Split into different columns a .csv file with characters ';'

I have a large .csv file with the below format ('Date;Time;Value').

'16/4/2017;05:12:02;1.21'

and need to import it into 3 collumns.

'16/4/2017' '05:12:02' '1.21'

My code is:

fid = fopen('KD.csv','r');

C = textscan(fid, '%s %s %f %f %s', 'HeaderLines',1, 'CollectOutput',true);

fclose(fid);

[dt,val,exch] = deal(C{:});

w=regexp(dt,'\s+','split')

out=reshape([w{:}],1,[])'

time = [dt regexp(dt, '\d\d:\d\d:\d\d', 'match', 'once')]

Any help?

Answer 1

[Guillaume]

Even better than xlsread, readtable would be a much better and modern option. If the header line that is skipped actually contains columns names readtable could even parse these. Skipping it, as in the original code:

kd = readtable('KD.csv', 'delimiter', ';', 'ReadVariableNames', false, 'HeaderLines', 1);
kd.Properties.VariableNames = {'dt', 'val', 'exch'};  %name the columns of the table

Note that readtable is usually clever enough to detect the delimiter, number of lines to skip and whether or not the variable names are included, so:

kd = readtable('KD.csv');

may be enough.

[Ancalagon8]

@Guillaume thanks a lot! "readtable" did exactly what i needed!

In order to plot the values (3rd collumn) with time (2nd collumn), i should convert time from non-numeric form to what?

Thanks in advance!

[Guillaume]

"thanks a lot! "readtable" did exactly what i needed!"

So why did you unaccept the answer? Because it didn't answer your completely different other question which wasn't asked in the first place?

"i should convert time from non-numeric form to what?"

If readtable didn't already convert your time to datetime. You can do it manually,

kd.val = datetime(kd.val, 'InputFormat', 'HH:mm:ss');

Modern versions of matlab know how to plot against datetime.

[Ancalagon8]

@Guillaume Yes. I did manually. Thanks again!