Converting cell to array data with specific string

Hi,

I am trying to convert data read from a file to a numeric array. The format that it comes in as from an importdata command is:

{'[2, 11, 5, 0] '           }
{'[1, 10, 6, 0] '           }
{'[9, 3, 6, 0] '            }
{'[8, 1, 4, 0] '            }
{'[2, 4, 7, 0] '            }
{'[1, 3, 6, 0] '            }
{'[2, 3, 5, 0] '            }
{'[1, 2, 4, 0] '            }
{'[1, 2, 3, 0] '            }

I want to convert this to a double array of size n by 4. I want to do so without loops, preferably in one line (combined with the importdata command) to keep my code relatively concise. I have tried string, but that spits out an odd answer.

The data file includes those brackets, so I guess if anyone knows how to exclude MATLAB reading those, readmatrix can work (right now, the first and last columns are NaN because of those if I use that command).

Thank you!

[Cris LaPierre]

Please attach your file (or a representative example) to your post using the paperclip icon.

[Publius]

Attached. It's originally a .dat file, but I couldn't upload it, so I changed it to a .txt. Should be the same.

Answer 1

[Stephen23]

The simplest approach:

M = readmatrix('test.txt', 'TrimNonNumeric',true)

M = 5073×4

648 698 4494 623 4180 4181 1406 1415 4179 4180 1422 1415 4178 4179 1429 1422 4177 4178 1429 1434 4177 1434 4123 1438 4134 1441 4123 1438 1441 1443 4134 4145 4176 4145 1443 1444 1456 4185 1371 1358

Answer 2

[Cris LaPierre]

There are many ways to do this. Here are two. The result of the 2nd option is a table. See this page on Accessing Data in Tables

file = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/1195758/test.txt';
data1 = readmatrix(file,'delimiter',["[","]",","," "],'ConsecutiveDelimitersRule','join',...
    'LeadingDelimitersRule','ignore')

data1 = 5073×4

648 698 4494 623 4180 4181 1406 1415 4179 4180 1422 1415 4178 4179 1429 1422 4177 4178 1429 1434 4177 1434 4123 1438 4134 1441 4123 1438 1441 1443 4134 4145 4176 4145 1443 1444 1456 4185 1371 1358

data2 = readtable(file,'format','%*c %f %f %f %f %*c')

data2 = 5073×4 table
    Var1    Var2    Var3    Var4
    ____    ____    ____    ____

     648     698    4494     623
    4180    4181    1406    1415
    4179    4180    1422    1415
    4178    4179    1429    1422
    4177    4178    1429    1434
    4177    1434    4123    1438
    4134    1441    4123    1438
    1441    1443    4134    4145
    4176    4145    1443    1444
    1456    4185    1371    1358
    1384    4185    1371    4183
    1384    1395    4182    4183
    1406    1395    4181    4182
    1557    1555    4515       0
    4474    1557    1551       0
    4474    4515    1557       0

[Publius]

This first option was exactly what I was looking for -- thank you!

[Cris LaPierre]

In that case, Stephen23's answer is better in my opinion

[Publius]

Agreed, thank you for pointing that out. It came in afterward.

Answer 3

[Jan]

data  = {'[2, 11, 5, 0] '; '[1, 10, 6, 0] '; '[9, 3, 6, 0] '};
value = str2num(sprintf('%s;', data{:}))

value = 3×4

2 11 5 0 1 10 6 0 9 3 6 0

Internally str2num is based on eval. Maybe this is mor secure:

data2 = erase(string(data), "[");
data2 = replace(data2, "]" | ",", " ");
s     = strjoin(data2);
value = reshape(sscanf(s, '%g'), 4, []).'

value = 3×4

2 11 5 0 1 10 6 0 9 3 6 0

[Steven Lord]

If you're using release R2022a or later you can specify Evaluation='restricted' to limit what expressions in the text to be converted are executed.

str2num('[1, 2; 3, 4; 5:6]', Evaluation='restricted')

ans = 3×2

1 2 3 4 5 6

To preemptively answer your next question no, I don't have a definitive list of what counts as "basic math expressions" for purposes of restricted evaluation. I suspect if you stick to the operators in the Arithmetic Operators, Relational Operators, and Logical Operators sections on this documentation page along with array creation characters like square brackets, comma, semicolon, and colon you're probably okay.