How to pass textscan data to array?

Hello fellows,

I have a write a code where I acess to a *.txt file and after get the information from the header, i close with fclose the open file.

After that I call again the open file, and the textscan to get the data without the header file, now I get blocked because I need acess to certain filds in the whole *.txt file to load a dinamic graph, can someone help me

Thanks in advanced

 %OPEN FULL PATH FILE
           fileID = fopen(fullfile( FilePath, FileName),"r");
           %START TEXTSCAN HEADER
           startHeader = textscan(fileID ,'%s', 'WhiteSpace', '\n');
           %CONVERT INT 1X1 MATRIX
           Header = startHeader{1};
           %END TEXTSCAN HEADER
           Header = Header(strncmp(Header, '#', 1));
           %COUNT HEATHERLINES
           headerLine = numel(Header);
           %GET TOTAL OF FILE ROWS EXCLUDE HEADER LINES
           totalLines = numel(startHeader{1})-headerLine;
           %PASSING HEADER VARIABLES TO FIELD
           for k = 1:headerLine
              %LABEL 
              label{k} = extractBetween(Header{k},'#',':');
              %FIELD
              field{k} = extractAfter(Header{k},':');
           end
           %LOAD LABELS AND FIELDS REQUIRED BY POSITION
           %
           %CLOSE THE INSTANCE TO READ THE HEADER
           fclose(fileID);
           %OPEN AGAIN THE FILE
           fileID = fopen(fullfile( FilePath, FileName),"r");
           %CREATE A INITIAL VARIABLE
           j=0;
           %MAKE A NEW SCAN EXCLUDIN THE HEADER LINES
          
           %PRE-CONSTRUCT A TABLE TO INTRODUCE SAMPLE_COUNT AND TIME
           for j = 1:12%totalLines
               %samples(j) = sscanf(Header{j}, 'Samples: %f');
              startAquisition(j,:) = textscan(fileID, '%d %f %f %f %f %f %f %f %f %f %f', ...
               'HeaderLines',headerLine,'Whitespace','\n')
                
           end
          

           fclose(fileID);

[Star Strider]

I am not certain what you are doing.

The textscan function produces a cell array as output, so it will be necessary to use the cell2mat function to create a matrix from it (which should be possible since all the data in ‘startAquisition’ appear to be numeric). It might be helpful to include the name-value pair 'CollectOutput',true in the textscan call.

[dpb]

As usual, it would be much easier to provide useful code specific to the situation if you would attach the input file (if it's large, just the header and a few representative data lines would suffice) as a .txt file (use the paperclip icon).

As a couple comments on above code,

%OPEN FULL PATH FILE
fileID = fopen(fullfile( FilePath, FileName),"r");
startHeader = textscan(fileID ,'%s', 'WhiteSpace', '\n');   %START TEXTSCAN HEADER
Header = startHeader{1};                                    %CONVERT INT 1X1 MATRIX
Header = Header(strncmp(Header, '#', 1));                   %END TEXTSCAN HEADER
headerLine = numel(Header);                                 %COUNT HEATHERLINES
totalLines = numel(startHeader{1})-headerLine;              %GET TOTAL OF FILE ROWS EXCLUDE HEADER LINES
for k = 1:headerLine                                        %PASSING HEADER VARIABLES TO FIELD
  label{k} = extractBetween(Header{k},'#',':');             %LABEL 
  field{k} = extractAfter(Header{k},':');                   %FIELD
end

%fclose(fileID);                                             %CLOSE THE INSTANCE TO READ THE HEADER
%fileID = fopen(fullfile( FilePath, FileName),"r");          %OPEN AGAIN THE FILE

% don't close file just to reopen it immediately, use
frewind(fileID)
% instead...

j=0;

% if you correctly read file, should know how many lines, don't use "magic"
% numbers in code -- this looks like sonmething must have gone wrong...
for j = 1:12%totalLines
  startAquisition(j,:) = textscan(fileID, '%d %f %f %f %f %f %f %f %f %f %f', ...
                             'HeaderLines',headerLine,'Whitespace','\n')
end

The use of 'HeaderLines',headerLine in each call in a loop such as the above is going to skip every other record and the use of 'Whitespae','\n' is peculiar here...

Without the specific file structure itself to see we can't really tell for absolute certain, but I'd wager you could replace all of the above with a single call to

data=readmatrix(fullfile( FilePath, FileName));

[Biza Ferreira]

Hello dpb and star strider,

Thanks for your support, sorry for not send the specific file structure, now I include the INPUT.txt file . My intention is to use the data aquisition to fill a graph dynamically. I have interest in the header too, becouse it have some important information, that is the motive for first part of the code, where I need the total dimention of the file And count the header lines before exclued it. So in the second part i need access to certain columns with intention to charge it in the graph.

Thanks in advance

Answer 1

[dpb]

OK, with the multiple-line header, and parsing it, the code isn't so bad, after all... :)

I'd probably still do a little different, but your first part suffices to locate the header and parse it -- but then, if you're going to close the file and reopen it anyway, then use the information you found before and read the rest as

...
dataAcquisition=readmatrix(fullfile(FilePath,FileName),'NumHeaderLines',headerLine);

and you've got the whole data array in memory -- you can then just remove the columns don't care about by something like

ixKeep=[1 3:6];                             % an arbitrary set of column indices wanted; salt to suit...
dataAcquisition=dataAcquisition(:,ixKeep);  % and keep those...

[Biza Ferreira]

OK, I understand the modification you have done in my code, but I have to introduce in the table or matrix two more filds, and then plot all data merged. Imagin a need to inplement data sample count like this:

j=0;
time = 0;
for j=1:totalLines
    j=j+1; %COUNT LINES TO CERTIFICATE THAT's ALL READ
    time = time + 0.008; %TIME BETWEEN EACH AQUISITION
end

[dpb]

You don't need a loop to do that in MATLAB. The header of your above file says the sample rate is 1000, one presumes it's Hz although that isn't given.

You parsed the header; one would presume for precisely such reason -- if not going to use it, why bother?

ix=matches(label,'SamplingFrequency');              % which is the sampling rate?
Fs=str2double(field{ix});                           % the sampling frequency value
dt=1/Fs;                                            % the sample time delta
dataAcquisition=dt*[0:size(dataAcquisition,1)-1].'; % append time vector as new column

Or, you might consider turning it into a timetable; there are many useful builtin features there depending on what next you want to do.

And, of course, you could have gone ahead and done the numeric conversion when you were parsing the rest of the header. A clever use of regexp could return the tokens directly.

[Biza Ferreira]

Well that is why i decide mantain the header, which you have mencioned I can rescue all information I need, like the sample frequency, witch can change if the user require... In this moment i am working with a frequency of 125Hz, which give me 0.008s per sample

This way I try to make my code the most adaptative to many situation.

"You don't need a loop to do that in MATLAB. " - That is awesome, becouse loop cause lots of problems in the processing.

"...there are many useful builtin features there depending on what next you want to do." - My next step is try to mount a dynamic graph, where I can analyse the ecg in a timeline. Turn it in a timetable would bring to me some advantages?