How can I run all executable files that are located in series of subfolders?

I have a hundreds of sub-folders (named in sequence like folder_1, folder_2 ...) that include executable file and all necessary input files in them. I just need to get in every sub-folders and run exe files, they don't even require typing in command window. So far, i got this code:

 currentFolder = pwd;
for n=1:100
cd(sprintf('D:\analysis\folder_%i',n))
system('runner.exe &');
cd(currentFolder);
end

But i think this code only runs the exe in main folder(in my case D:\analysis) repeatedly, not the ones in sub-folders. How can i do it?

Answer 1

[Walter Roberson]

projectdir = pwd;   %or hard-code the name

folderinfo = dir(projectdir);
folderinfo(~[folderinfo.isdir]) = [];  %discard the non-directories
folderinfo(ismember({folderinfo.name}, {'.', '..'})) = [];  %get rid of . and ..

for K = 1 : length(folderinfo)
  thisdir = fullfile( projectdir, folderinfo(K).name );
  cmd = sprintf('cd "%s" & runner.exe &');
  system(cmd);
end

[GeoMSI]

Thank you, it worked, but it needed to add 'thisdir' to sprintf.

[Walter Roberson]

Yes, you are right. The perils of composing code on the fly at 2:30 in the morning!

[GeoMSI]

I used the same structure to load the output files and calculate the standard dev. When i plot the results it seemed to me that for loop is not following a reqular order numerically for folders. I mean it didnt calculate the stdv. from folder1 to folder50, it somehow mixed the order. How can i fix that?

clc, clear;
projectdir = pwd;   %or hard-code the name
folderinfo = dir(projectdir);
folderinfo(~[folderinfo.isdir]) = [];  %discard the non-directories
folderinfo(ismember({folderinfo.name}, {'.', '..'})) = [];  %get rid of . and ..
for K = 1 : length(folderinfo)
thisdir = fullfile( projectdir, folderinfo(K).name );
A = load(sprintf('%s/OUTPUT1.DAT',thisdir));
B = load(sprintf('%s/OUTPUT2.dat',thisdir));
C = A(:,3)-B(:,3);
m(K)=std(C,1);
end
y=linspace(5,240,48);
plot(y,m);

[Walter Roberson]

http://www.mathworks.com/matlabcentral/fileexchange/47434-natural-order-filename-sort

Your folders are named numerically, so the sort order would typically be

1, 10 : 19, 2, 20:29, 3, 30:39, 4, 40:49, 5, 50

You can use the routine I linked to above to get the files in a numeric sorted order.

[GeoMSI]

I couldn't do it. natsortrows or natsortfiles requires string, but folderinfo is structure so it cannot do. When i convert structure to string in that case i cannot use that as directory in the for loop:(

[Image Analyst]

Why does it need to be in that order?

[GeoMSI]

well, when i try to plot the standard deviation with y=linspace(5,240,48); , which is the pattern of increase in folder name, figure is wrong cause it didn't plot in right order. For instance, in figure y=150(folder_150) corresponds to different folder's standard deviation, not 150 itself.

[Image Analyst]

OK, so they don't need to be in order. You just need to figure how to use sscanf() or something to parse the number out of the folder name. Do that.

[Walter Roberson]

projectdir = pwd;   %or hard-code the name
folderinfo = dir(projectdir);
folderinfo(~[folderinfo.isdir]) = [];  %discard the non-directories
folderinfo(ismember({folderinfo.name}, {'.', '..'})) = [];  %get rid of . and ..
foldernames = natsort({folderinfo.name});
for K = 1 : length(foldernames)
  thisdir = fullfile( projectdir, foldernames{K} );
  A = load(sprintf('%s/OUTPUT1.DAT',thisdir));
  B = load(sprintf('%s/OUTPUT2.dat',thisdir));
  C = A(:,3)-B(:,3);
  m(K)=std(C,1);
end
y = linspace(5,240,48);
plot(y,m);

[GeoMSI]

Thank you again Walter, it worked.

[Stephen23]

I recommend using natsortfiles for sorting filenames with file extensions, rather than natsort. The file extension and separator character (period) can affect the sort order when using natsort. I created natsortfiles to remove this effect, so shorter filenames always sort first (like a dictionary, and like the most popular desktop OS).

[Wesser]

Hi Walter,

Same question many moons later...

I don't quite understand the code in your original reply to this question. Right now, I believe I'm only runing the .exe in the original directory file, but I'd like to run the copied .exe in each of the created path{i}. The .exe doesn't cooporate when it's not in the same directory as all the required files. Thank you for any help you may be able to provide with this!!

clear all;
clc;
%Solute transport parameters in row 47 of SELECTOR.IN file
SolTrans=[0.035 0 1 0.0479 0 0 0 0 273895 1 0 0 0 0]; %[Kd Nu Beta Henry SnkL1 SnkS1 SnkG1 SnkL1p SnkS1p SnkG1p SnkL10 SnkS10 SnkG10 Alfa] !! UPDATE NUMBERS IN BRACKETS !!

