Reading CSV data and preserving empty cells?

[forkandwait]

Hi all,

Does anyone have any advice on reading csv files with a function that converts 
empty data into NaN's.  

dlmread converts missing columns into zeros, which is probably matlab 
compatible but also pretty stupid (golly, is a missing data point the same 
thing as a measurement of zero... duh...).   textread would probably work, but 
I don't want to write a format string for every dataset with 200+ columns.  
Perhaps an awk script to pre-process, but that sounds gross.

csv2cell in octave-forge converts them into blank text -- this is approaching 
reasonable, but how to do I mass-convert all the blanks into NaNs?

I think I will probably try to figure out a cellfun/ cellidx solution tomorrow, 
but if anyone has done this before, please share your approaches.

Best!