[Paraview] Accessing point data from python

[Tony McDaniel]

I'm currently using the csv writer to create a CSV file for subsets of 
data, then reading those csv files in to generate the numpy arrays for 
additional analysis, report generation, etc.

Is it possible to bypass the csv step, and directly convert the point 
data to numpy arrays?

A separate, but related question has to do with the file naming for CSV 
files. If I save data to slice.csv, Paraview will write to slice0.csv, 
(and sometimes slice1.csv, slice2.csv, etc.). Usually the other files 
are empty, but since the name passed to the csv writer is different from 
the name of the csv file that I have to read in, I'm having to change 
the name in the script. Is there an option to force Paraview to only 
write to a single CSV file? Is there a limit to the number of rows of 
data per file?

Thanks,

Tony