Cannot Use MATLAB Coder to Generate C

I would like to generate c code from my function.

It's a function about reading data from .csv file generated by a sensor.

Below is my function code:

function [data_file, Sample_Rate, head, tail] = read(data_source_name)
data_file = readmatrix(data_source_name);

filen2 = char(data_source_name);
filen  = [filen2];                                      
files1 = [filen '.csv'];                               

textstring = textread(files1,'%s',16);                  
naxsenid   = textstring(2);                             
recordd    = textstring(6);                             
recordt    = textstring(7);                             
comnm      = textstring(8);                             
comns      = comnm{1,1};                                
comma      = strtok(comns,',');                         
if ( comma == "")                                       
    sri = 16;       
else                                                    
    sri = 15;
end
srf        = split(textstring(sri),'-');                
sdata      = split(textstring(sri-4),',');              
srt        = textscan(srf{1,1},'%f%c');                 
datans     = textscan(sdata{1,1},'%f');                 
if (srt{1,2} == 'K')                                   
    Sample_Rate = srt{1,1}*1000;
else
    Sample_Rate = srt{1,1};
end
datanc=datans{1,1};                                     

fprintf('naxsen_id   = %s\n',naxsenid{1,1});
fprintf('record_date = %s\n',recordd{1,1});
fprintf('record_time = %s\n',recordt{1,1});
fprintf('total_data_number = %d\n',datans{1,1});
fprintf('Sample_Rate = %f\n',Sample_Rate);

start = (1/Sample_Rate)*Sample_Rate-1+5;                 
stop = start+(datanc-1);                                 
stop_time = (stop-5+1)/Sample_Rate;                       
fprintf('%f\n',stop_time);
datan = 1:(stop-start+1);                                 
t = datan/Sample_Rate;                                    
Array = csvread(files1,start,0,[start,0,stop,5]);         
f = Sample_Rate*((-datanc/2):(datanc/2)-1)/datanc;      

data_length = size(Array);                                                    
classn = data_length(1,2);                             

head = 1;
tail = data_length(1,1);

end

and test code be like

data_source_name = "test";
[data_file, Sample_Rate, head, tail] = read(data_source_name);

the content of "test" is

### NAXSEN800387 Data0
### Time 2021/09/08  16:26     
### Data Length 9
### Sample Rate 100-Hz
accX,accY,accZ,gyroX,gyroY,gyroZ
1.01611328125,0.00927734375,0.0244140625,-0.3662109375,0.732421875,0.3662109375
1.0166015625,0.00830078125,0.0224609375,-0.30517578125,0.732421875,0.48828125
1.017578125,0.01025390625,0.02490234375,-0.3662109375,0.732421875,0.48828125
1.01513671875,0.01171875,0.02294921875,-0.30517578125,0.79345703125,0.48828125
1.0146484375,0.01025390625,0.025390625,-0.3662109375,0.79345703125,0.48828125
1.01513671875,0.0087890625,0.0244140625,-0.3662109375,0.79345703125,0.48828125
1.0185546875,0.00732421875,0.0224609375,-0.30517578125,0.732421875,0.3662109375
1.01611328125,0.00927734375,0.0244140625,-0.30517578125,0.732421875,0.42724609375
1.01513671875,0.0087890625,0.02587890625,-0.3662109375,0.732421875,0.42724609375

Output of test code

naxsen_id   = NAXSEN800387
record_date = 2021/09/08
record_time = 16:26
total_data_number = 9
Sample_Rate = 100.000000
0.090000

the last row "0.090000" means execution time, due to 9 data divided by sample rate 100 Hz = 0.09

When I executed MATLAB Coder, I got some error information below

I have no idea how to instead these function like readmatrix, split, etc.

Is there any way to resolve this issue?

If so, could you provide clear instructions/examples using my code?

Thanks a lot!

Answer 1

[Nathan S]

I think your best option is to use a combination of fscanf and fgetl, both of which are supported for code generation.

The best way to write such code will depend on what assumptions you can make about your file (which strings have the same length all the time, do you always have the same number of columns, are there likely to be any extra lines or extreneous data, etc.)

I have made some assumptions about what you might want, and this works on your example file in codegen for me. It will probably not be as robust to files that are slightly wrong as readmatrix would be.

function [data_file, Sample_Rate, head, tail] = readmyfile(data_source_name)

fid = fopen(data_source_name);

naxsenid = simpleSplit(fgetl(fid), 2);

[record_date, record_time] = simpleSplit(fgetl(fid), 3:4);

total_data_number = fscanfForScalar(fid, '### Data Length %d');

fgetl(fid); %skip any remaining whitespace / end of line

Sample_Rate = fscanfForScalar(fid, '### Sample Rate %f');

sample_modifier = fscanf(fid, '%c', 1);

fgetl(fid); %skip any remaining whitespace / end of line

if sample_modifier == 'K'

Sample_Rate = Sample_Rate * 1000;

end

headerLine = fgetl(fid);

numCols = sum(headerLine==',')+1;

fprintf('naxsen_id = %s\n',naxsenid);

fprintf('record_date = %s\n',record_date);

fprintf('record_time = %s\n',record_time);

fprintf('total_data_number = %f\n',total_data_number);

fprintf('Sample_Rate = %f\n',Sample_Rate);

data_file = zeros(total_data_number, numCols);

for r = 1:total_data_number

for c = 1:(numCols-1)

data_file(r,c) = fscanf(fid, '%f,', 1);

end

data_file(r,numCols) = fscanf(fid, '%f\n', 1);

end

head = 1;

tail = size(data_file, 1);

fclose(fid);

end

function [varargout] = simpleSplit(str, idx)

spaces = [0, find(str==' '), numel(str)+1];

spaces(diff(spaces)==1) = [];

coder.unroll()

for i = 1:numel(idx)

varargout{i} = strtrim(str((spaces(idx(i))+1):(spaces(idx(i)+1)-1))); %#ok<AGROW>

end

end

function out = fscanfForScalar(fid, format)

tmp = fscanf(fid, format, [1,1]);

if isempty(tmp)

out = 0;

else

out = tmp(1,1);

end

end