Is it possible to section a very long string into a matrix?

I am trying to break up a very long character string into sections that I can put into a matrix.

example: '56748946489148461898434919946189189454961894564349894389' and I want to extract the numbers in groups of say 4 [(5674),(8946), etc], and put them into a (25,1) matrix, each group of 4 being the content for the row. Is there a way to use a reshape or a loop, or any other ideas about how I would do this?

Answer 1

[Matt J]

You can do as follows, although I don't know what you mean by a 25x1 matrix. If each row is to contain 4 digits, then it would have to be 14x4

    >> A='56748946489148461898434919946189189454961894564349894389';
    >> B=reshape(A,4,[]).',

    B =

    5674
    8946
    4891
    4846
    1898
    4349
    1994
    6189
    1894
    5496
    1894
    5643
    4989
    4389

    >> whos B
      Name       Size            Bytes  Class    Attributes

      B         14x4               112  char

[Matt J]

Or if you mean that you want to convert the strings to actual numbers, you can do something like this

      >> B=(reshape(A,4,[]).' - '0')*10.^(3:-1:0).'

      B =

              5674
              8946
              4891
              4846
              1898
              4349
              1994
              6189
              1894
              5496
              1894
              5643
              4989
              4389

      >> whos B
        Name       Size            Bytes  Class     Attributes

        B         14x1               112  double

Answer 2

[Cedric]

One way would be:

 >> s =  '56748946489148461898434919946189189454961894564349894389' ; 
 >> num = sscanf(s, '%4f')
 num =
        5674
        8946
        4891
        4846
        1898
        4349
        1994
        6189
        1894
        5496
        1894
        5643
        4989
        4389

EDIT 1: Matt answered while I had the answer opened and had to take a break for managing a hungry cat ;-) .. so, note that his answer is more efficient than SSCANF. I leave my answer for the record though.

EDIT 2: you can check this out:

http://www.mathworks.com/matlabcentral/answers/72285-how-to-read-formated-data-with-implied-decimal-point

to get an idea about how "more efficient" Matt's answer is actually.

[Matt J]

I'm a little surprised to discover that SSCANF is so slow!

[Cedric]

Yes, I was too, but parsing format spec is probably time consuming even when the string is short/simple (if only to detect that it is simple). The input parser probably also adds a non-negligible overhead.

[Jan]

Strange. SSCANF is surprisingly fast for some tasks, e.g. for converting a cell string to a numerical vektor. See e.g. FEX: Fast str2double, which should be much faster actually than this naive M-version:

d = reshape(sscanf(sprintf('%s#', c{:}), '%g#'), size(c));

Inspite of the time-consuming creation of the large temporary string, this beats a C-implementation, which scans each string separately by sscanf(), strtod() of the C-libs. Therefore I conclude that sscanf must be extremely efficient for this case. Perhaps use '%4d' instead of '%4f'.