How to detect, open and process nxm matrices?

Hi i have a code to open .txt files with nx1 matrices, it works very well:

[filename1,filepath1]=uigetfile({'*.txt*','Text Files'},...
  'Select Data File 1');
   cd(filepath1);
 fp= fopen(filename1);
 fgets(fp); 
 A = textscan(fp, '%f');
 fclose(fp);
 result=A{:};

But i want to open 1xm matrices with this code also. Do do this, i though to create an if-else structure to deteckt if the data matrix is nx1 or 1xm and then textscan it with textscan(fp, '%f') or textscan(fp, '%??'). Or is there another way to make this without creating an if-else structure? Thanks in advance!

Answer 1

[Jan]

fp = fopen(filename1);
if fp == -1, error('Cannot open %s', filename1); end
fgets(fp);  % Skipping a header line?!
pos = ftell(fp);

tmpLine = fgets(fp);  % Read first line
tmpData = sscanf(tmpLine, '%g ', Inf);
nDataPerLine = length(tmpData);

fseek(fp, pos, -1);  % Spool one line back
Data = fscanf(fp, '%g ', [nDataPerLine, Inf]);
fclose(fp);

Answer 2

[Azzi Abdelmalek]

If your files are not big, you can use

v=dlmread('file.txt')

[Ekin]

I want to first open my file with "uigetfile", by browsing it. Depending on the data in the file, the code should detect if it is 1xm or nx1 and process.

[Ekin]

And files will be really big.

[Jan]

uigetfile does not open files, but determine the file name only. Of course you can apply Azzi's suggestion together with your uigetfile part.

Some users think, that 1MB is big, others use this term for 20 TB files.

Answer 3

[Walter Roberson]

textscan(fp, '%f')

is fine to read either format. End-of-line is considered to be "whitespace" for textscan(), so it will be fine without you having to code differently for the two situations.

If it is important that you process the two forms in different ways, then:

currentloc = ftell(fp);  %find out where we are now
small_input = fread(fp, 20, '*uint8');   %read a little
fseek(fp, currentloc, 'bof');  %go back to where we were

if any(small_input == 10)    %control J, newline, \n
  %there was a newline, so it is n x 1 rather than 1 x n.
  ...
else
  %it was 1 x n
end

[Ekin]

It is not important to process them in different ways for me, i just want to process both of them. I am currently using

   textscan(fp, '%f')

It works fine with mx1 matrices, but gives error when input is 1xn.

[Walter Roberson]

What error do you get?