The field width parameter of sprintf and fprintf (e.g., "n" in "sprintf('%nf',1234)") specifies the minimum width of the field. This means that shorter numbers will be padded but longer numbers will be allowed to extend beyond the specified field width when this is necessary to maintain the specified precision. I am looking for a way to specify the maximum field width such that width supersedes precision.
For example, I would want 1234.56, 1234.56789, and 1.2345 to have the width of 8 characters (including the decimal) and therefore print as 1234.560, 1234.567, and 1.234500, respectively.
Here is a similar question, but without a MATLAB-specific solution Specifying maximum printf field width for numbers (truncating if necessary)?
I feel like someone has to have encountered this problem before, but I was unable to locate anything relevant. Please provide a link if a similar question exists. Thanks
You can use floor(log10()) to determine how many non floating digits you have:
X = [1234.56, 1234.56789, 1.2345]
n = 8 % number of total digits.
% We create an array of string containing the format:
%[ '%.4f\n'
% '%.4f\n'
% '%.7f\n' ]
form = strcat('%.',num2str(n-1-floor(real(log10(X))).'),'f\n')
% We concatenate this array into one single string and we print the result:
fprintf(reshape(form.',1,[]),X)
Or you can use a for loop (using the same logic):
for x = X
form = strcat('%.',num2str(7-floor(real(log10(x)))),'f\n');
fprintf(form,x)
end
And we obtain:
1234.5600
1234.5679
1.2345000
Edit:
I use floor(log10()) instead of ceil(log10()) so the code won't fail if a number is part of the set 10^n (1,10,100...).
And now this snippet can deal with negative floating number since I only take the real part of the log10 output.