Read a .txt except one column

Hello,

I come to you because I have a problem to import a .txt. In fact, I have some .txt files following the same format, namely : 30 lines of headers, and then a format like that :

       D;       Q; valQ; Ptot; Fsol;  ETP; Temp; Vent; Humi;     DLI;     SSI;    IHGR;     SWI;
19580801;     -99;    -;  0.0; 0.00;  4.0; 19.9;  2.3;  10.;  2827.9;  2495.2;  -0.999;   0.435;

I tried different function (dlmread, textscan, ...) but I have a problem with the third column because it's not an integer. I'm not interested by these column. So, I would like to have a matrix in Matlab, from the .txt, without the third column. Do you know how to help me ? Thanks !

[per isakson]

Try something like

cac = textscan( fid, '%f%f%*s%f%f%f%f%ff%f%f%f', 'Headerlines',30+1 ...
'Delimiter',';', 'CollectOutput',true )

and study the documentation of textscan

Answer 1

[dpb]

Looks like an extra count for 'headerlines', maybe? 'Pends on whether the above line is the 30th or the previous 30 were some other text and then the variable names...

I'd only point out here's a place where the use of repmat to count repeat fields is handy rather than trying to count how many '%f' fields one has typed in...

fmt=['%f%f*%f' repmat('%f',1,10)];
cac = cell2mat(textscan(fid,fmt, 'Headerlines',30+1 ...
                                 'Delimiter',';', 'CollectOutput',true ));

Also cell2mat wrapped around textscan gets the double array directly instead of a cell array which is more pain to dereference when not needed.

[Walter Roberson]

I think you meant textscan not testscan

[dpb]

Ayup...corrected typo...

Answer 2

[Thomas DENIS]

Allright, thanks you a lot both of you ! I managed to do it by using :

fid = fopen('abc.txt');
fmt=['%f%f%*s' repmat('%f',1,10)];
C = textscan(fid,fmt,'Headerlines',30,'Delimiter',';');
Deb = cell2mat(C(:,:));
fclose(fid);

Thanks you !

[dpb]

What's the point of not simply wrapping textscan with cell2mat directly instead of the temporary C? Use the 'collectoutput' named parameter to collect like terms, too...