[Paraview] Custom plugin reader with time series

[Dominik Szczerba]

Hi again,

Unfortunately, this was a pure coincidence and the problem is somewhere
else (at the end, a segfault occured as well). I am still very grateful 
for suggesting this path (deriving from Algorithm and not Reader) as I 
would not discover the problem without it. I still believe the problem 
is not on my side. Will eventually report my findings in a new thread.

bye
Dominik

Dominik Szczerba wrote:
> Dear Mike,
> 
> Sorry for warming up the old meal but it is only now that I could fully 
> appreciate your solution. After long weeks of frustration with my hdf5 
> reader derived from vtkDataReader - and subsequent segfaults reported 
> before - I tried your way of deriving from UnstructuredGridAlgorithm : 
> and my problem is magically gone, with no other changes than that. I 
> have not yet tried time series, so it is not unlikely I will have some 
> questions later. For the moment, thanks a lot.
> 
> Dominik
> 
> Mike Jackson wrote:
>> Take a look at 
>> <http://titanium.imts.us/viewvc/Task_4/PVDislocation/src/DislocationReaders/vtkPrdsMultiOutDislocationReader.cxx?view=markup> 
>>
>>  to see how I did a time series reader.
>>
>> Basically in RequestInformation  you will need to determine the the
>> range of times that your data covers, probably by opening each file or
>> some other means. Then set
>>     double timeRange[2];
>>     timeRange[0] = lowestTimeValue
>>     timeRange[1] = highesetTimeValue
>>     outInfo->Set(vtkStreamingDemandDrivenPipeline::TIME_RANGE(),
>> timeRange, 2);
>>
>> Also, you will need to have the following in your header:
>>
>>    vtkSetMacro(TimeStep, int);
>>    vtkGetMacro(TimeStep, int);
>>
>>  int GetNumberOfTimeSteps();
>>
>> Now, down in the RequestData() Method you can get the requested time
>> step from the
>> pipeline and then open and read the data from the proper data file.
>>
>> int tsLength = 
>> outInfo->Length(vtkStreamingDemandDrivenPipeline::TIME_STEPS());
>>   double* steps = 
>> outInfo->Get(vtkStreamingDemandDrivenPipeline::TIME_STEPS());
>>   // Check if a particular time was requested.
>>   if(outInfo->Has(vtkStreamingDemandDrivenPipeline::UPDATE_TIME_STEPS()))
>>   {
>>     // Get the requested time step. We only supprt requests of a 
>> single time
>>     // step in this reader right now
>>     double *requestedTimeSteps =
>> outInfo->Get(vtkStreamingDemandDrivenPipeline::UPDATE_TIME_STEPS());
>>     this->TimeValue = requestedTimeSteps[0];
>>
>>     // find the first time value larger than requested time value
>>     // this logic could be improved
>>     int cnt = 0;
>>     while (cnt < tsLength-1 && steps[cnt] < this->TimeValue)
>>       {
>>       cnt++;
>>       }
>>     this->ActualTimeStep = cnt;
>>   }
>>
>>   if (this->TimeStep > -1)
>>   {
>>     this->ActualTimeStep = this->TimeStep;
>>   }
>>
>>   if (this->ActualTimeStep >= this->TimeStepRange[0] &&
>> this->ActualTimeStep <= this->TimeStepRange[1]) {
>>     vtkSmartPointer<DislocationReader> reader =
>> vtkSmartPointer<DislocationReader>::New();
>>     reader->SetFileName(this->FileName);
>>     reader->SetTimeStep(this->ActualTimeStep);
>>     //reader->SetInformationVector(outputVector);
>>     reader->SetFileOffsets(this->FileOffsets);
>>     reader->read();
>>     vtkUnstructuredGrid* grid;
>>     grid = reader->GetDislocations();
>>     dislocations->ShallowCopy(grid);
>>
>>    // dislocations->SetBlock(0, grid);    // Block 0 is the 
>> Dislocation Data
>>     grid = reader->GetSimulationBox();
>>     simBox->ShallowCopy(grid);
>>     //dislocations->SetBlock(1, grid); // Block 1 is the Simulation 
>> Limits
>>   } else {
>>       std::cout << "vtkPrdsMultiOutDislocationReader: Timestep was out
>> of Range: " << this->ActualTimeStep << std::endl;
>>       std::cout << "vtkPrdsMultiOutDislocationReader: Timestep Range:"
>> << this->TimeStepRange[0] << " -> " << this->TimeStepRange[1] <<
>> std::endl;
>>       return 0;
>>   }
>>
>>
>> This code is specific to my case where I have a whole separate reader
>> for my data where I feed it the name of the file to read and then tell
>> it to read the file and get the output. This may also be the case for
>> you also.
>>
>> Hope some of the helps.
>>
>> Mike
>>
>> On Sat, May 10, 2008 at 6:03 AM, Dominik Szczerba
>> <domi at vision.ee.ethz.ch> wrote:
>>> Hi,
>>> I have successfully implemented a reader for my data as a PV plugin. It
>>> loads separate files correctly, refuses, however, to read the whole 
>>> series
>>> (file_01.h5, file_02.h5, etc.). How do I go about achieving the same 
>>> effect
>>> as the native readers to have a series as transient data?
>>> Thanks for any hints.
>>> -- 
>>> Dominik Szczerba, Ph.D.
>>> Computer Vision Lab CH-8092 Zurich
>>> http://www.vision.ee.ethz.ch/~domi
>>> _______________________________________________
> 

-- 
Dominik Szczerba, Ph.D.
Foundation for Research
on Information Technologies in Society
http://www.itis.ethz.ch