Strange symbol in file name - outputting file from Xcode and loading into Matlab

Hi all, I am trying to output a text file in C++ (using xcode on a Mac), and then upload the same data into Matlab.

However, the text file I output from xcode (which should be called '/Users/robinthompson/Documents/JTBWhenIsEpidemicOverPaper/⁨MCMCResults_18351555Time_212.txt') only loads into Matlab from the command window if I write

load('/Users/robinthompson/Documents/JTBWhenIsEpidemicOverPaper/⁨?MCMCResults_18351555Time_212.txt')

, where the question mark before the file name is instead a weird rectangle symbol...

Furthermore, because the weird rectangle symbol is not a standard character, the same command doesn't work to upload the file when I run the load command from a Matlab script...

This is driving me crazy - is there a solution? What is this strange rectangle? Thanks in advance!

[Walter Roberson]

For debugging,

cd('/Users/robinthompson/Documents/JTBWhenIsEpidemicOverPaper');
dinfo = dir('*Results_18351555Time_212.txt');
for K = 1 : length(dinfo)
    fprintf('Visible file name: "%s"\n', dinfo(K).name);
    fprintf('Numeric equivalent: ');
    disp(dinfo(K).name + 0);
    fprintf('\n');
end

and show us the result

[Robin Thompson]

Hi Walter, thanks for your help - please see attached screenshot!

Thanks, Robin

Answer 1

[Robin Thompson]

When I run this from a script (rather than command window), I get:

help

Visible file name: "⁨MCMCResults_18351555Time_212.txt"

Numeric equivalent: Columns 1 through 8

8296 77 67 77 67 82 101 115

Columns 9 through 16

117 108 116 115 95 49 56 51

Columns 17 through 24

53 49 53 53 53 84 105 109

Columns 25 through 32

101 95 50 49 50 46 116 120

Column 33

116

[Walter Roberson]

https://www.w3.org/International/questions/qa-bidi-unicode-controls

The 8296 is unicode U+2068, "First Strong Isolate", which is sort of explained in the above link.

It is not at all clear where that came from. It might or might not be related to the fact that U+2068 could be broken up into space and 'h' -- that is, if a section of code was confused over whether a character vector was "wide" characters or not, then plausibly it might have tried to store space and 'h' there but accidentally ended up with U+2068 . But there is no apparent reason why your code might have wanted to store space 'h' .

Answer 2

[Robin Thompson]

And the weird rectangle is in front of the filename!

Answer 3

[Robin Thompson]

Still no luck fixing this - if anyone has any ideas, I would be really grateful! Thanks! Robin

[Walter Roberson]

This is the point where we have to start looking at the code that created the file.

[Robin Thompson]

std::stringstream sswwww;
    sswwww << "⁨MCMCResults_" << (int)(seedVal) << "Time_" << (int)T  <<  ".txt" ;
    std::string swwww(sswwww.str());
       std::cout << swwww << std::endl;
    
       ofstream Ntime4;
       Ntime4.open(swwww.c_str());
    
    for (int i = 0; i < nStepsMCMC; i++) {
        Ntime4 << recordedValues[i][0] << " " << recordedValues[i][1] << " " << recordedValues[i][2] << " " << recordedValues[i][3] << " " << recordedValues[i][4] << endl;
    }
    
       Ntime4.close();

[Walter Roberson]

What is the line

std::string swwww(sswwww.str());

doing? Is it somehow constructor format for std::string ?

I notice you are dumping swwww to stdout: does the char(8296) show up in that?

[Robin Thompson]

Yes, that's right.

When I cout << swwww, then it comes up with the correct filename - i.e. it outputs MCMCResults_18351555Time_2000.txt

Very confusing!!!

[Walter Roberson]

Time to use the debugger to see what the c_str returns. If it is okay then it would be time to debug Ntime4.open

[Walter Roberson]

Also, when you did the cout, did you happen to pipe the output to "od" or write it to a file and check the contents of the file in detail? With it being a non-printing non-spacing character, there is the possibility that it is there but just did not display in that context.