Assign an array of structs to a cell

Hi, I have a variable sized array of structs that I read and want to assign to cell within a cell array. I found a results for "Conversion to cell from struct is not possible." in prior questions but that maps a single struct to a single cell. I would like to be able to store an array of structs to a cell. Is that possible? How? Thanks, Jerry

[Rik]

Cells in Matlab are container variables in the truest sense of the word: use the syntax you would normally use and the replace the variable name with the name of the cell and the index in curly braces. It is as simple as that.

If you show the line of code you are attempting to execute, I'll try to modify it.

[Stephen23]

Jerry's "Answer" moved here:

Hi Rik, thanks for the quick response.

I believe this to be the gist of the code that I am using. Meanwhile I found a workaround by using a struct of structs... so if this isn't possible, the problem is essentially solved from my end. Even if it's not exactly the way I intended.

% final try was to allocate sufficient space for the worst case struct growth
% (I have some that are 973 ... up to 11869)
%carray = cell(11869,4);
% first try for carray
carray = cell(1,4); % 'results' assigned to 1 of 4 cells

for k = 1:4
    file1 = some_data{k}; % grab from file
    for i = 1:1000
        % gather results in array of structures
        results(i).rvr_time = str2num(file1{i,1});
        results(i).chan     = str2num(file1{i,2});
               ...
    end
    % Assign array of structs to a cell in carray
    carray{:,i} = results;
    % I tried this first:
    carray{i} = results; 
    % and this...
    carray(i) = results;
end

What happens as I iterate through the four 'files' is that the cell array grows from 1x973 (first dataset) to 1x11869 (for the third), then the fourth (1x364) is lost... ending up with a carray of 1,11869 of garbage...

thanks again, Jerry

[Jerry Olup]

Thanks for moving the response Stephen. I haven't used this interface very much.

[Stephen23]

@Jerry: it is hard to see what the problem is. A structure can be allocated to a cell just like any other variable can be:

C{k} = S

This seems to satisfy your description. If you are actually trying to extract data from a structure then maybe struct2cell could help.

More importantly you should probably import that data correctly in the first place: it seems that you import data as strings, and then do some slow string-to-numeric conversion inside of MATLAB. Most likely you should just import the data correctly in the first place (as numeric data), and this may well help you to resolve the question of how it is stored as well. Beginners often import data as string because they do not use the importing tools very effectively. If you want help importing data, then you could upload a sample file by clicking the paperclip button.

What you are doing seems overly complex for the mere importing of data. It might help if you read this:

http://xyproblem.info/

[Jerry Olup]

Hi Stephen, Yep I know I can assign a struct to a cell. However I am trying to assign an array of structures to a single cell.

I worked around this by using a struct of struct of array of struct.

However then I can't simply index through the cell array to get access to my structures, which is my goal. Instead I hav to access each structure through dot indexing, which forces me to manage the structure names... still a lookup but not as convenient.

Like I said the problem is worked around... just not how I wanted to do it.

>> What you are doing seems overly complex for the mere importing of data The data is already imported from the csv, this is the act of arranging the data, reducing some of it, and organizing it. So my

      file1 = some_data{k}; % grab from file

isn't really grabbing from a file but an array that is already imported. This is complex data - each point uses a structure... multiple points comprising the array. Thanks, Jerry

[Jerry Olup]

Stephen, >> Beginners often import data as string because they do not use the importing tools very effectively.

Ah, regarding importing of data:

The data is mixed between numbers (all modes, literally) and strings. It is also variable length with fields missing or having a few modes (including numeric or string, causing decisions in interpretation). That's why I ran it all as strings (with a lot, hence the ..., of code omitted). from that I can parse variations of the file effectively.

This imports all data in a few mins, so I am satisfied with the performance.

It's not regular or academic data...!

The code excerpts are meant to point out that this is storing an array of structs into a single cell.

Thanks, Jerry

[Stephen23]

"However I am trying to assign an array of structures to a single cell."

I still do not see what the problem is. Any variable can be allocated to a cell, the fact your structure is non-scalar is irrelevant. Here is a non-scalar structure array being allocated to a cell:

>> S(1).data = 1;
>> S(2).data = 3;
>> C{1} = S;

That test code throws no errors, no problems doing exactly what you state: "assign an array of structures to a single cell."

"Instead I hav to access each structure through dot indexing"

You put the data into an structure yourself! How else would you expect to access it?

Answer 1

[Stephen23]

In your example code you need to change the indexing into the cell array from i to k:

carray{k} = results;

[Jerry Olup]

ugh, yes, thanks!

[Jerry Olup]

And... tested, so it's fine to store any data in cell as expected. Simple indexing error. Thx again.