Parse a cell array of strings without looping

I've got a nx1 cell array of strings that correspond to spatial coordinates. Here's an example where n=4,

'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'

Each row of the cell array is a string containing x,y, and z coordinates separated by a comma. I need to parse each string to separate the coordinates, convert the coordinates to numbers, do an operation on them, then put them back in their original format. Is there a way to do this without looping through each row of the cell array? Any tips or insight would be greatly appreciated.

Regards, Justin

[per isakson]

Why are these strings in a cell array? Did they originate in a file?

[Justin Solomon]

Yes. They are read in using the textscan() function.

[per isakson]

Why not parse the data with textscan when reading? If you need the cell array of strings read twice with textscan. When the file is in the system cache it is faster to parse with textscan( fid, ...) than to parse the cell array of strings.

[Justin Solomon]

The file has stuff in a format other than what is listed above. The text file actually looks like this (but much longer of course):

head

39 :M

63 :N

U Knot Vector

0

0

0

0

0.0167

0.0333

0.05

.

.

.

Control Points

122.8233,216.1323,671.8740

122.8233,216.1323,671.8740

122.8233,216.1323,671.8740

122.8233,216.1323,671.8740

.

.

.

chest_surface

50 :M

53 :N

U Knot Vector

.

.

.

My strategy is to read in the whole file with textscan() then parse it after I have the whole thing read in. I suppose I could parse everything while reading in. This will still require me looping through each line and checking if its a line that I want to mess with right? The overall goal is to read in the file, change all the 'Control Points', then write out a new file. Is this possible without looping?

Thanks for your insight.

[Azzi Abdelmalek]

this is another question

[per isakson]

With 'Headerlines' and 'N' of textscan it would still be possible to read and parse in one step. However, you need to find out the values of the two first.

Answer 1

[Azzi Abdelmalek]

clear
A={'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'
'122.8233,216.1323,671.8740'}
B=cellfun(@(x) regexp(x,',','split'),A,'uni',false)
out=cell2mat(cellfun(@(x) [cellfun(@(y) str2num(y),x)],B,'uni',false))

%or easier

 out=str2num(cell2mat(A))

[Sogand Karbalaieali]

Thank you for : out=str2num(cell2mat(A))

Answer 2

[Matt Fig]

C = {'122.8233,216.1323,671.8740'
     '122.8233,216.1323,671.8740'
     '122.8233,216.1323,671.8740'
     '122.8233,216.1323,671.8740'};

% Extract the numbers:
D = cellfun(@str2num,C,'Un',0);
% Do the 'operation' on the coords with cellfun:
E = cellfun(@(x) x+[1 2 3],D,'Un',0)
% Put them back in original format.
F = cellfun(@(x) sprintf('%.4f,',x),E,'Un',0)

Note that the step from D to E is ambiguous given your description, but you get the idea.

Answer 3

[per isakson]

On possibility for your first step - if the file is not huge:

    str = {
    '122.8233,216.1323,671.8740'
    '122.8233,216.1323,671.8740'
    '122.8233,216.1323,671.8740'
    '122.8233,216.1323,671.8740'
    };
    M = str2num( char(str) );

or I didn't get it.

[Justin Solomon]

My file is pretty big (> 1 million lines). This works pretty well despite the file size! Any ideas on how to convert M back to str after I perform the desired operation?

[per isakson]

cellstr( num2str( M, '%f,%f,%f\n' ) )

However, write the numerical array directly to the file with fprintf in a loop I guess that fast enough. Otherwise, transpose the numerical array and write with fprintf, but only if profile showed that it could be worth the trouble.

[Justin Solomon]

Perfect. Thanks everyone for your help. I didn't know I could only accept one answer or I would have accepted everyone's.

Justin