MATLAB: Does textread fails to read files which contain non-ASCII characters in their filename on MAC operating system

MATLABnon-asciitextreadtextscan

I tried running following simple function which creates a file which contains non-ASCII character in its filename. I wrote some text in it and tried to read it back using "textread" but while doing so I received error saying file was not found.

I tried following code:

filename = char([65 181 66]);
fid = fopen(filename, 'w'); fprintf(fid, '%s\n', 'hello'); fclose(fid);
textread(filename, '%s')

I received following error after executing code:

Error using dataread

File not found or permission denied.

Error in textread (line 168)

[varargout{1:nlhs}]=dataread('file',varargin{:}); %#ok<REMFF1>

Why does textread fail while working with files which contain non-ASCII characters in their filename? Is there any workaround I can use to read these files?

Best Answer

As mentioned in the documentation of 'textread' for MATLAB, it is recommended to avoid using 'textread' by using 'textscan' instead. The issue seen can be resolved by using 'textscan' instead of 'textread'. Please refer to the attached demo script. The only difference between the scripts is, the attached script uses 'textscan' to read from the file instead of 'textread'.

Related Solutions

MATLAB: Different Results When Using “textscan” vs “fgetl” in MAC

When a file is opened in vim and Sublime Text on Mac, there is no extra header line. However, the MATLAB Editor does display an extra header line. This is probably what is throwing off the detection code. 'fgetl' and 'textscan' are interpreting the number of header lines differently.

Two workarounds exist:

1. You can use the following call for "textscan" function:

>> dataset = textscan(fid, '%f %f', 'HeaderLines', skipLines, 'CollectOutput', 1, 'EndOfLine', '\r\n');

2. Use "readmatrix" function instead:

dataset = readmatrix(filename);
x = dataset (:,1); 
y = dataset (:,2);

MATLAB: How to replace ‘textread’ with ‘textscan’

The 'textscan' function offers several advantages over 'textread'.

1. It is much more efficient, especially when reading very large files or a very large number of files

2. It has more features and supports more types and formats

3. It offers greater flexibility when reading from arbitrary points in the file.

4. Using fopen/fclose offers more power and safety in dealing with files and errors.

Your original code might look something like this:

[x,y,z] = textread(filename,format);

When using 'textscan', you must open the file with 'fopen' first and close it later. In addition, the output is returned as a cell array instead of separate output variables.

Your new code might look something like this:

fid = fopen(filename);
C = textscan(fid,format);
fclose(fid);
[x,y,z] = C{:};

There is likely no need to go back and rewrite all of your old code to use 'textscan' instead, but future code should. However, you may still want to do this if you desire the features, stability, and performance of 'textscan' in your older code. For example if you are reading very large files or a very large number of files you may find that the significant speedup is worth the effort in some cases.

Best Answer

Related Solutions

MATLAB: Different Results When Using “textscan” vs “fgetl” in MAC

MATLAB: How to replace ‘textread’ with ‘textscan’

Related Question