Re-organize LS-DYNA data from text file removing header and footer

Hello,

I'm trying to extract data from an LS-DYNA output text file and write it into a new file with a new format, but I find the initial format a bit complicated and I have problem getting what I want, which is:

• Header: keep the same (LS-DYNA keyword ... date)
• One line with the legend of the data, with timestep and time moved to columns 1 and 2, respectively (two new columns should be created)
• N lines with the actual data
• No footer

The starting file looks like this:

and I'd like to have it like this

From that point I can use textscan to manipulate the data the way I want.

Numpy library in Python has the command genfromtxt that allows to remove header and footer, then choose which data to extract from a specific column of a specific line. I couln't find such a command in MATLAB, and i tried with a combination of, fgets, fgetl, and textscan, but impossible to achieve what I'm looking for.

Does anyone have an idea of which function would work best?

Thanks a lot!

Answer 1

[Sam McDonald]

I understand that you would like to know what MATLAB functions would work best to format your data so it can be easily imported using textscan.

Because your original data file is not formatted the same on each line, you will likely need to call 'fgetl' to read in each line as a character array. Use 'eof' to determine when the end of the file has been reached.

To extract the numerical values from lines that contain many characters, you can first remove the characters by using the 'replace' function.

https://www.mathworks.com/help/matlab/ref/replace.html

Then, you can use 'textscan' on the remaining string to extract the numerical values.

Once you have read the data into MATLAB, you can use a function like 'fprintf' to write the data to another file.

https://www.mathworks.com/help/matlab/ref/fprintf.html

These functions should help you write custom code to achieve the desired format.

[Davide De Cicco]

Thanks a lot for your answer.

I tried the replace function but I couldn't really understand how to use it. So I decided to go with sscanf instead. In fact, in the documentation, there is an example that shows how to extract only the numbers when there are different formats in a string. And it works pretty much in the same way as textscan (or I just didn't understand the difference between them).

I ended up with a code that gives me what I need (even if it is probably possible to optimize it). Here it is, in case it might be useful to someone else:

function numsTot = transfNodout(resufile, ext)

      %--------------------
      % Read data from file
      %--------------------

      % Open the file
      fid3 = fopen(resufile,'r');

      % Read the header
      header = 5;    % nb of header lines
      for line = 1:header
          tline3 = fgetl(fid3);
          head{line} = tline3;
      end

      % Loop to read the NODOUT data
      line = 1;
      while ischar(tline3)
          tline3 = fgetl(fid3);
          formatSpec = [repmat('%*s ',[1,24]) '%f %*s %*s %*s %f %*s']
          nums1(line,:) = sscanf(tline3, formatSpecs, [1, inf]);
          for j = 1:3
              tline3 = fgetl(fid3);
          end
          nums2(line,:) = sscanf(tline3, '%f');
          for j = 1:3
              tline3 = fgetl(fid3);
          end
          numsTot(line,:) = [nums1(line,:) nums2(line,:)];
          line = line+1;
      end

      % Close the file
      fclose(fid3);

      % Get the legend
      fid3 = fopen(resufile,'r');
      lin_legend = 8; % line where the legend is
      for line = 1:lin_legend
          tline3 = fgetl(fid3);
          legend{1} = tline3;
      end
      fclose(fid3);

      %---------------------
      % Write data into file
      %---------------------

      % Write the header
      newname = [ext resufile];
      fid4 = fopen(newname', 'w');
      for line = 1:header
          fprintf(fid4,'%s\n', head{line});
      end

      % Write the legend, with addition of 2 columns
      fprintf(fid4, '  time step      time     ');
      fprintf(fid4, '%s\n', legend{1});

      % Write the data
      formatSpec = [repmat('%E ',[1,14]) '%f\n'];
      for line = 1:size(numsTot,1)
          fprintf(fid4, formatSpec, numsTot(line,:));
      end
      fclose(fid4);

end