Doug's MATLAB Video Tutorials

July 2nd, 2009

Advanced: loading files using their names for variable names

In this short video we will show how to bring text files into MATLAB and use the original file name as the variable name. Actually, the file name will be used as a field name in a structure.

Some people would use the EVAL statement or ASSIGNIN to do this, however doing so would disallow the JIT accelerator from speeding the code up as much as it could. The skills taught in this video are related to the ones in these videos:

Processing files in a directory with MATLAB

Dynamic field name usage in MATLAB

By Doug Hull

16:22 UTC | Posted in Format: Video, Level: Advanced | Permalink | 10 Comments »

You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

10 Responses to “Advanced: loading files using their names for variable names”

jason replied on July 2nd, 2009 at 19:31 UTC :

I was just thinking about this last week. Is there a comparable solution, using load instead of eval, that would also keep the Matlab compiler happy with scripts (.m files)?

For example, if I had a series of parameters as a script (a .m file), this won’t work with the compiler:

eval([filename '_model_parameters']);

Is there an equivalent with load that would work?

Thanks for the videos.
Jason.
jason replied on July 6th, 2009 at 19:58 UTC :

So, thanks to Doug’s help, it turns out that the EVAL in was not in fact the real problem. Rather, it was that the associated m-file was not compiled with the main function. In this case, I have hundreds of these associated m-files based on site × date combinations and didn’t want to have to recompile each time I had a new dataset with a new site × date combination. The solution is then to not use the script-based m-file (that required being compiled) and instead use a csv file.

See here for the details on compiling with associted scripts:

http://www.mathworks.com/access/helpdesk/help/toolbox/compiler/f13-1003481.html#f13-1003495

A basic example of the old script-based m-file is:
```
Latitude = 49.3824;
Longitude = -80.4862;
Altitude = 300;
```
As a csv file, this is:
Latitude,49.3824
Longitude,-80.4862
Altitude,300

To make those variables available and to replace the
```
eval([filename '_model_parameters']);
```
command, I have this:
```
% Get data from csv files
fid = fopen(filename '_model_parameters.csv');
inputs = textscan(fid,'%s%f','delimiter',',');
fclose(fid);

% Separate variables from values
vars = genvarname(inputs{1,1});
vals = inputs{1,2};

% Assign values to variables
for i = 1:size(vars,1)
        eval([cell2mat(vars(i)) '= vals(i);']);
end
```
Thanks (again) to Doug,
Jason.
dhull replied on July 6th, 2009 at 20:04 UTC :

Thanks for the follow up. Hopefully at least one other MATLAB user will find this useful!

-Doug
Carlos Adrian Vargas Aguilera replied on July 9th, 2009 at 21:50 UTC :

Nice post Doug.

I used to work with my ascii files as you explain. I even write a code called INT2STRZ
http://www.mathworks.com/matlabcentral/fileexchange/12973
to work around with files named data001.txt, data002.txt, etc.

Now I prefer to use another way. First I search for the files names and create a list with the LS command (in windows) which allows me to use the wildcard ‘*’ and then I just pick the file I need. So, your code will look like this:
```
list = ls('data*.txt'); % Files names on rows (WIN OS)
for k = 1:size(list,1)
 fileName = deblank(list(k,:)); % Name of k-file
 dataStruc.(fileName) = load(fileName);
end
```
Of course, there is an extra line, and one should be aware that all ‘data*.txt’ files will be read! Other safer way is to save the filenames on a file and read instead of use LS. Well, just a thought

Carlos
Mesut replied on February 7th, 2011 at 16:45 UTC :

Hi
quick question
can I pipe the names of files in a directory in to a variable array
meaning if I have bunch of text files *.asf
how can I (if possible) assign results of ls *.asf
into a variable that I can use (like a vector member by member)
thanks
dhull replied on February 9th, 2011 at 14:53 UTC :

@Mesut,

The Dir command should help.

Doug
Aaron replied on March 31st, 2011 at 19:20 UTC :

Thanks!!

spent hours trying to figure this one out and it turns out to be beautifully simple.

One very grateful newbie
Ramin replied on June 23rd, 2011 at 13:39 UTC :

I have same problem but my files are in “.CSV” format (Microsoft Office Excel Comma Separated Values File) and I use “csvread“to load one of my files, so I used:
```
fileName= [ 'data' num2str(i)];
datastruct.(fileName)=csvread([fileName'.csv'],0,0,[0,0,32766,0]);
```
But didn’t work.
I have also some “.txt” files that have some header line and I use “importdata“to load one of my files, so I used:
```
fileName= [ 'data' num2str(i)];
datastruct.(fileName)=importdata ([fileName '.txt'], ',', 18);
```
But this didn’t work, too.
dhull replied on June 29th, 2011 at 18:25 UTC :

@Ramin,

Please make sure you are constructing the name of the file correctly. For testing, just write out the file name. does it import then?

Doug
liz replied on November 11th, 2011 at 21:25 UTC :

Hi,

adding onto the discussion, I wanted to do the same thing as Jason, but instead of just doubles, I had arrays of numbers and strings that I wanted to assign to variables. A slight modification of Jason’s code is below which allows one to read in variables from file assign the variables in the matlab workspace.

my input looks something like this:

mtraj{1}=’src/ww1.assn.traj’;
mtraj{2}=’src/ww2.assn.traj’;
mtraj_to_use=[1, 2];
lag_i=[ 15 ];
lag=[1, 2, 4, 8, 10, 16, 25, 50, 100, 150, 200, 250 ,300, 400, 500, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 8000, 10000];
low_mem_map_matrix=0;
occ_threshold=0;
symm=1;
macrostate_defs=[2, 4, 6, 8, 10, 20, 40, 50, 100, 150, 200, 250, 300, 350, 400, 450, 500, 600, 700, 800];

it requires a semicolon separating all variable definitions as well as commas between array elements. Besides those two qualifications, it follows matlab syntax. I hope this is useful to other matlab users!
```
fid = fopen(input_file);
inputs = fscanf(fid,'%s');
fclose(fid);                                              

limits = strfind(inputs,';');
ndx = 1;
for(i = 1:length(limits))
  t = inputs(ndx:limits(i));
  t = strrep( t, ',' , ' ');
  eval(t);
  ndx = limits(i) + 1;
end
```