vw.Draw(1, -1, (IntPtr)0, (IntPtr)0, (IntPtr)0,
(IntPtr)hdc, ref bounds, (IntPtr)0, (IntPtr)0, 0);
grp.ReleaseHdc(hdc);
image.Save(fileName);
image.Dispose();
}
}
I am hopefule that we will have some mechanism to extract the RGBA data out of it. But I am not getting the clue of handling the mouse events. here the rendering is happening to a memory and not to a window (off-screen rendering).
The display sdk (a C dll) requires the application to register a call back funtion to get the mouse related data. The information that I will get is the left down/up, move and x and y co-ordinates.
Now, what I want to know is, how to register a call back function and how to pass the mouse information to the IE Core so that it responds as if it was clicked on a normal window.
Any help is appreciated. Please do the needful.
Regards,
HalliHaida
|
|
|
|
|
There is a problem with reading large TCP data transfer.
Server on one end serializes some class object with BinaryFormatter and send it to the client over TCP. Serialized object in MemoryStream may take about 2Mb in size.
In the client I've got 1Mb byte[] array to read to with asynchronous NetworkStream.BeginRead and all the data is concatenated to MemoryStream object which is used as parameter passed to every BeginRead.
If more data available to read from server NetworkStream.DataAvailable that byte[] array is added to MemoryStream, otherwise deserialization happen and MemoryStream is set to 0 length for next transfer.
The problems are:
- data sent != data recieved in size
- synchronization failure
Server sent say 1.6Mb byte[] to client.
Client gets it in chunks of upto 1Mb which are added to MemoryStream for deserialization in case NetworkStream.DataAvailable == false
1) Instead of 1.6Mb byte[] array client reads 1Mb+1Mb+1Mb+0.9Mb = 3.3Mb and deserializes correct object.
2) Sometimes after 1Mb+1Mb it gets NetworkStream.DataAvailable == false thus deserialization fails, but 3ms later it reads remaining data 1Mb+0.9Mb and deserialization is successfull
3) Sometimes after immediatly 3-5ms after 1) it gets more data from server say 1Mb+0.3Mb chunks and tries to deserialize them again with a failure of course since the data is somewhat extraneous and I do not know who is sending it
If I set to 0 length MemoryStream in case of deserialization failure 3) is handled but 2) becomes invalid.
If I do not set to 0 length MemoryStream after deserialization failure, then after 3) any other deserialization attempt will be invalid since MemoryStream will contain that extraneous data in the beggining.
Also 1Mb+1Mb+1Mb+0.9Mb might be different as 1Mb+0.3Mb+0.5Mb+0.3Mb or 1Mb+0.7Mb for the same 1.6Mb transfers from the server.
Чесноков
|
|
|
|
|
Don't attempt to send such large chunks across the network. Break the stream up into smaller sizes (I find 4K blocks tend to be about right). Reading is easy with the following method:
internal static object DeserializeFromNetworkStream(IFormatter formatter, NetworkStream ns)
{
using (MemoryStream ms = new MemoryStream())
{
byte[] buffer = new byte[4096];
while (true)
{
int read = ns.Read(buffer, 0, buffer.Length);
if (read > 0)
{
ms.Write(buffer, 0, read);
continue;
}
ms.Seek(0, SeekOrigin.Begin);
break;
}
return formatter.Deserialize(ms);
}
}Serializing the object is just as easy:
private static void SerialiseToNetworkStream<T>(NetworkStream ns, IFormatter formatter, T data)
{
using (MemoryStream ms = new MemoryStream())
{
byte[] buffer;
formatter.Serialize(ms, data);
buffer = ms.GetBuffer();
int offset = 0;
int size = 4096;
int remaining = buffer.Length;
while (remaining > 0)
{
if (size > remaining)
{
size = remaining;
}
ns.Write(buffer, offset, size);
offset += size;
remaining -= size;
}
}
}
|
|
|
|
|
What is the reason for 4k limit?
I'm using asynchronous RW when sending and reading data, does it apply?
I had another 3rd party component which sends packets of about 400k.
You may recieve those packets at a maximal rate, say, up to 20 times per second.
If you do it with 4k and blocking read, reading operations takes too much time due to those chinking and you're limited to say 5 requests per second.
If you use as much as 400k or more you're in time to achieve that 20 times per second capture.
In other words with 4k limit reading those 400k takes 200ms as with 400k+ limit you can get them at 50ms.
Чесноков
|
|
|
|
|
4K works for me - you can up the limit as you see fit; I didn't say this had to be the one you chose. I wouldn't use too large a size though - if there's a problem, it's quicker to detect it with a smaller size than a larger one.
|
|
|
|
|
Message Closed
modified 23-Nov-14 5:56am.
|
|
|
|
|
That is error prone, suppose scenario 4) when some extraneous data was recieved.
The better is to use some custom header before serialized data.
