reading hdf5 file with dataset name containing special characters

I have a hdf5 file, hdf5disp give me the following

Group '/'

..........

Group '/histogram_plots'

Group '/histogram_plots/last_plot'

......

Dataset '23: Total Intensity = |Eσ|² + |Eπ|²'

Size: 400

MaxSize: 400

Datatype: H5T_IEEE_F64LE (double)

ChunkSize: []

Filters: none

FillValue: 0.000000

note the Dataset '23: Total Intensity = |Eσ|² + |Eπ|²', obviously the following line works in matlab console but not in a script. What can be the solution?

y= h5read(fname,'/histogram_plots/last_plot/23: Total Intensity = |Eσ|² + |Eπ|²');

Answer 1

[Walter Roberson]

If the command works in a console but not in a script, and you are using an old enough version of MATLAB (I see you are using R2018b which is a bit older), then potentially your MATLAB is not saving .m files as UTF-8 and so might be getting incorrect characters.

If so, then there are a couple of approaches you can take:

• you could upgrade to R2020a or newer, which changes the default encoding to UTF-8 for all operating systems
• you could use one of the methods for convincing your older MATLAB to encode the file in UTF-8 . If I recall correctly, that is most difficult on Windows
• you could construct the character vector without using the special characters in the code
• you could create a uint8 stream of the utf8 bytes that encode the characters, and use native2unicode to convert into character vector. This is a quite effective way, but it is not readable

  sigma_char = char(hex2dec('03C3'));  %https://www.compart.com/en/unicode/U+03C3
  pi_char = char(hex2dec('03C0')); %https://www.compart.com/en/unicode/U+03C0
  raised_2_char = char(hex2dec('00B2')); %https://www.compart.com/en/unicode/U+00B2
  
  Dataset = sprintf('/histogram_plots/last_plot/23: Total Intensity = |E%s|%s + |E%s|%s', sigma_char, raised_2_char, pi_char, raised_2_char);
  
  y= h5read(fname, Dataset);
  

[Yuelin Li]

A straight upgrade to R2022B solved the problem. Thanks!