%~~~~~~
%ADD NOISE TO SOLUTE TRANSPORT PARAMETERS 
num_sim=100;    %100 Monte Carlo Simulations
Kd=SolTrans(1)+.02*rand(1,num_sim); %a vector with 100 values of Kd parameters. We assume that the error is normally distributed with a standard deviation of ##.
Nu=SolTrans(2)+.02*rand(1,num_sim);
Beta=SolTrans(3)+.02*rand(1,num_sim);      %!!! CHANGE .02 FOR ALL THESE TO APPROPRIATE STANDARD DEVIATIONSN !!!
Henry=SolTrans(4)+.02*rand(1,num_sim);
SnkL1=SolTrans(5); 
SnkS1=SolTrans(6);
SnkG1=SolTrans(7);
SnkL1p=SolTrans(8);
SnkS1p=SolTrans(9)+.02*rand(1,num_sim);      %!!! FOR AWI MODIFIED MODEL - KL - LANGMUIR EXPONENT FOR AWI FOR LANGMUIR SORPTION or 0 FOR FREUNDLICH SORPTION
SnkG1p=SolTrans(10);
SnkL10=SolTrans(11);
SnkS10=SolTrans(12);
SnkG10=SolTrans(13);
Alfa=SolTrans(14);



%~~~~~~
%INITIALIZE FILES AND DIRECTORIES

hydrus_exec='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\H1D_CALC.exe';   %'C:\Program Files (x86)\PC-Progress\Hydrus-1D 4\H1D_calc.exe';   
hydrus_ref='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS';   
profileDAT='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\PROFILE.DAT';    
options='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\OPTIONS.IN';
atmosph='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\ATMOSPH.IN';
selectorIN='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\SELECTOR.IN';  
level='C:\Users\jessi\Desktop\Hydrus\Projects\ATSDR\AFFF1PFOS\LEVEL_01.dir';
work_dir='C:\Users\jessi\Desktop\Simulations\';
mkdir(work_dir);


%~~~~~~
%CREATE DIRECTORIES AND INPUT FILES - This block is doing the work. Creates the input files for all realizations. Copies the mandatory file Profile.dat to 
% the corresponding folder and creates the corresponding Selector
% .in. The Selector.in file is different for each realization since the solute transport parameters is 
% variable. The code reads the reference Selector.in, copies the first 46 lines to the new file, write the van Genuchten parameters to the 47th line, 
% and finally copies the last 10 lines from the reference Selector.in to the new file.

path=cell(1,num_sim);
for i=1:num_sim
    path{i}=strcat(work_dir,'run_',num2str(i)); %create folder for each run           
    mkdir(path{i});    

    copyfile(profileDAT,path{i}); %copy profileDAt from reference directory to the simulation directory 
    copyfile(options,path{i}); %copy options from reference directory to the simulation directory 
    copyfile(atmosph,path{i}); %copy atmosph from reference directory to the simulation director
    copyfile(hydrus_exec,path{i}); %copy options from reference directory to the simulation directory 
    
    %create a new level_01.dir in each subdirectory with associated file
    %directory path
    fid_level = fopen(fullfile(path{i},'LEVEL_01.dir'),'w');
    fprintf(fid_level,'%s',path{i});
    fclose(fid_level);

    
    fileID_out=fopen(strcat(path{i},'\selector.in'),'wt'); %manipulate Selector.in for each run
                        % "wt"= permision for file axis type
                        % w=Open or create new file for writing. Discard existing contents, if any.
                       % To open files in text mode, attach the letter 't' to the permission argument,
    fileID_in=fopen(selectorIN);
    skip_lines=46;      %!!! THIS IS THE LINE THAT CHANGES IN SELECTOR.IN FILE , i.e. solute transport parameters!!!
    for k=1:(skip_lines)
        x=fgetl(fileID_in); %x = fgetl(fileID)= returns the next line of the specified file, removing the newline characters.
        fprintf(fileID_out,'%s\n',x);  % %s in the formatSpec input indicates that the values of the variables url and sitename, should be printed as text. 
                                       % '\n' as a newline indicator.   
                                       % "x" = prints the values from variable x
    end
    
    out_Sol=SolTrans; % Renaming...
    
    %Whatever the solute transport parameter is for that monte carlo run (i)
    out_Sol(1)=Kd(i);       % Solid phase sorption Kd
    out_Sol(2)=Nu(i);       % Van Genuchten parameter
    out_Sol(3)=Beta(i);     % Van Genuchten parameter
    out_Sol(4)=Henry(i);     % Kh = KL*Rmax
    out_Sol(9)=SnkS1p(i);    % KL - LANGMUIR EXPONENT FOR AWI FOR LANGMUIR SORPTION or 0 FOR FREUNDLICH SORPTION
  
    fprintf(fileID_out,'%f %f %f %f %f %f %f %f %f %f %f %f %f %f\n',out_Sol'); 
                             % %f = Fixed-point notation (Use a precision operator to specify the number of digits after the decimal point.)
                             % '\n' as a newline indicator. 
    fgetl(fileID_in);        % returns the next line of the specified file, removing the newline characters.
    skip_lines_end=5;       % CORRECT NUMBER???  DOUBLE CHECK
    for k=1:(skip_lines_end)
        x=fgetl(fileID_in);
        fprintf(fileID_out,'%s\n',x);
    end
    fclose('all');
end

%   THE SECTOIN THATIS WONKY AND NEEDS HELP!
path=cell(1,num_sim);  
for i=1:num_sim
    exec_path=['"' hydrus_exec '" "' path{i} '"'];
    [x, y]= dos(exec_path);
    if x % unsuccessful
         error('exe failed'); % or take some other action besides throwing an error
    end
end

Answer 2

[Image Analyst]

If your folders are more than one level deep, you can use genpath(), like in my attached example.