Чесноков
|
|
|
|
|
In order to handle this situation correctly, you need a more sophisticated communications protocol between the server and client. In my case, from the server, I put an MD5 hash at the beginning of the message, along with a length, and then the message itself. When the receiver receives the message, it then calculates the MD5 hash on the received data. If they match, it then sends an ACK to the sender. If not, it sends a NACK and then the sender resends the message. The overall message itself can be as long as it needs to be.
You can define ACK and NACK however you want, I use the text ACK and NACK for mine, so if I want to look at the raw ethernet packets in ethereal or something like that, I can easily track what is happening.
I also do this for a UDP broadcast, except that there is no response from the clients. This ensures that the clients only act on known good messages from the server.
David
|
|
|
|
|
|
If that works for you, great. We've had problems in the past with unreliable network connections which caused real problems with larger buffer sizes, but if you can guarantee this is not a problem for you then great.
|
|
|
|
|
What were the problems? How small buffer size resolved the issue in your case?
Чесноков
|
|
|
|
|
Basically, they were to do with clients who had infra-red connections (don't ask). The smaller buffer size meant that we received drop out notifications straight away. Granted, we had multi-gigabyte files that were being transferred - the drop out allowed us to trigger reconnects to alternate servers.
|
|
|
|
|
How to check Stream contains valid BinaryHeader to avoid deserialization exceptions such as Binary stream '0' does not contain a valid BinaryHeader?
Чесноков
|
|
|
|
|
Could you post some code ??? so we can see what your trying to do. ?
With great code, comes great complexity, so keep it simple stupid... 
|
|
|
|
|
I do not want that deserialization call formatter.Deserialize(ms); will result in Exception. Thus I need to check that MemoryStream ms; contains valid header which is used internally in BinaryFormatter.Deserialize() call.
Чесноков
|
|
|
|
|
Its not that simple. please provide an sample code from where you get the error. else im unable to help you.
With great code, comes great complexity, so keep it simple stupid...
|
|
|
|
|
It is very simple, MemoryStream contains your data for deserialization.
With MemoryStream.GetBuffer(); you may get access to underlying byte[] array.
It needs to check that in that array first bytes are valid BinaryHeader structure used inernally by BinaryFormatter during deserialization call.
If there is garbage in the MemoryStream then BinaryFormatter will throw exception. I need to avoid that exception and make sure correct data will be passed for deserialization.
Finally I found that link that resolves the issue http://primates.ximian.com/~lluis/dist/binary_serialization_format.htm[^]
Чесноков
|
|
|
|
|
dose anyone one know how to change the opacity of a picture box or anything like a buttons opacity? and still keep the background of the form at 100%?
|
|
|
|
|
It depends what you are trying to do. Are you just trying to make a control transparent? If so, take a look at this[^] article.
|
|
|
|
|
Hi In my system i has Visual Studio Team System(2008), i don't have sharepoint. My system o/s is Windows XP. I want to use SPFarm commands in my code. Here my question is without installing sharepoint is it possible to use SpFarm commands by adding the reference? will it work? Please help me. I'm very new to Sharepoint.
Very urgent for me.
Thanks in advance.
|
|
|
|
|
Please do not post your question in multiple forums.
It's time for a new signature.
|
|
|
|
|
OK, This is driving me insane. Yes I am a beginner and I am trying but this is confusing me so much.
I have the following classes...(This is a patient simulation program)
1) Form (my main form which displays an ecg etc)
2) ecg class
I would like the ecg class to be able to draw itself (ecg.Draw())
Now, I tried to create a custom panel class derived from panel, and overide the onpaint method.
But I cannot seem to initialise it with a panal control on my main form, and nothing hapens.
Am I even doing this right in the first place.
Should I pass a panel object into the constructor og my ecg class, or should i set a static member of ecg in the form control and invoke paint from here?
I am so confused!
Thank you
Steve
|
|
|
|
|
Message Closed
modified 23-Nov-14 5:56am.
|
|
|
|
|
Hi.
Thats what I am trying to do, but the onpaint method never gets called
Steve
Also,
If I do it that way, how do I add custom functions to the control?
Thank you
Steve
|
|
|
|
|
It works fine for me, so try recreating the steps I used:
1) In solution explorer, right click your project, and add a user control. Call it something sensible.
2) Highlight the new control, and open the code.
3) Change the derivation from "UserControl" to "Panel" - you don't need to do this, but WTF. If you do, compile the program, and delete the line the compiler complains about (AuoScaleMode)
4) In your new control class, add the following:
protected override void OnPaint(PaintEventArgs e)
{
e.Graphics.FillRectangle(Brushes.Green, e.ClipRectangle);
}
Now, when you add your new control, it will fill itself with Green - this shows OnPaint in being called.
I would then add a Paint method to my ECG class, and hand it either e.Graphics so it can draw itself.
Real men don't use instructions. They are only the manufacturers opinion on how to put the thing together.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
