A client is having problems reading a file created with our software. I'm reading a Diagram from a stream as follows:

try
{
using (Stream stream = new FileStream(pathname, FileMode.Open, FileAccess.Read, FileShare.Read))
{
flowChart.LoadFromStream(stream);

// read other data
.
.
}
}

catch(Exception ex)
{
// Exception caught as follows:
// End of stream encountered before parsing was completed
}

I'm getting an exception thrown from the LoadFromStream function. I don't have the source code for FlowChart .NET. Is there a diagnostic I can use see what went wrong? Other files created with the same version of the software are reading fine.

Regards
DavidL

Stoyan

I don't use ControlNodes but each ShapeNode does have a Tag attached.  They have not been changed recently, but when they are changed I don't delete any fields and new fields are only ever added at the end of the class.  I also think that the file was saved and then read with the same version of the software, but I'm checking on this now.

File is too big to attach. Can I email it directly?

DavidL

Our developer traced through the deserialization code, the last thing it tries to do is load a BoxDataL tag value for ShapeNode with ZIndex = 1269. At that point the binary formatter reaches the end of the stream. If you look at the end of the file using VS hex editor you will see it cuts off abruptly at the position of some type descriptor for a string (perhaps for the Swim_Agent field). The List<int> field before it was loaded successfully.



Our guess is that either there was a full-disk condition when saving, or some transfer / download hasn't completed successfully if your customers exchange files by email or http. Also from your code it seems you write additional values into the stream; if you use the stream.Seek method check if there isn't some chance for it to return current position to a point inside diagram's data.

All data up to that point is well formatted, and from the 3000 diagram items in the file about 2850 were read successfully. If the customer doesn't have a backup we might try to extract them for you. Swimlane information and some groups are missing from the file too.

I hope that helps,
Stoyan

This build adds a TryLoadCorruptedFile method that should let you load the file in your application:
https://mindfusion.eu/_beta/fcnet634.zip

Otherwise you would not be able to load tag values from file saved by us as their types will be defined in a different assembly (our test project). That method manages to load 2242 nodes and their text and tag values.

Unfortunately, the information for ConnectionPoint objects used to connect nodes to links is missing from the file you sent us, along with control point coordinates, so the loaded DiagramLink objects aren't of much use. If you have some Tag values associated with links that will help you identify target nodes, try reassigning links' Origin and Destination values. If you store some kind of parent-node identifiers in nodes' Tags, you could delete the unconnected links and create new ones from parent to child nodes.

I hope that helps,
Stoyan