unicode characters in .m file

Matlab claims to be Unicode compliant, but only characters whose code # is ≤255 can be saved in .m files. I can accept that Matlab code words must be in ASCII, but there should be no such constraints on characters in comments or text strings. It would be nice if useful characters, like "≈", "Ω", "≤","≥", etc., could be saved in comments. On a similar note, while a character whose code is ≥128 and ≤255 can be stored in .m files, it is not possible to include them directly in graphics labels or text; typing '\circ' (1970s LaTex convention) seems clunky compared to "Alt+0176" (or pasted from another document) for the degrees symbol(°).

[Paul McKenzie]

While there now seem to be solutions for storing Unicode characters in .m files, I would still like a better alternative to the clunky LaTex method for special characters on plots. With the advent of Unicode, I don't see why labels, titles, and texts shouldn't be specified directly in Unicode. That would make the .m file text look like the text on the plot. (I think they call that approach 'WYSIWYG'.) Adding an 'Insert Symbol' feature to the editor would make labels, titles, and texts so much more convenient then remembering arcane \<symbol> sequences.

[Walter Roberson]

unicode characters can be included directly for text() objects when using Interpreter Tex (the default) or Interpreter None . However, at this time unicode characters cannot be included directly for latex.

Answer 1

[Rahul Arora]

For R2017a, the steps are similar to the ones described in https://www.mathworks.com/matlabcentral/answers/280988-how-do-i-get-my-matlab-editor-to-read-utf-8-characters-utf-8-characters-in-blank-squares-in-editors. The only difference is that the matlabroot/bin/lcdata.xml file no longer describes all standard locales and encodings. But you can still use this file for customizing the locale database.

For my case, the encoding is windows-1252 (you can find your encoding using the feature('locale') command), and therefore, I added the following lines to the <lcdata> ... </lcdata> block in lcdata.xml.

<codeset>
  <encoding name="UTF-8">
    <encoding_alias name="windows-1252" />
  </encoding>
</codeset>

Just modify based on your own system encoding.

[Paul McKenzie]

OK, this fix seems to work. Why isn't this version of lcdata.xml the Mathworks default? I can't see any downside, except for those (Luddites) who actually want Windows-1252 (ANSI) instead of Unicode

[Tom DeLonge]

being a bit late here: can I change this property for each user? Unfortunately I do not have access to lcdata.xml, I am not admin on my machine... Also lcdata.xml does not have any

<codeset>..</codeset>

entries on my machine, just a few commented lines

[Andrey Kazak]

The fix does not work on 2018b for Windows.

Please suggest a workaround.

Thank you.

[Rahul Arora]

Hi Andrey

I just tried doing this on my MATLAB installation (R2019b, Windows 10) and it works fine. Please note that you have to restart MATLAB after performing the edit I suggested.

Answer 2

[MathWorks Support Team]

As of R2020a, the MATLAB Editor supports UTF-8 characters, and uses UTF-8 as the default encoding for new plain text files, including MATLAB code files with a .m extension.

Answer 3

[Walter Roberson]

The instructions for OS-X (Mac) are given at in the Release Notes https://www.mathworks.com/help/matlab/release-notes.html?rntext=internationalization&startrelease=R2012a&endrelease=R2017a&groupby=release&sortby=descending&searchHighlight=internationalization

[Stephane]

This one worked for me (MacOS Mojave, R2018b) - Thanks !

Answer 4

[Jan]

Please read https://www.mathworks.com/matlabcentral/answers/262114-why-i-can-not-read-comments-in-chinese-in-my-mfile . It is possible to display Unicode in the editor.

[Paul McKenzie]

Answer does not seem applicable to R2017a.

[Paul McKenzie]

The answer from Jinghao Lei ('UTF- encoding')seems to have been successful on some earlier (Chinese?) version of Matlab, but again seems not applicable to R2017a.

[Yueyan Li]

The answer from Jinghao Lei https://cn.mathworks.com/matlabcentral/answers/280988-how-do-i-get-my-matlab-editor-to-read-utf-8-characters-utf-8-characters-in-blank-squares-in-editors does work on MATLAB R2017a, just rename the lcdata_utf8.xml to lcdata.xml first.

[Paul McKenzie]

An elegantly simple solution! The file names even make sense now. I wish that Mathworks had provided an option during installation to make this file name change, since many corporate users aren't allowed to make changes inside the "Program Files" directory.

[Paul McKenzie]

What happened? This solution was working a week ago; today, unicode characters are displaying as garbage. Did some Windows update foul Matlab up? Why is only Matlab susceptible?

[Walter Roberson]

Paul McKenzie: did you happen to do Windows Creator Spring Update within the last week?

[Paul McKenzie]

I don't think so; I'm running Win 7. I've seen this problem on two Win 7 machines.

[Duijnhouwer]

explanation in https://www.mathworks.com/matlabcentral/answers/262114-why-i-can-not-read-comments-in-chinese-in-my-mfile . does not apply to 2018a.

[Paul McKenzie]

After multiple conversations with tech support, it is clear that the editor design team did not intend to make it Unicode compatible. R2018b will not cure the problem. Perhaps some future release will be truly Unicode compatible, but in the interim, Notepad++ seems the only alternative that supports Unicode.

[Walter Roberson]

Mathworks needs to do more to make it easy to use Unicode in source code, UTF-8 at least.

[Paul McKenzie]

R2019a is out, but the editor still cannot handle Unicode!

Note to Mathworks Editor Design Team: Check out the comprehensive and elegant way Octave allows one to select text encoding.

[Paul McKenzie]

Now R2019b is out, and the editor still cannot handle unicode. Since that's the case, why does Matlab still include the file "lcdata_utf8.xml", since it no longer provides any unicode capability?

[Paul McKenzie]

R2020a is out and solves the problem! All .m files are saved in UTF-8 format.

[Shaul Shvimmer]

I'm using MATLAB 2020a and I still have the problem - I cannot change encoding to UTF-8 using Windows 10.

Answer 5

[Muhammad Kashif]

Assalam O Alikum and good day all

For those using languages having Arabic script e.g. Arabic ,Urdu.

1)Change the system Locale to Urdu(Pakistan) for Urdu.For Arabic, change the system locale accordingly.

2)For displaying Urdu and Arabic in Matlab Command prompt,use the following command:

>>slCharacterEncoding('UTF-8')

3) In the matlab editor, to make the string functions like strsplit() work properly for Urdu and Arabic, this should be the first line of code:

feature('DefaultCharacterSet', 'UTF8');

Best regards

[Walter Roberson]

Unfortunately, slCharacterEncoding() is for Simulink use, not the MATLAB command prompt.

Answer 6

[Jean-Michel Rousseau]

Hello,

Is ther any way to definitively change the 'DefaultCharacterSet' without the need to use the feature function at every start of Matlab?

I need to use 'windows-1252' and Matlab (& Simulink) always start with "UTF-8". It as recently canged after updating to the R2019b. As I can't solve the problem into R2019b (my folders and files with accent can't be used anymore without renaming them), i go back to a fresh R2016b but the problem remain :-(