How to use textscan on a cell array without a loop?

Hi everyone!

So, I have a structure with multiple fields. One of the fields I filled with strings from multiple, fairly large text files (~100,000 lines), and it is now a cell array. Here's the layout of the structure, and a sample of mystruct.out.

mystruct = 

  1×5 struct array with fields:

    filename
    out
    size
    
mystruct.out = 
{'%ACCDA,50,123.99,W,07512.001,E,2,11,2.2,1002.2,Z,,,,*22' }
{'%ACCDA,50,123.99,W,04412.001,E,2,11,2.2,1002.2,Z,,,,*20' }
{'%ACCDA,50,123.99,W,07112.001,E,2,11,2.2,1002.2,Z,,,,*2A' }
{'%ACCDA,50,123.99,W,08512.001,E,2,11,2.2,1002.2,Z,,,,*2E' }
{'%ACCDA,50,123.99,W,06512.001,E,2,11,2.2,1002.2,Z,,,,*2B' }
...
...

I want to avoid using a loop to perform the following, but I haven't been able to figure out how to do so on a mass scale without a loop.

I want to be able to use textscan, or some other method, to store the delimited strings into a new cell array, inside of an new mystruct field (like *** below)

Note: 'ii' is an index I'm using in a separate loop, but it's irrelevant to this issue.

for nlr = 1:mystruct(ii).size
    fields = textscan(mystruct(ii).out{nlr},'%s','delimiter',',');
end

***(e.g. store into mystruct(ii).split(nlr))

How can I use textscan, or some other method, to "vectorize" the above loop?

Thanks!

[Walter Roberson]

I am a little confused about the sizes involved. Your show a 1x5 struct array, and 5 example output lines, but imply that there are more. Is it that case that mystruct(K).out is a 1 x 1 cell array containing a character vector for any given K? Or is mystruct(K).out an N x 1 cell array containing character vectors for any given K? If mystruct(K).out is an N x 1 cell array of character vectors, then should the data for each be processed separately, or can it all be put together into one big array ?

[rbme17]

Hi Walter, sorry for the confusion.

The size of the structure varies, but in this case it could be expanded like so:

mystruct(1) =
  filename: some value
  out:      Nx1 array
  size:     some value

mystruct(2) = ... etc.

mystruct(K).out is an Nx1 cell array

I'm hoping to process each line of mystruct(K).out, from 1:N, and store it in the new structure field but without using a loop. Does that make more sense, or should I rephrase?

Thanks

Answer 1

[rbme17]

Ok, so I actually ended up figuring something out that doesn't require textscan. I used the function 'split'.

Here's a shortened sample:

mystruct.out = {
        '%ACCDA,50,123.99,W,07512.001,E,2,11,2.2,1002.2,Z,,,,*22' 
        '%ACCDA,50,123.99,W,04412.001,E,2,11,2.2,1002.2,Z,,,,*20' };

s = split(mystruct(ii).out,',',1); 

s = 
    {'%ACCDA'   }    {'%ACCDA'   } 
    {'50'       }    {'50'       }    
    {'123.99'   }    {'123.99'   }    
    {'W'        }    {'W'        }    
    {'07512.001'}    {'04412.001'}   
    {'E'        }    {'E'        }   
    {'2'        }    {'2'        }    
    {'11'       }    {'11'       }  
    {'2.2'      }    {'2.2'      }   
    {'1002.2'   }    {'1002.2'   }    
    {'Z'        }    {'Z'        }   
    {0×0 char   }    {0×0 char   }    
    {0×0 char   }    {0×0 char   }    
    {0×0 char   }    {0×0 char   }    
    {'*22'      }    {'*20'      }   

Sorry if I wasn't clear enough in what I was asking, I was able to accomplish what I needed too so thank you all for your help!

Answer 2

[Walter Roberson]

See my answer at https://www.mathworks.com/matlabcentral/answers/476452-splitting-one-column-into-multiple-columns# on how to process more efficiently.

[rbme17]

Hmm this doesnt seem to be what I'm looking for. I don't need to join strings together, they need to be separated in order to be processed in another script.

[Walter Roberson]

textscan can process a character vector that has embedded newline characters, treating each as a separate line. The processing is quite efficient. This permits you to parse and convert each column of your input without having to loop sscanf or textscan over the cell entries. Give it a try.

[Walter Roberson]

mystruct.out = [{'%ACCDA,50,123.99,W,07512.001,E,2,11,2.2,1002.2,Z,,,,*22' }
{'%ACCDA,50,123.99,W,04412.001,E,2,11,2.2,1002.2,Z,,,,*20' }
{'%ACCDA,50,123.99,W,07112.001,E,2,11,2.2,1002.2,Z,,,,*2A' }
{'%ACCDA,50,123.99,W,08512.001,E,2,11,2.2,1002.2,Z,,,,*2E' }
{'%ACCDA,50,123.99,W,06512.001,E,2,11,2.2,1002.2,Z,,,,*2B' }];
S = strjoin(mystruct.out, '\n');
fields_cell = textscan(S, '%s%f%f%s%f%s%f%f%f%f%s%s%s%s%s', 'Delimiter', ',');
fields_cell{5}

ans =

                  7512.001
                  4412.001
                  7112.001
                  8512.001
                  6512.001

Answer 3

[per isakson]

Am I on the right track? Run

%%    
    mystruct.out = {
        '%ACCDA,50,123.99,W,07512.001,E,2,11,2.2,1002.2,Z,,,,*22' 
        '%ACCDA,50,123.99,W,04412.001,E,2,11,2.2,1002.2,Z,,,,*20' 
        '%ACCDA,50,123.99,W,07112.001,E,2,11,2.2,1002.2,Z,,,,*2A' 
        '%ACCDA,50,123.99,W,08512.001,E,2,11,2.2,1002.2,Z,,,,*2E' 
        '%ACCDA,50,123.99,W,06512.001,E,2,11,2.2,1002.2,Z,,,,*2B' };
%%
    chr = strjoin( mystruct.out, '\n' );
    cac = textscan( chr, repmat('%s',1,15), 'Delimiter',',' );

Inspect the result

>> cac{1}
ans =
  5×1 cell array
    {'%ACCDA'}
    {'%ACCDA'}
    {'%ACCDA'}
    {'%ACCDA'}
    {'%ACCDA'}
>> cac{3}
ans =
  5×1 cell array
    {'123.99'}
    {'123.99'}
    {'123.99'}
    {'123.99'}
    {'123.99'}
>> cac{15}
ans =
  5×1 cell array
    {'*22'}
    {'*20'}
    {'*2A'}
    {'*2E'}
    {'*2B'}
>> 

Don't you want to convert the numeric fields to double?

[rbme17]

This was very close to what I needed. I ended up figuring out another way that worked well for me. Thanks!