Strange issue encountered using textscan and cell arrays

When I run the following snippet of code:

str = '(3.3940269999999999,1,0) (0,3.3940269999999999,0) (0,0,2.024902)';
format = '(%f, %d, %d) (%d, %f, %d) (%d, %d, %f)';
A = textscan(str, format);

matrix = [A{1,1}, A{1,4}, A{1,7} ;  A{1,2}, A{1,5}, A{1,8}; A{1,3}, A{1,6}, A{1,9}]

I get the following output matrix =

           3           0           0
           1           3           0
           0           0           2

Why were the floating point numbers converted to integers?

Answer 1

[Star Strider]

When I run a slightly modified version of your code (because format is a function that defines the format displayed in the Command Window and Tooltips), I get:

format_string = '(%f, %d, %d) (%d, %f, %d) (%d, %d, %f)';
A = textscan(str, format_string);

matrix =

    3×3 int32 matrix

     3   0   0
     1   3   0
     0   0   2

That is likely because of the '%d' format descriptors. Convert them to '%f':

format_string = '(%f, %f, %f) (%f, %f, %f) (%f, %f, %f)';

to get:

matrix =

          3.394            0            0
              1        3.394            0
              0            0       2.0249

[Walter Roberson]

Right. %d format descriptors are explicitly for converting whatever result that is read into int32, as by int32(whatever_was_read)

[Peter Ryan]

Thanks!

[Star Strider]

My pleasure!

Answer 2

[Eamon]

I think if you use the format function like below before the rest of your code, it will show the original floating point numbers.

format long

[Eamon]

Nevermind, I tried this and it doesn't work.